Food Chemistry 199 (2016) 760–767

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Localized lipid autoxidation initiated by two-photon irradiation withinsingle oil droplets in oil-in-water emulsions

http://dx.doi.org/10.1016/j.foodchem.2015.12.0700308-8146/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: mola@food.ku.dk (M.L. Andersen).

Piret Raudsepp a,b, Dagmar A. Brüggemann a,b, Jes C. Knudsen b, Mogens L. Andersen b,⇑aMax Rubner-Institut, E.-C.-Baumann-Straße 20, D-95326 Kulmbach, GermanybDepartment of Food Science, University of Copenhagen, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmark

a r t i c l e i n f o a b s t r a c t

Article history:Received 8 October 2015Received in revised form 14 December 2015Accepted 16 December 2015Available online 18 December 2015

Keywords:BODIPY665/676

RadicalsTwo-photon irradiationLipid oxidationOil dropletCLSM

The initiation of lipid autoxidation within single oil droplets in Tween-20-stabilized oil-in-water emul-sion was achieved by highly focused two-photon (2P) irradiation at excitation wavelength (kex)700 nm. The radical formation was enhanced by inclusion of the photo-cleavable radical initiator di-tert-butyl peroxide (DTBP) into the droplets, and demonstrated with confocal microscopy usingradical-sensitive probe BODIPY665/676. The radical chain reactions progressed up to 60 lm; however,there were no indications of oxidation in neighboring droplets demonstrating that radicals and oxidizedprobe molecules were not able to migrate between oil droplets. In addition, the spatial propagation oflipid autoxidation increased with the degree of oil unsaturation.

� 2015 Elsevier Ltd. All rights reserved.

1. Introduction

The positive nutritional properties of polyunsaturated fattyacids have generated an interest in developing foods with highlevels of unsaturated lipids. However, polyunsaturated fatty acidseasily develop quality degrading rancidity due to autoxidation dri-ven by radical chain reactions. Control of lipid autoxidation is ofmajor importance in order to be able to produce stable and deli-cious foods with high levels of polyunsaturated fatty acids. Mostfoods are multiphase systems with complex microstructures,where lipids are dispersed in aqueous phases. Consequently, theoxidative stability of oil-in-water emulsions has received a lot ofinterest (Berton-Carabin, Ropers, & Genot, 2014; Waraho,Cardenia, Decker, & McClements, 2010). This has included studiesof the role of various physical and chemical characteristics of lipidphase, aqueous phase, and interface, although, in most cases theautoxidation phenomena in emulsion systems have only beenstudied and characterized on bulk macroscopic scale. Thus, theprogression of lipid oxidation taking place on microscopic scalein individual oil droplets is still rather poorly understood.

Recently, it was shown that the lipid-soluble radical-sensitivefluorescent probe BODIPY665/676 can be used to monitor radicals

involved in lipid oxidation on microscopic scale in bulk oils as wellas in oil-in-water emulsions (Raudsepp, Brüggemann, & Andersen,2014a, 2014b). Reactions between radicals and the probe changesits fluorescence, thus enabling following the concentration of rad-icals by two excitation wavelengths (kex): (1) decrease in fluores-cence intensity at kex 670 nm with maximum emission peak at685 nm, and (2) more pronounced change as increase in fluores-cence at kex 580 nm with maximum emission peak at 605 nm.Moreover, BODIPY665/676 is highly lipophilic, it is evenly distributedinside emulsion oil droplets, and the low water solubility preventsit from diffusing between the droplets. It is therefore possible tomonitor the presence of radicals in selected single oil droplets, thusallowing for studies of lipid oxidation in emulsions on microscopicscale. Moreover, new research directions such as autoxidation pro-gression in different lipid compositions of oil droplets would bepossible (Raudsepp et al., 2014a).

The objective of this study was to investigate the spatial pro-gression of radical-driven autoxidation reactions on microscopicscale inside individual oil droplets in Tween-20-stabilized oil-in-water emulsions. In order to achieve this goal, two major subjectmatters needed to be investigated: (1) the feasibility of generatingradicals from photoinducible radical initiator di-tert-butyl perox-ide (DTBP) by two-photon (2P) irradiation inside the confinedspace of a single oil droplet, and (2) the potential use of the fluores-cent probe BODIPY665/676 with confocal laser scanning microscopy(CLSM) to follow the propagation of radical reactions. The

P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767 761

advantage of the 2P-technique in generating radicals is that it canbe used to initiate spatially highly resolved photochemicalreactions on the length scale of a few micrometers inside singleoil droplets. The approach of controlled radical generation withthe 2P irradiation and following lipid peroxidation under a micro-scope has not been utilized or studied before. However, it has beenpreviously used to form radicals (Adam, Kita, & Oestrich, 1994; Sun& Kawata, 2004) and uncage chemical compounds (Brown, Shear,Adams, Tsien, & Webb, 1999; Idoux & Mertz, 2011; Nikolenko,Yuste, Zayat, Baraldo, & Etchenique, 2005) demonstrating that 2Plaser does possess enough energy to induce photolytic reactions.Therefore, applying 2P irradiation at kex 700 nm to photochemi-cally cleave the O–O bond of DTBP should be possible. DTBP hasbeen reported to be cleaved with a single-photon absorption inthe range of 213–350 nm (Keller-Rudek, Moortgat, Sander, &Sörensen, 2013), and the homolytic O–O bond dissociation energyof the peroxide bond in DTBP has been reported to be 38 kcal/mol(Montalti, Credi, Prodi, & Gandolfi, 2006) or 41 kcal/mol (dosSantos, Muralha, Correira, & Simoes, 2001). The calculated 2P laserenergy at kex 700 nm is 82 kcal/mol, which should be sufficient tocleave the O–O bond and produce alkoxyl radicals:

ðCH3Þ3C� OO� CðCH3Þ3 þ hm ! 2ðCH3Þ3C� O�

The spatial spreading of radicals in the interior of individual emul-sion oil droplets after highly localized radical generation within anarea of few micrometers at the center of the droplet was investi-gated in different oil media in order to evaluate the ability of thefluorescent probe to monitor and report on the presence of differenttypes of radicals. Emulsions were made with saturated medium-chain triglyceride (MCT) oil, where autoxidation reactions wereunlikely to occur, or unsaturated sunflower oil (SFO) or MCT oilwith added unsaturated methyl linoleate (Me-LH). In the lattertwo emulsions, lipid oxidation reactions were expected to takeplace. Moreover, compared to SFO, Me-LH is a much smaller mole-cule, which is beneficial in studying temporal and spatial progres-sion of radical reactions.

2. Experimental

2.1. Materials

Medium-chain triglyceride (MCT) oil was purchased fromCognis GmbH, Ludwigshafen, Germany. Sunflower oil (SFO) waspurchased from a local supermarket and purified before use byalumina column chromatography (Yoshida, Kondo, & Kajimoto, 1992).The lipophilic fluorescent probe (E,E)-3,5-bis-(4-phenyl-1,3-buta-dienyl)-4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY665/676

(B-3932)) and internal fluorescent standard beads InSpeckTM DeepRed (633/660) Microscope Image Intensity Calibration Kit, 2.5 lmwere obtained from Life Technologies Corporation, Oregon, USA.Methyl linoleate (Me-LH) and Tween-20 were purchasedfrom Sigma–Aldrich Inc., St Louis, Missouri, USA, and di-tert-butyl peroxide (DTBP) from Merck Schuchardt OHG, Hohenbrunn,Germany.

2.2. Viscosity measurements of the oils

Viscosity of medium-chain triglyceride (MCT) oil and sunfloweroil (SFO) was measured with AR-G2 magnetic bearing rheometer(TA Instruments, West Sussex, UK) at a steady state flowwith shearrate from 1000 s�1 to 0.1000 s�1 by cone/plate geometry ofØ40 mm 1�00260 0 aluminum steel cone at a temperature of 37 �C.Measurements were done in quadruplicate.

2.3. Absorbance spectrum of BODIPY665/676

UV–Vis spectroscopy and the recording of the absorbance spec-trum of BODIPY665/676 (1 lM) dissolved in methanol were con-ducted on UV/VIS spectrometer (Cintra 40, GBC ScientificInstruments Pty Ltd., Australia). The spectrum was recorded from250 nm to 800 nm with a slit width of 2.5 nm and speed 300 nm/min. The spectrum was baseline corrected.

2.4. Oil-in-water Tween-20-stabilized emulsions for confocal laserscanning microscopy (CLSM)

Oil-in-water Tween-20-stabilized emulsions were made by fol-lowing the method of Berton, Genot, and Ropers (2011). Emulsionsconsisted of 0.5 wt% of Tween-20, 69.5 wt% sodium acetate-aceticacid buffer (pH = 4.65), and the oil phase comprised of 30 wt% ofMCT oil, or 28 wt% MCT oil containing 2 wt% of methyl linoleate(Me-LH), or 30 wt% of sunflower oil (SFO). The mixture was emul-sified by Ultra Turrax T25 (IKA Works GmbH & Co. KG, Staufen,Germany) at 8000 rpmwith a dispersion element of 8 mm in diam-eter. The fluorescent probe BODIPY665/676 at concentration of 1 lMand radical initiator DTBP at concentration of 5.7 lM were addedinto the oil phase prior to mixing the emulsion.

2.5. The two-photon (2P) irradiation of a single oil droplet of Tween-20-stabilized emulsions

The two-photon (2P) irradiation experiments of a single oil dro-plet of Tween-20-stabilized emulsions made with MCT oil andmethyl linoleate (Me-LH) containing BODIPY665/676 (1 lM), andDTBP (5.7 lM) were conducted on a confocal Leica TCS SP5-X MPmicroscope by employing a Chameleon Ultra pulsed laser (Coher-ent, Santa Clara, California, USA) at excitation wavelength (kex)700 nm, output power 1.615 W (stdev 0.016), pulse width 140 fs,and repetition rate 80 MHz. The images were recorded with 40�HCX PL APO CS, NA 1.30 oil immersion objective, pinhole was setto 1 AU, line average was 2�, and image resolution 1024 � 1024.The 2P irradiation was conducted by positioning the laser pointscanner of the 2P laser in an area of a zoomed-in oil droplet(Scheme 1A). The area subjected to 2P irradiation was chosensmaller than the diameter of the oil droplet to ensure affecting onlythe oil phase and not damaging the interface with the laser light.For the experiments of the 2P irradiation of a large area, a dropletsize around 85 lm in diameter was selected with an area of irradi-ation around 50 lm in diagonal, and the change in the fluorescenceintensities after the 2P irradiation was given relative to the non-irradiated droplet. In the case of highly localized 2P irradiationstudies, the droplet size was selected to be around 120 lm indiameter, and the irradiation area was kept to a minimum beingaround 10 lm in diagonal. After the 2P irradiation, the samplewas scanned with a white light supercontinuum laser with twoexcitation wavelengths (kex) and the signal was detected usingeither a photomultiplier tube or a hybrid detector in photon count-ing mode: (1) kex 670 nm and signal detection at 675–785 nm, and(2) kex 580 nm and signal detection at 585–620 nm. In the photoncounting mode at kex 580 nm, a pixel on the image appears whitewhen a photon is detected and black when it is not detected; thus,an increase in the fluorescence intensity will be visualized as whitepixels. The changes in the fluorescence intensities were calculatedrelative to the internal fluorescent standard beads. The fluores-cence intensity of the oil droplets was measured by drawing a cir-cle of region of interest (ROI) covering the oil droplet, andexcluding the border regions. The average value of the fluorescenceintensity inside the ROI was acquired, and the change in the fluo-rescence intensity was calculated. In the highly localized radicalgeneration experiments, seven ROIs were drawn across the

Scheme 1. (A) Setup of the 2P laser irradiation experiment of single oil droplets in oil-in-water emulsions. (B) Illustration of region of interest (ROI) selection and spheres ofthe irradiated oil droplet used for quantifying the effect of distance of the changes in fluorescence. The diameters of each ROI were 15 lm.

762 P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767

diagonal of the oil droplet and the data was classified into four dif-ferent regions: (1) the 2P irradiated area (focus), (2) the 1st circle,(3) the 2nd circle, and (4) the 3rd circle (Scheme 1B) in increasingdistance order from the 2P irradiated area.

3. Results

3.1. The 2P irradiation of a large area of a single oil droplet: Radicalgeneration from DTBP and oxidation reporting by BODIPY665/676

Temporal relationship between the concentration of radicalsand the response of the radical-sensitive probe BODIPY665/676 inoil droplets made with MCT oil + methyl linoleate (Me-LH) con-taining the radical initiator DTBP was quantified by graduallyincrementing 2P irradiation times from 1 s to 15 s (Fig. 1). Radicalswere generated by cleaving DTBP with a 2P pulsed laser at kex700 nm at the center of oil droplets covering the majority of thedroplet’s area, but not compromising the integrity of droplet inter-faces. For that purpose, all the oil droplets chosen had diametersaround 85 lm, and the zoomed-in area of the 2P irradiation wasaround 50 lm in diagonal. This approach would produce a largenumber of radicals. A control emulsion, which did not contain

Fig. 1. The effect of increasing 2P irradiation times of single emulsion oil droplets onemulsion was made with MCT oil + 2 wt% methyl linoleate (Me-LH), and radical sensitivewhereas the controls did not contain DTBP (s). A single oil droplet was 2P irradiated ausing a photomultiplier tube at 670 nm or a hybrid detector in photon counting mode at 5measurements were done at least in triplicate. The error bars depict the standard devia

DTBP, was included in the investigation to evaluate the direct influ-ence of the irradiation on the probe. The fluorescence changes weremeasured at excitation wavelengths (kex) 580 nm and 670 nm.Based on the recent research, the fluorescence intensity ofBODIPY665/676 measured at kex 580 nm is expected to increase inthe presence of radicals, and the fluorescence at kex 670 nm willsimultaneously decrease (Raudsepp et al., 2014a, 2014b). It wasobserved that the fluorescence intensity at kex 580 nm increasedlinearly with the irradiation time in both emulsions (Fig. 1A),whereas in the emulsion containing DTBP, a plateau intensity val-ues were reached after 10 s irradiation. In the control emulsion, themaximum of the fluorescence intensity reached around 60% of theintensity of the emulsion with DTBP. The intensity measured at kex670 nm decreased with increasing irradiation times having identi-cal intensities for up to 5 s of the 2P irradiation. In the controlemulsion, the intensity changes at kex 670 nm fluctuated around2–3% from the initial values (Fig. 1B), and in the emulsion contain-ing DTBP, the fluorescence had around 10% decrease after 10 s of 2Pirradiation.

These results demonstrated that radicals were generatedby 2P irradiation from DTBP; the radicals readily reactedwith BODIPY665/676, and subsequently caused a change in the

the fluorescence intensities of BODIPY665/676. A Tween-20-stabilized oil-in-waterfluorescent probe BODIPY665/676 (1 lM). The samples contained DTBP (5.7 lM) (d),

t kex 700 nm under a confocal microscope. The fluorescence intensity was detected80 nmwith two excitation wavelengths: (A) at kex 580 nm, and (B) at kex 670 nm. Alltions.

P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767 763

fluorescence spectra. The direct photobleaching effect of the probeby the 2P laser was negligible in the present emulsion systemsand under present conditions, even though, the wavelengths ofthe radical generation and excitation of the probe, 700 nm and670 nm respectively, were in the same narrow range of the visiblespectrum. Nevertheless, the absorbance spectrum of BODIPY665/676

confirmed that 2P laser did not directly affect the fluorescence prop-erties of the probe (Fig. 2) and photobleaching did not occur. Theirradiation results observed at kex 580 nm indicated that whilecleaving DTBP with the 2P laser at kex 700 nm is feasible, DTBP didnot seem to be the only source of radicals – the fluorescence inten-sity of the oil droplet of the control emulsion without DTBPincreased up to 60% of the emulsion that did contain the peroxide.

3.2. The 2P irradiation of a large area of single oil droplets: Radicalmigration into neighboring droplets?

Oil droplets made with MCT oil + methyl linoleate (Me-LH) con-taining DTBP and BODIPY665/676 were 2P irradiated for 10 s to studythe possible mobility of radicals across interfaces into neighboringdroplets. After the irradiation, a substantial increase in the fluores-cence intensity of around 600% relative to the non-irradiated emul-sion measured at kex 580 nm inside the irradiated droplet wasobserved, whereas the intensities in the neighboring dropletsremained at the initial level (Fig. 3A). Concurrently, the changesin the fluorescence intensity in the irradiated droplets measuredat kex 670 nm were relatively small with only 7–8% of decrease,and similarly the fluorescence in the neighboring droplets didnot change (Fig. 3B). Moreover, the stability of the fluorescencealterations evoked by the 2P irradiation measured at kex 580 nmand 670 nm remained constant for up to 60 min in an experiment,where the droplet sizes, irradiation times, and the physical condi-tions of the emulsion (dehydration) varied (Fig. 4). Thus, theabsence of fluorescence changes in the neighboring droplets indi-cated the radicals did not migrate between the oil droplets butremained isolated, and the fluorescence changes were stable forat least an hour in changing physical conditions.

3.3. Highly localized 2P irradiation of oil droplets with different lipidcomposition: Propagation of radical reactions inside oil droplets

After establishing that neither BODIPY665/676 molecules nor rad-icals were able to move between non-polar oil droplets of an oil-in-water emulsion, the propagation of radical chain reactions was

Fig. 2. Absorbance spectrum of the fluorescent radical-sensitive probeBODIPY665/676 dissolved in methanol.

studied on microscopic scale in oil droplets with varying degreesof unsaturation: (1) Highly saturated MCT oil, (2) MCT oil contain-ing 2 wt% of methyl linoleate (Me-LH), and (3) unsaturated sun-flower oil (SFO). In addition to the different degree ofunsaturation, the viscosity of MCT oil and SFO were also very dif-ferent, 16.83 cP (stdev 0.15 cP) and 31.08 cP (stdev 0.67 cP),respectively. All emulsions contained BODIPY665/676 and DTBP,and were irradiated with the 2P laser at kex 700 nm for 7 s. Sincethe area of the 2P irradiation (focus) was approximately 10 lmin diagonal at the center of a 120 lm droplet (Scheme 1A and B),radical generation by this approach could be considered as highlylocalized. The spatial propagation of radicals in the focus of theoil droplets was followed immediately after the irradiation byrecording a series of 10 images for 24 s at kex 580 nm. The fluores-cence changes in different regions of interest (ROIs) in the focus,but also outside the focus within the limits of the droplet(Scheme 1B) were measured. The differences arising from the lipidcomposition were very prominent. The fluorescence intensities inthe focus increased with increasing unsaturation of triglycerideswith the lowest increase in the MCT oil droplets, around 0.45 fluo-rescence intensity units (UI) relative to the internal fluorescentstandard, and the highest in the SFO droplets, around 1.1 UI(Fig. 5). Moreover, the change in the fluorescence intensity alsodepended on the distance from the focus. The intensities in thefocus and in the first circle in the case of MCT oil and MCT oil+ Me-LH oil droplets had equal intensities, whereas in the SFO dro-plets, the difference was around 0.7 UI. That means that spreadingof the radicals and radical reactions was limited to a distance ofroughly 60 lm from the focus. In all three emulsions, the furtheraway from the focus the less pronounced differences in the fluores-cence intensities, and almost no changes occurring in the most dis-tant circles, where the intensities remained around the initialvalues. In addition, differences in the fluorescence intensitiesbetween the focus and other circles in the MCT oil and MCT oil+ Me-LH oil droplets decreased also with time by levelling outwithin 24 s. In the SFO droplets, on the other hand, the increasedfluorescence intensities inside the focus and in the first circle werestill observable after 24 s.

4. Discussion

The combined use of localized radical generation with two-photon (2P) irradiation and a radical-sensitive lipophilic probehas made it possible to observe and follow the temporal and spatialprogression of radical reactions within single oil droplets in an oil-in-water emulsion. This approach is a novel methodology in emul-sion science, and it opens up for microscopic studies of autoxida-tion of unsaturated lipids in complex matrices such as foods,where the microstructure can profoundly affect the distributionand properties of critical compounds such as antioxidants(Berton-Carabin et al., 2014; Zhao, Elias, & Coupland, 2015). More-over, tracking autoxidation reactions on microscopic scale couldunfold new opportunities for kinetic studies in structurally com-plex systems.

The 2P irradiation has previously been used to photolyze cagedcompounds (Idoux & Mertz, 2011; Brown et al., 1999; Nikolenkoet al., 2005). However, in the present study, controlled generationof radicals was targeted by cleaving the O–O bond of the radicalinitiator di-tert-butyl peroxide (DTBP) with the 2P laser at excita-tion wavelength (kex) 700 nm in a non-polar triglyceride environ-ment in the presence of the radical-sensitive probeBODIPY665/676. The fluorescence of BODIPY665/676 changes uponreacting with radicals and the emission spectra at kex 670 nmand 580 nm could be used to follow the presence of radicals(Raudsepp et al., 2014b). A recent study demonstrated the high

Fig. 3. Fluorescence changes in individual oil droplets of an emulsion irradiated with 2P laser. The Tween-20-stabilized oil-in-water emulsion was made with MCT oil + 2 wt%methyl linoleate (Me-LH) containing the radical-sensitive fluorescent probe BODIPY665/676 (1 lM) and the radical initiator DTBP (5.7 lM). A single oil droplet was 2Pirradiated at kex 700 nm for 10 s under a confocal microscope. Fluorescence changes inside the 2P irradiated droplet (d), and inside neighboring droplets in the close vicinityof the irradiated droplet (j) were measured. The fluorescence intensity was detected using a photomultiplier tube or hybrid detector in photon counting mode (A) at kex580 nm and (B) at kex 670 nm. The measurements were done in triplicate. The error bars depict the standard deviations.

Fig. 4. Stability of the change in the fluorescence intensities in oil droplets made with MCT oil + 2 wt% methyl linoleate (Me-LH) containing the radical-sensitive fluorescentprobe BODIPY665/676 (1 lM) and the radical initiator DTBP (5.7 lM). Differently sized droplets were 2P irradiated for various times: (1) 54.1 lm, 3 s, (2) 57.4 lm, 4 s, (3)76.3 lm, 3 s, (4) 76.7 lm, 10 s, (5) 45.1 lm, 5 s, (6) 97.2 lm, 5 s, (7) 75.7 lm, 5 s. The fluorescence stability was measured 60 min after the first irradiation time at (A) kex670 nm and (B) kex 580 nm.

764 P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767

lipophilicity of BODIPY665/676 is a key prerequisite for studying thelocation of radical formation in oil droplets in emulsions.(Raudsepp et al., 2014b) In order to follow radical chain reactions,it is vital that the probe molecules do not migrate between the oildroplets. Moreover, the photochemical influence of the 2P laser onthe probe is also of high importance since the probe is reporting onoxidation through becoming oxidized. Therefore, the direct effectof the 2P irradiation at kex 700 nm was investigated. The impactof the 2P laser on the fluorescent properties of the probe, such asphotobleaching, were negligible, since the fluorescence intensityat kex 670 nm changed only around 2–3% (Fig. 1B). Thus, directphotobleaching of the probe only occurred to a minor extent dur-ing the 2P irradiation in the present emulsion systems. On theother hand, the 2P irradiation of the control emulsion led to anincrease of the fluorescence intensity measured at kex 580 nm thatwas around 60% of the increase of the corresponding emulsion con-taining DTBP. This indicates that radicals generated by the 2P irra-diation were not only originating from the cleavage of DTBP, butmay also have been generated by photochemically-induced dam-ages of the lipid medium creating radicals directly and causingradiation-induced oxidation of the lipids, which has previously

been reported by Slater, Cheeseman, Davies, Proudfoot, and Xin(1987).

The stability studies of the evoked fluorescence changes of theprobe at both excitation wavelengths demonstrated that fluores-cence intensities remained stable for 60 min. Therefore, the exper-iments of the 2P irradiation of an oil droplet made with MCT oil+ Me-LH containing BODIPY665/676 and DTBP demonstrated that(1) it was possible to generate radicals with a 2P laser at kex700 nm without photobleaching the probe, (2) BODIPY665/676 wasable to report the oxidation events, and (3) the 2P irradiation hada substantial effect on the oxidation of the probe measured at kex580 nm, although the background fluorescence of the controlemulsion without the radical initiator was around 60% (Fig. 1).Even though the effect of irradiation reported by the probe fluores-cence intensity was high also without addition of the peroxideDTBP, inclusion of the radical initiator led to elevated generationof radicals and enhanced fluorescence changes. The propagationof radical reactions to neighboring droplets was investigatedbecause a recent study had shown that radical mobility betweenoil droplets of an emulsion was possible (Raudsepp et al., 2014a).Therefore, a single oil droplet of an emulsion made with MCT oil

Fig. 5. Highly localized 2P irradiation at kex 700 nm of a small area inside single oil droplets of emulsions with different degrees of unsaturation. Emulsions were made eitherwith MCT oil, MCT oil + 2 wt% methyl linoleate (Me-LH), or SFO. BODIPY665/676 (1 lM), and DTBP (5.7 lM) were incorporated into the oil droplets. The droplets were 2Pirradiated for 7 s, and the change in the fluorescence intensity was measured at kex 580 nm using a hybrid detector in photon counting mode. A series of 10 images over 24 swere recorded. The fluorescence intensity was measured in four different regions illustrated in Scheme 1B: Fluorescence intensity in the focus of the oil droplet (j), intensityin the 1st circle (d), intensity in the 2nd circle (▲), and intensity in the 3rd circle (.). Data shown in the figure are average results of two independent measurements. Theerror bars depict the standard deviations.

P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767 765

+ Me-LH containing BODIPY665/676 and DTBP was 2P irradiated for10 s. However, no changes in the fluorescence intensities of theneighboring droplets compared to the initial level before the 2Pirradiation were observed. Since the oxidized and non-oxidizedprobe molecules could not travel between the oil droplets(Raudsepp et al., 2014b), the unchanged fluorescence intensitiesof the neighboring droplets in the vicinity of the 2P irradiated dro-plet indicated that also the radicals generated either from DTBP orformed during radical chain reactions were not able to cross theinterfaces of the droplets or diffuse through the aqueous phase.Since the fluorescence intensities within the irradiated oil dropletsand in the neighboring droplets remained stable for at least60 min, this may be due to (1) the short life-time of the generatedradicals, which limited the distance radicals could diffuse awayfrom the location of their formation, and (2) radical generation atthe core of the droplet, which in combination with short-lived rad-icals, would be too remote for radicals to be able to transmit radicalchain reactions further away towards the interface. If radicalscould reach the interface of a droplet, the absence of transport-mechanisms for radicals could not be a factor because the excessof Tween-20 with a critical micelle concentration of 0.035 mMensured the presence of micelles in the aqueous phase (Huang,

Hopia, Schwarz, Frankel, & German, 1996). It has been shown pre-viously that micelles could be successfully exploited to transferradical chain reactions between the droplets, and induce oxidationin neighboring oil droplets (Raudsepp et al., 2014a). Although, theradical transfer seemed to greatly depend on the species of radi-cals. In the work of Raudsepp et al. (2014a), the radicals were gen-erated from temperature-dependent azo-initiator 2,20-azobis (2,4-dimethylvaleronitrile), which decomposed into carbon-centeredradicals that readily reacted with oxygen forming peroxyl radicals,while in the present work DTBP is cleaved into alkoxyl radicals. Theperoxyl radicals are less reactive than alkoxyl radicals, whichexplains the different propagation ranges observed in the twostudies (Pryor, 1986; Schuchmann & von Sonntag, 1987).

Under the present conditions, BODIPY665/676 molecules and rad-icals were not able to migrate between oil droplets. Therefore, thepropagation of highly localized radical reactions initiated by the 2Pirradiation within small areas inside single oil droplets of varyingdegrees of unsaturation was investigated. The spatial spreadingof radicals was followed by the fluorescence changes from the 2Pirradiated area (focus, around 10 lm in diameter) up to the bound-aries of a relatively large droplet (around 120 lm), excluding theboundaries themselves (Scheme 1 and Fig. 5). The fluorescence

766 P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767

intensity measurements indicated that propagation of radical andautoxidation reactions was limited to the distance from the focusto the 1st circle, which was less than 60 lm in diameter insidethe oil droplet. In addition, the differences in the spatial propaga-tion were strongly dependent on lipid composition, such as unsat-uration and viscosity of the triglycerides.

Enhanced increase in the change in the fluorescence intensitiesin SFO droplets after the 2P irradiation compared to moderateincrease in droplets containing MCT oil + Me-LH, and low increasein MCT oil droplets demonstrated that higher degree of unsatura-tion led to increased amount of radicals detected by the probe.Moreover, radicals were able to initiate autoxidation radical-driven chain reactions in the vicinity of the focus. The number ofchain reactions was higher in SFO compared to MCT oil or MCToil + Me-LH droplets because the unsaturated substrates were clo-ser to the 2P irradiated area, and could sustain a series of chainreactions initiated at the focus towards the outer boundaries ofthe droplet. Accordingly, radicals could be present for an extendedtime in SFO droplets, thereby increasing the probability ofBODIPY665/676 reacting with radicals. In the MCT oil + Me-LH dro-plets, the concentration of unsaturation (2 wt% Me-LH) was prob-ably too low to have an effect.

Nevertheless, 2P-induced radical generation was highly local-ized in a very confined space at the center of the oil droplet, andthe migration and impact of radicals, detected and measured as achange in the fluorescence intensities of the probe, were limitedto the dimensions of a single 2P irradiated oil droplet. Thus, theradicals did not cross the interface and did not progress to neigh-boring droplets. However, there were clear differences in the fluo-rescence changes between the ROIs in all of the samples, andespecially in SFO, where intensities of the focus, the 1st circle,and the 2nd circle were substantially different. This indicates thatdiffusion of irradiated and non-irradiated molecules in SFO dro-plets was slower than in the other two oil droplets. These observa-tions may be linked to different viscosities of the oils, which mayaffect radical generation and radical chain reaction efficiency(Yoshida, Itoh, Saito, Hakayawa, & Niki, 2004). Also, the reactivemolecules in SFO were bulky triglyceride molecules as comparedto the smaller methyl linoleate. The above mentioned factors couldaffect the diffusion of radicals and irradiated probe molecules, andthereby the progression of lipid oxidation. It has been found thatdiffusion of non-irradiated fluorescent probe molecules through2P irradiated damaged area could be slower due to photodamageand irreversible sample modifications caused by the 2P laser(Koester, Baur, Uhl, & Hell, 1999). In MCT oil, and MCT oil + Me-LH droplets the diffusion was faster resulting in similar fluores-cence changes in the focus and the 1st circle. In addition, the slightincrease in the final fluorescence intensity of the most distant cir-cle in the MCT oil + Me-LH droplet compared to the initial valuemight also indicate prolonged radical reactions spreading and last-ing longer due to the mobility of relatively small Me-LH. Further-more, this effect did not occur in MCT oil or SFO droplets.

There is currently an increased focus on the role of microstruc-ture during lipid oxidation. This includes emulsions, colloidal lipidparticles, as well as bulk lipids (Berton-Carabin et al., 2014;Budilarto & Kamal-Eldin, 2015; Raudsepp, Brüggemann,Lenferink, Otto, & Andersen, 2014c; Tikekar & Nitin, 2012; Zhaoet al., 2015). The present study has demonstrated that the combi-nation of the 2P irradiation and the probe BODIPY665/676 allows forstudies of lipid oxidation taking place within single emulsion dro-plets. Together with the possibility of using emulsions with differ-ent types of lipid droplets (Raudsepp et al., 2014a, 2014c), newexperimental strategies are possible for studying lipid oxidationon microscopic scale including the complex interplay betweenphysical and chemical factors and the location of important delete-rious radicals.

5. Conclusions

A two-photon (2P) laser was used to generate radicals from theradical initiator DTBP in a very localized and confined space in sin-gle oil droplets in a Tween-20-stabilized emulsion. The propaga-tion of radical reactions within the oil droplets could be followedwith the radical-sensitive fluorescent probe BODIPY665/676. Life-times of radicals formed from DTBP were extremely short; thus,the radical chain reactions took place within a few seconds andprogressed only up to �60 lm from the initiation site inside thedroplet. The radical reactions were not able to cross the interfacesand progress to neighboring oil droplets. The propagation and dif-fusion of radicals was dependent on the degree of unsaturation ofthe oil, and the viscosity of the lipid medium.

Acknowledgements

This work was funded by the Danish Agency of Science,Technology and Innovation. The CLSM work was done in Centerfor Advanced Bioimaging (CAB), Denmark.

References

Adam, W., Kita, F., & Oestrich, R. S. (1994). Two-photon chemistry in the laser jet:Generation of radical intermediates and their photochemical transformation.Journal of Photochemistry and Photobiology A, 80, 187–197.

Berton, C., Genot, C., & Ropers, M.-H. (2011). Quantification of unadsorbed proteinand surfactant emulsifiers in oil-in-water emulsions. Journal of Colloid andInterface Science, 354, 739–748.

Berton-Carabin, C. C., Ropers, M.-H., & Genot, C. (2014). Lipid oxidation in oil-in-water emulsions: Involvement of the interfacial layer. Comprehensive Reviews inFood Science and Food Safety, 13, 945–977.

Brown, E. B., Shear, J. B., Adams, S. R., Tsien, R. Y., & Webb, W. W. (1999). Photolysisof caged calcium in femtoliter volume using two-photon excitation. BiophysicalJournal, 76, 489–499.

Budilarto, E. S., & Kamal-Eldin, A. (2015). The supramolecular chemistry of lipidoxidation and autoxidation in bulk oils. European Journal of Lipid Science andTechnology, 117, 1095–1137.

dos Santos, R. M. B., Muralha, V. S. F., Correira, C. F., & Simoes, J. A. M. (2001).Solvation enthalpies of free radicals: O–O bond strength in di-tert-butylperoxide. Journal of the American Oil Chemists’ Society, 123, 12670–12674.

Huang, S. W., Hopia, A., Schwarz, K., Frankel, E. N., & German, J. B. (1996).Antioxidant activity of a-tocopherol and Trolox in different lipid substrates:Bulk oils vs oil-in-water emulsions. Journal of Agricultural and Food Chemistry,44, 444–452.

Idoux, E., & Mertz, J. (2011). Control of local intracellular calcium concentrationswith dynamic-clamp controlled 2-photon uncaging. PLOS One, 6, 1–8.

Keller-Rudek, H., Moortgat, G. K., Sander, R., & Sörensen, R. (2013). Earth SystemScience Data, 5, 365–373.

Koester, H. J., Baur, D., Uhl, R., & Hell, S. W. (1999). Ca2+ fluorescence imaging withpico- and femtosec two-photon excitation: Signal and photodamage. BiophysicalJournal, 77, 2226–2236.

Montalti, M., Credi, A., Prodi, L., & Gandolfi, M. T. (2006). In M. Montalti, A. Credi, L.Prodi, & M. T. Gandolfi (Eds.), Handbook of photochemistry (3rd ed.,pp. 529–534). Boca Raton: CRC Press, Taylor and Francis Group.

Nikolenko, V., Yuste, R., Zayat, L., Baraldo, L. M., & Etchenique, R. (2005). Two-photon uncaging of neurochemicals using inorganic metal complexes. ChemicalCommunications, 1752–1754.

Pryor, W. A. (1986). Oxy-radicals and related species: Their formation, lifetimes, andreactions. Annual Reviews of Physiology, 48, 657–667.

Raudsepp, P., Brüggemann, D. A., & Andersen, M. L. (2014a). Evidence for transfer ofradicals between oil-in-water emulsion droplets as detected by the probe (E,E)-3,5-bis-(4-phenyl-1,3-butadienyl)-4,4-difluoro-4-bora-3a,4a-diaza-s-indacene,BODIPY665/676. Journal of Agricultural and Food Chemistry, 62, 12428–12435.

Raudsepp, P., Brüggemann, D. A., & Andersen, M. L. (2014b). Detection of radicals insingle oil droplets of oil-in-water emulsions with the lipophilic fluorescenceprobe BODIPY665/676 and confocal laser scanning microscopy. Free RadicalBiology and Medicine, 70, 233–240.

Raudsepp, P., Brüggemann, D. A., Lenferink, A., Otto, C., & Andersen, M. L. (2014c).Oxidative stabilization of mixed mayonnaises made with linseed oil andsaturated medium-chain triglyceride oil. Food Chemistry, 152, 378–385.

Schuchmann, M. N., & von Sonntag, C. (1987). Hydroxyl radical-induced oxidationof diisopropyl ether in oxygenated aqueous solution. A product and pulseradiolysis study. Zeitschrift für Naturforschung, 42b, 495–502.

Slater, T. F., Cheeseman, K. H., Davies, M. J., Proudfoot, K., & Xin, W. (1987). Freeradical mechanisms in relation to tissue injury. Proceedings of the NutritionSociety, 46, 1–12.

Sun, H.-B., & Kawata, S. (2004). In N. Fatkullin (Ed.), NMR � 3D analysis �photopolymerization (pp. 169–273). Berlin: Springer.

P. Raudsepp et al. / Food Chemistry 199 (2016) 760–767 767

Tikekar, R. V., & Nitin, N. (2012). Distribution of encapsulated materials in colloidalparticles and its impact on oxidative stability of encapsulated materials.Langmuir, 28, 9233–9243.

Waraho, T., Cardenia, V., Decker, E. A., & McClements, D. J. (2010). Lipid oxidation inemulsified food products. Oxidation in foods and beverages and antioxidantapplications. In E. A. Decker, R. J. Elias, & D. J. McClements (Eds.).Management indifferent industry sectors (Vol. 2, pp. 306–343). Cambridge: WoodheadPublishing Limited.

Yoshida, Y., Itoh, N., Saito, Y., Hakayawa, M., & Niki, E. (2004). Application of water-soluble radical initiator, 2,20-azobis[2-(2-imidazolin-2-yl)propane]-dihydro-chloride, to a study of oxidative stress. Free Radical Research, 38, 375–384.

Yoshida, H., Kondo, I., & Kajimoto, G. (1992). Participation of free fatty acids in theoxidation of purifies soybean oil during microwave heating. Journal of theAmerican Oil Chemists Society, 69, 1136–1140.

Zhao, Y., Elias, R. J., & Coupland, J. N. (2015). Effect of food structure on thedistribution and reactivity of small molecules. Current Opinion in Food Science, 4,19–24.