Influence of gelation on the retention of purple cactus pear extract in microencapsulated double emulsions

A purple cactus pear (Opuntia ficus-indica) extract (CP) was encapsulated in double emulsions (DE) gelled with gelatin (DE-CP-G) and with gelatin and transglutaminase (DE-CP-GT), as well as in a DE with a liquid external aqueous phase (DE-CP), in order to study the retention of betanin as colorant agent. Both gelled DEs showed a predominantly elastic behavior, in contrast with DE-CP. The degradation rate constant of betanin was significantly higher in DE-CP-GT (90.2 x 10−3 days-1) than in DE-CP-G (11.0 x 10−3 days-1) and DE-CP (14.6 x 10−3 days-1) during cold-storage (4 °C). A shift towards yellow color was found in all the systems during cold-storage (4 °C) and after thermal treatment (70°C/30 min), especially in DE-CP-GT, denoting a higher degradation of betanin. Betalamic acid, cyclo-Dopa 5-O-β-glucoside, 17-decarboxy-betanin and neobetanin were identified by UHPLC-MS/MS as degradation products of betanin.


Introduction
Water-in-oil-in-water (W 1 /O/W 2 ) double emulsions (DEs) are multi-compartmentalized systems in which droplets of a water-in-oil (W 1 /O) emulsion are dispersed within an outer continuous aqueous phase (W 2 ) [1]. In recent years, there has been a growing interest in the incorporation of DEs in food applications, since they can be used as encapsulation and delivery systems for both hydrophilic and hydrophobic bioactive compounds due to their multi-compartmentalized structure. Furthermore, DEs have been proposed as a strategy for masking unpleasant flavors, developing reduced-fat and/or salt products and, by suitable formulation of the oil phase, improving the lipid profile [2].
Although DEs offer several advantages over simple oil-in-water emulsions, mainly the encapsulation, protection and release of both water-soluble and lipid-soluble compounds, as well as the reduction of the overall fat content of oil-in-water emulsions by loading the oil a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 phase with water droplets [1,3], they also show higher instability. Furthermore, plastic properties are desirable for DEs when they are used for replacing animal fat in food products, which is highly saturated and solid at room temperature, to mimic adipose tissue in its physical properties. However, DEs do not provide solid-like texture unless oil droplets are closely packed [4,5]. Gelling of the internal [3,6,7] or external aqueous phase [8][9][10][11][12][13], together with the structuring of the oil phase [3,14] are among the different strategies attempted for stabilizing DEs [15]. However, the gelation of the external aqueous phase, embedding the emulsion droplets in a continuous hydrogel matrix, is the most promising strategy to also provide a fat analog with solid-like properties [4]. Different compounds such as pectin, alginate, gellan gum, glycyrrhizic acid nanofibrils have been incorporated into the external aqueous phase of DE to form double emulsion gels [4,7,16]. Cold-set gelation is an attractive alternative for gelling DEs when heatlabile bioactive compounds are encapsulated [17]. In this context, the incorporation of microbial transglutaminase and gelatin into W 2 led to the formation of thermostable cold-set double emulsion gels [9,10,12,17]. Transglutaminase covalently binds different protein residues, leading to the formation of relatively thermostable gels [4].
Betalains are natural water-soluble pigments composed of two structural groups: betacyanins (reddish-purple) and betaxanthins (orange-yellow). Although several betalains sources have been described (amaranth, djulis, some cactaceas such as cactus pear, pitayas, pitahayas; [18], red beet (Beta vulgaris L.) is the main source of commercial betalains, which is classified as additive E-162 (EU) and 73.40 (FDA, USA). However, red beet has an earth-like flavour attributed to geosmin and 3-sec-butyl-2 methoxypyrazine, and besides it has high nitrate levels [19]. In this context, the purple cactus pear (Opuntia ficus-indica) is an interesting alternative as a source of betalains, where betanin is the main colourant compound, and indicaxanthin and isobetanin have been detected at a low level [20]. Furthermore, betalains have been associated with several beneficial effects on human health, such as protection against oxidative stress-related and cardiovascular diseases [18]. The stability of betalains is an important aspect to consider for using these bioactive compounds in foods, since they are affected by pH, concentration of buffer, water activity, light, oxygen, temperature and enzymatic activities [21,22]. Encapsulation technology could be a useful strategy to prevent betalains degradation in food matrices [23] (Khan, 2016). Among the different encapsulation technologies used for betalains encapsulation, spray-drying is one of the most commonly used [24]. However, the encapsulation of betalains in DEs, mostly from reed beet, is scarce [25][26][27] and, to the best of our knowledge, gelled DEs have not been evaluated as encapsulation strategy. Gelled DEs would allow widening the application of betalains as antioxidants and natural food colorants in food matrices where solid-like properties are desirable. The aim of this work was to study the stability of betanin from a purple cactus pear extract when they are encapsulated in the internal aqueous phase of DEs with a liquid or gelled external aqueous phase. Gelation of the external aqueous phase was addressed with gelatin (G) or a mixture of gelatin and transglutaminase (T).

Preparation of cactus pear extract
An ethanolic extract was prepared from the cactus pear pulp according to Sáenz et al. [20]. Briefly, the pulp was defrosted and macerated with ethanol:water (1:1 v/v) for 12 h, with ratio pulp:solvent 1:1 (w/w). The supernatant of the extraction was collected, and the sediment was re-extracted twice for 2 h. The supernatants of the three extractions were combined and concentrated in a rotary evaporator (R-100, Büchi, Switzerland) at 40˚C, giving a cactus pear extract (CP) with a betanin concentration of 0.5 mg/mL.
The betanin standard was obtained from the cactus pear extract according to Forni et al. [28]. Briefly, silica (70-230 Mesh ASTM, Merck) was activated at 150˚C for 1 h, allowed to cool for 15 min, packed in a glass column and conditioned with McIlvaine buffer (citrate phosphate buffer, pH 6.5). The cactus pear extract (6 mL) was applied in the column, and the McIlvaine buffer was used as mobile phase. The purple fraction was collected, and the isolation of betanin was ascertained by HPLC, according to Fernández-López and Almela [29], using a Merck Hitachi L-6200 pump, a Waters 996 photodiode-array detector and a C18 column (5 μm, 4.6 i.d. x 250 mm, Symmetry, Waters). The quantification of betanin in the fraction was performed spectrophotometrically (Orion Aquamate 8000, Thermo Scientific, USA) according to Stintzing et al. [30] by using this equation: Where A is the absorption value, calculated from the difference between absorption values at 538 nm and absorption values at 600 nm, DF is the dilution factor, and l is the path length of the cuvette (1 cm). MW and ε are the molecular weight and molar extinction coefficient of betanin, respectively (MW 550 g/mol; ε = 60 000 L/(mol cm) in H 2 O).

Preparation of DEs
Double emulsions were formulated following a two-step emulsification process [31] with slight modifications. The cactus pear extract with a betanin concentration of 0.5 mg/mL was used as the internal aqueous phase (W 1 ). A blend of olive, linseed and fish oils (70:20:10, w/ w/w) was used as lipid phase [31]. PGPR was added to the oil blend (6%, w/w), and a W 1 /O coarse emulsion was prepared by drop-wise addition of W 1 (20%) to the lipid phase (80%) using a blender (Thermomix, Vorwek, Germany; 3250 rpm, 5 min, 40˚C Bovine gelatin (4% w/w) and microbial transglutaminase (2% w/w) were dissolved in the W 2 of the freshly prepared DE-CP at 40˚C for 10 min [10]. DE-CP-G and DE-CP-GT were aliquoted in 50 mL tubes that were closed and chilled at 4˚C in darkness. After a 12 h overnight (day 0), the three emulsion systems (DE-CP, DE-CP-G and DE-CP-GT) were subjected to a thermal treatment (70˚C/30 min).

Characterization of DEs
Determination of droplet size and size distribution of DE-CP. The size and distribution of oil droplets in DE-CP were determined using a Malvern Mastersizer 3000 particle size analyzer (Malvern Instrument Ltd., Worcestershire, UK), equipped with a He-Ne laser (λ = 633nm) and a measurement range 0.02-2000 μm. Particle size calculations were based on the Mie Scattering theory. Ten measurements were performed on each emulsion. The results were expressed as the average of the volume-weighted mean globule size (D 4,3 ).
Confocal laser scanning microscopy. Morphology of freshly prepared DE with a liquid (DE-CP) or gelled (DE-CP-G and DE-CP-GT) external aqueous phase was evaluated by confocal laser scanning microscopy (CLSM, LSM 700, Carl Zeiss, Germany) at 40x magnifications. The oil phase of DEs was fluorescently labelled with Nile red (0.02% w/w). Samples were placed onto microscope slides, covered with a cover glass and subjected to 488 nm excitation wavelength, while fluorescent signal was collected at 580 nm. The software used for the CLSM imaging was ZEN 2012 (Blue Edition, Carl Zeiss, Germany).
Encapsulation efficiency of DE-CP. The EE of betanin was determined in DE-CP after preparation, by measuring the betanin concentration in W 2 spectrophotometrically. A 3-fold dilution of DEs in water with the adjusted osmolarity was centrifuged (2500 g, 30 min) to obtain W 2 that was filtered (0.22 μm, Millipore filter). Betanin was quantified as described for the concentrated extract, using a betanin standard calibration curve (0.2-6.7 mg betanin/mL; R 2 = 1). The EE was defined as the percentage of betanin in W 1 that remained in the primary emulsion W 1 /O after the second emulsification process, and it was calculated following the equation (Eq 2): Where Betanin W2 is the betanin concentration recovered in W 2 after preparation, Betanin 0T is the total concentration of betanin in the system after preparation, determined as described in the next section. The encapsulation stability (ES) of betanin was also determined at specific time intervals during storage at 4˚C. ES was defined as the percentage of betanin in W 1 during the storage, and it was calculated according to Eq 3: Where Betanin W2(t) is the betanin concentration recovered in W 2 at specific storage time (t), and Betanin t is the total concentration of betanin in the system at specific storage time (t), determined as described in the next section. Total content of betanin in DEs. The total content of betanin was determined in DEs with a liquid (DE-CP) or gelled (DE-CP-G and DE-CP-GT) W 2 at specific time intervals during storage time at 4˚C, and after the thermal treatment (70˚C/30 min) at day 0. For that, 2 g of each system were homogenized sequentially with methanol (2.8 mL), chloroform (2.8 mL) and distilled water (1.4 mL) by using a rotor-stator (T18 IKA, Germany) at 20,000 rpm for 15 s (each solvent). Afterwards, the samples were centrifuged (9,610 g for 30 min) at 4˚C. The supernatants were collected, and the total content of betanin was determined spectrophotometrically as described for the concentrated extract, using a betanin standard calibration curve (0.2-6.7 mg betanin/mL; R 2 = 1).
Rheology of DEs. Dynamic rheological analyses were performed on the three systems (DE-CP, DE-CP-G and DE-CP-GT), using a rheometer (HR-1 Discovery Hybrid Rheometer, TA Instruments, USA), equipped with a cone-plate geometry (1.008˚, 60 mm diameter, 27 μm gap for DE-CP) and a parallel-plate geometry (40 mm diameter, 500 μm gap for DE-CP-G and DE-CP-GT). Samples were analyzed after cold maturation at 5˚C for 18 h, and they were allowed to stand for 5 min before rheological analysis. Strain amplitude sweeps (from 0.01 to 10%) at a frequency of 1 Hz were performed to determine the linear viscoelastic (LVE) region. The strain amplitude was set at γ = 0.8%, within the LVE, in the three systems for both frequency and temperature sweeps. For frequency sweeps, the samples were subjected to stress that varied harmonically with time, at variable frequencies from 0.1 to 10 Hz and 20˚C. The viscous modulus (G"; Pa) in the case of DE-CP, and both the elastic modulus (G'; Pa) and G" in the case of DE-CP-G and DE-CP-GT, were plotted as a function of frequency. Furthermore, temperature sweeps were run in DE-CP-G and DE-CP-GT from 5˚C to 50˚C (2˚C/min), at a frequency of 1 Hz. G', G" and phase angle (δ) values were plotted as a function of temperature. Results were average of at least three measurements.
Degradation rate constants of betanin in DEs during storage. The aliquoted samples (DE-CP, DE-CP-G and DE-CP-GT) were stored at 4˚C in absence of light for 28 days (DE-CP-GT) and 49 days (DE-CP and DE-CP-G). Samples were withdrawn at specific time intervals (in triplicate), and the total betanin content was determined as described above (Total content of betanin in DEs). Data were fitted by a first-order kinetic model, ln C = ln C 0k (t) . Degradation rate constants (k obs ) were obtained from the slope of a plot of the natural log of the percentage retention of betanin vs. time.
Color and pH of DEs. Color measurements were performed with a portable spectrophotometer (MiniScan XE Plus, HunterLab, USA) on DE-CP, DE-CP-G and DE-CP-GT at specific time intervals during storage at 4˚C, and after the thermal treatment (70˚C/30 min) at day 0. Hue angle [h˚= (arctan b � /a � )] was calculated from a � and b � values using illuminant D 65 .
To determine pH, the samples (10 g) were homogenized with 90 mL of distilled water and pH values were measured by using a pH metro (HI 5221, Hanna Instruments, USA).
Identification of betanin degradation products in DEs. The samples were analyzed according to Herbach et al. [32] with some modifications, at the end of the storage period at 4 C and after the thermal treatment (70˚C/30 min) at day 0. An Ultra-High Performance Liquid Chromatography (UPLC) Ultimate 3000 RSLC system was used, coupled to a Linear Ion Trap Mass Spectrometer LTQ XL (Thermo Scientific, USA), with a C18 column (1.7 μm particle size, 3 mm i.d. x 100 mm; Kinetex, Phenomenex, USA). The mobile phase A was 0.1% formic acid in MiliQ water, and 0.1% formic acid in acetonitrile was the mobile phase B. Separation was performed at a flow rate of 0.3 mL/min, using a gradient program as follows: 5% B (1 min), 5% B to 80% B (19 min), 80% B (7 min), 80% B to 5% B (5 min). The mass detection was carried out using electrospray ionization (ESI), with the spray voltage set at 3 kV (at 300˚C). Detection was performed in full scan mode in the 100-1000 m/z range in positive mode. MS2 spectra were obtained using He for collision-induced fragmentation (CID) with a normalized collision energy of 35 units and detection of fragments in full scan mode.

Statistical analysis
All the experiments were performed in triplicate. Analysis of variance (ANOVA) and Tukey's multiple range tests were applied to determine statistical differences among samples, using Statgraphics Centurion XVI software (Statistical Graphics Corporation, USA).

Size and size distribution of DE-CP
The size distribution and D 4,3 values of DE-CP are shown in Fig 1a and Table 1, respectively. DE-CP showed a monomodal size distribution at day 0, with oil globule populations in the range 0.4-15.1 μm, and D 4,3 values of 2.56 μm. Larger oil droplets were reported for DEs with betanin (purified from red beet) encapsulated in W 1 (D 3,2 45.8 μm; [27]) or DEs with a spraydried beetroot juice encapsulated in W 1 (D 3,2 72.4 μm; [27]), both obtained by microchannel emulsification. In other study, where DEs with a beetroot juice encapsulated in W 1 were prepared by hybrid membrane premix emulsification system, a droplet size (D 3,2 ) of 20.4 μm was obtained [26]. Thus, several factors may affect both droplet size and size distribution in DEs, such as oil composition and viscosity of the phases, type and concentration of emulsifiers, processing conditions (homogenization), type and source of bioactive compounds encapsulated in W 1 [27,33,34].
Storage of DE-CP at 4˚C produced some changes both in oil globule size and oil globule size distribution. Larger oil globules, with sizes in the range 0.4-58.7 μm, were found at day 28 of storage. The shift towards larger sizes, may be due to coalescence of the oil globules during storage. The increase in size throughout the storage led to a slight but significant increase of  (Table 1), since D 4,3 is sensitive to the presence of large particles within a polydisperse system [34]. At the end of the storage, approximately 95% of the oil globule population had sizes below 10 μm, denoting high emulsion stability during storage.
As DE-CP is intended to be used as food ingredient, it was subjected to 70˚C for 30 min, to mimic a conventional thermal treatment used in the food industry. The heating (70˚C for 30 min) of DE-CP at day 0 did not involve noticeable changes in D 4,3 values (p>0.05; Table 1), denoting an excellent thermal stability. However, the heating of DEs from 7 days of storage at 4˚C onwards, led to the formation of small oil globule populations with smaller sizes and, therefore, to a slight but significant decrease (p�0.05) of D 4,3 values (Table 1). This reduction in size may be related to the breakup of the oil globules and the appearance of new smaller oil globules [27], with the consequent partial loss of W 1 and release of the cactus pear extract. These results agree with other studies [35,36], where DEs showed a decrease of thermal stability with storage time at 4˚C. Pagano et al. [27] also found a reduction in size of DEs with betanin encapsulated in W 1 when they were continuously stored at 25˚C and 60˚C for 7 days.  Fig 1d). The three systems showed the characteristic microscopic structure of DE, with oil globules containing smaller water droplets inside. Therefore, the gelation process did not affect the typical multi-compartmentalized structure of DE. However, the incorporation of both gelatin (Fig 1c) and gelatin/transglutaminase (Fig 1d) to the external aqueous phase of DE led to a higher packing of oil globules. Flaizt et al. [12] also reported that the incorporation of gelatin and transglutaminase to the continuous phase of DEs with hydroxytyrosol encapsulated in W 1 did not affect the typical DE compartmentalized structure, but oil globules were closely packed.

Rheology of DEs
Dynamic oscillatory shear tests were performed on DEs to characterize its viscoelastic properties. The dynamic mechanical spectra of DE-CP showed that G" values were higher than G' values all over the frequency range. Furthermore, G' values were very low and irregular and only G" values were plotted as a function of frequency (Fig 2a).
Therefore, DE-CP showed a viscous behavior within the frequency range studied, and G" showed high frequency dependence, in agreement with the results reported for DE in other studies [36,37]. A dominant viscous behavior was also reported in DEs formulated at the same  [36,37]. According to Russ-Murphy [38], high G" to G' ratios are typically found in dilute solutions, with both moduli showing a strong frequency dependence. In contrast, a dominant elastic behavior has been reported for DEs formulated at higher volumetric ratios of W 1 /O in W 2 (70/30 and 80/20; [5]. The behavior of DE-CP-G and DE-CP-GT was predominantly more elastic than viscous, since G' values were higher than G" values all over frequency range (Fig 2b). Furthermore, both G' and G" showed low frequency dependence, contrary to DE-CP, since the influence of frequency is reduced by increasing the structural strength of these systems [9]. According to the low values obtained for δ in both DE-CP-G and DE-CP-GT throughout the frequency range (δ � 5˚), both samples can be considered gels with a strong structure at 20˚C. The viscoelastic properties of the gelled DEs were also analyzed after cold maturation at 5 C for 18 h. Fig 2 shows the values obtained for G' and G" (Fig 2c) and δ (Fig 2d), during heating from 5 to 50˚C. DE-CP-G showed a melting temperature (G" > G') of 35.5˚C (Fig 2c), and high δ values at 50˚C (δ~66˚; Fig 2d). Under subsequent cooling from 50˚C to 5˚C, DE-CP-G showed a good refolding ability, reaching high G' values (G' = 1349 Pa) and low δ values (δ = 1.6˚) at 5˚C, denoting the thermoreversible property of this gelled DE (S1 Fig). As expected, the incorporation of gelatin and transglutaminase in the W 2 of DE led to the formation of a thermostable gel, as can be deduced from δ values found in DE-CP-GT throughout the heating range (δ < 10˚; Fig 2d). However, a ten-fold decrease in G' values was found during heating, probably due to the melting of the gelatin fraction not involved in the thermostable gel. In contrast, Freire et al. [9] found a slight decrease in G' values in thermostable DEs, gelled with gelatin (4% w/w) and transglutaminase (2% w/w), when they were heated from 20˚C to 80˚C; although these samples were analyzed after 30 days chilling, much longer time for enzyme activity. Table 2 shows the EE, ES and evolution of total betanin retention in DE-CP, DE-CP-G and DE-CP-GT during storage at 4˚C, as well as betanin degradation rate constants for the emulsion systems studied. Encapsulation efficiency of betanin in DE-CP was 87.7% at day 0.

Stability of betanin in DEs
Slightly higher encapsulation efficiency values (89.1%) were reported for DEs loaded with red beet extract and obtained with a high-speed mixer [25]. Eisinaite et al. [26] also reported higher EE than this study in food-grade DEs containing native (98.62%) and concentrate red beet juice (100%). Several factors could influence the EE in DEs, such as the nature of the bioactive compound (molecular weight, water solubility, polarity), the composition of DE and the emulsification method used [7,12]. The ES of betanin in DE-CP showed a slight decrease during the storage, from 87.7 to 80.9% (p�0.05). However, the ES values did not explain the internal changes of betanin content in DE-CP, since the content of betanin in both W 2 and W 1 showed a marked decrease during the storage (Table 2), due to the release and/or degradation of betanin from W 1 , and degradation of betanin in W 2 .
The degradation of total betanin followed first order reaction kinetics at 4˚C in all the studied systems. The same kinetic order was reported for the degradation of betanin in water solution (pH = 4) at 85˚C [39], degradation of betacyanins from Amaranthus at temperatures in the range 40-100˚C [40] and betanin from beet root in cow milk at 70, 80 and 90˚C [41]. Betanin degradation rate constants (k obs ) were obtained from the slope of a plot of the natural log of the betanin retention (%) vs. time (days), with coefficients of determination (R 2 ) above Table 2 [42]. Therefore, though DE-CP-GT showed higher thermal stability according to its rheological behavior (Fig 2c and 2d), it was not suitable for betanin retention.

. Encapsulation efficiency (EE), encapsulation stability (ES) and evolution of betanin total retention in DE
As mentioned above, these systems (DE-CP, DE-CP-G and DE-CP-GT) were subjected to thermal treatment (70˚C/30 min) after 12 h overnight (day 0) to mimic a conventional thermal treatment used in the food industry. A similar loss of betanin was found in DE-CP and DE-CP-G, reaching values of 40.6 ± 3.4% and 38.1 ± 5.6%, respectively. However, DE-CP-GT showed the highest betanin degradation, with 55.7 ± 3.2%. The higher betanin degradation in DE-CP-GT may be related to its pH values, since thermal stability of betanin is decreased at pH values outside the range 3-7 [21]. Several factors have been reported to affect the chemical stability of betanin, such as pH, water activity, oxygen, light, metals, antioxidant and temperature [21,22]. In this study, two isomers of betalamic acid (bright yellow,  Table 3).
The degradation of betanin through hydrolysis of aldimine bond, giving cyclo-Dopa 5-O-βglucoside and betalamic acid, has been proposed as the main pathway for decomposition and this may be favored in systems with high water activity such as W 1 /O/W 2 double emulsions [21,22,24]. Alkaline conditions, such as those in DE-CP-GT, have been also reported to bring out aldimine bond hydrolysis of betanin [21]. In other study, the same degradation products of betanin were identified at both pH 2 and pH 9, with a difference in their formation rate [22]. Furthermore, betalamic acid, cyclo-Dopa 5-O-β-glucoside and neobetanin have been identified as thermal degradation products of betanin in juices from purple pitaya (Hulocereus polyrhizus) or red beet [32,43].  In other study, a high correlation was reported between betalains content from beetroot and both L � and chroma values in colored milk during thermal treatment [41].

Color stability of DEs
When the systems were subjected to thermal treatment (70˚C/30 min) after a 12 h overnight (day 0), a significant increase (p�0.05) in hue angle values was observed in all the systems, indicating a shift towards the yellow region. A similar increase in hue angle values was found in DE-CP and DE-CP-G, reaching values of 14.5 ± 1.7˚and 16.5 ± 0.8˚, respectively; whereas DE-CP-GT reached values of 43.3 ± 2.4˚. This increase in hue angle values was in agreement with the betanin degradation found due to thermal treatment, higher in the case of DE-CP-GT (55.7 ± 3.2%). Pagano et al. [27] reported that betanin from different sources, encapsulated in W/O/W emulsions, was also temperature sensitive, and a decrease of a � values together with an increase of b � values were reported with increasing temperature from 4˚C to 60˚C. These authors suggested that the shift towards yellow colors was probably due to the formation of betalamic acid, decarboxylated derivatives or oxidation products [22], but an identification of degradation products of betanin was not carried out. In our study, yellow Influence of gelation on the retention of purple cactus pear extract in microencapsulated double emulsions (betalamic acid and neobetanidin 5-O-β-glucoside) and orange-red (17-decarboxy betanin) degradation products of betanin were identified in the samples subjected to 70˚C for 30 min (Table 3), explaining the shift towards yellow colors. These compounds were also reported as degradation products of betanin in red beet juice subjected to thermal treatment at 85˚C, explaining the color shift towards yellow observed in the juice [32]. In contrast, Eisinaite et al. [26] reported that the red color (evaluated by a � ) of DE with beet root juice as W 1 , was stable after thermal treatment at 70˚C and 75˚C, although these treatments were applied only for 5 min; while significantly decreased after thermal treatment at 85˚C and 90˚C for 5 min.

Conclusions
The incorporation of gelatin or a mixture of gelatin and transglutaminase in the external aqueous phase of a DE, with a purple cactus pear extract encapsulated in the internal aqueous phase as colorant agent, led to the formation of cold-set gelled DEs with solid-like behavior, which is desirable for certain food applications of DEs such as replacement of animal fat, without affecting the typical multi-compartmentalized structure of DE. Betanin from cactus pear gave red-purple color to DEs with a liquid or gelled external aqueous phase. However, the gelling agent influenced the stability of betanin and color. In the case of DE-CP-GT, higher pH values favored the degradation of betanin during cold storage and decreased their thermal stability, leading to a shift towards yellow colors. Betalamic acid, cyclo-Dopa 5-O-β-glucoside, 17-decarboxy-betanin and neobetanin were identified by UHPLC-MS/MS as degradation products of betanin. DEs with an external aqueous phase gelled with gelatin protected betanin of a purple cactus pear extract from degradation, and, given their solid-like properties, were promising to be used as potential fat replacers in food products consumed at temperatures under the gelatin melting point.