Formation of the oxidized flavor compounds at different heat treatment and changes in the oxidation stability of milk

Abstract The oxidized flavor could affect the sensory acceptability of consumers. In this study, the oxidized flavor compound (OFC) in different heated milk was analyzed using the solid phase microextraction–gas chromatography–mass spectrometry (SPME–GC–MS). The concentration of OFC increased with increasing the intensity of heat treatment. The concentrations of heptanal, nonanal, 2‐heptanone, and 2‐nonanone in the heated milk samples were in the range of 1.88–5.51, 1.03–3.26, 6.89–88.04, and 1.46–12.78 μg/kg, respectively. The correlation coefficients between the OFC and the heat intensity were above 0.86. It was found that the intensity of heat treatment (below 80°C for 10 min) could not cause the significant changes in the distribution of fat globules. The contents of partial proteins in milk fat globule membrane gradually destructed with increasing the intensity of heat treatment. The DPPH˙ scavenging activity reduced by 23.15%, and the peroxide value (POV) increased by 37.23% as raw milk was heated at 90°C for 20 min. Values of thiobarbituric acid reactive substances (TBARS) had a tendency to increase as the heated time was operated for 20 min (p < 0.05). It indicated the heat treatment could change the oxidative environment of milk and influence the distribution of milk fat and the formation of oxidized flavor compounds.

illumination, drugs, and spoilage also could contribute the different off-odors in milk.
Heat treatment was an indispensable process in the production of dairy products. The heat intensity could be monitored by adjusting heating temperature or heating time. During the heating process, the flavor compounds and the flavor precursors in milk changed from different physicochemical reactions, which made the flavor compounds in the heated milk different from raw milk (Bendall, 2001;Contarini, Povolo, Leardi, & Piero, 1997). The main oxidized flavor precursor in milk was milk fat, and it was the key factor to decide the flavor of milk. Milk fat could be oxidized and decomposed during the process of milk heating, which resulted from the release of fatty acids and formed lots of flavor compounds from the main reactions of the automatic oxidation, dehydration, decarboxylation, reduction, and hydrolysis (Li & Wang, 2016;Li, Zhang, Wang, & Han, 2013;Nursten, 1997). Among the flavor compounds, the volatile aldehydes and ketones were the important components of the oxidized flavor in milk. In the previous studies, the aldehydes and the ketones were also evaluated as the main flavor compounds in the concentrated milk, distillate, and milk powder (Li & Wang, 2016;Li et al., 2013).
Compared with ultra-high temperature (UHT) milk and milk powder, few studies focus on the flavor compounds in pasteurized milk. Contarini et al. (1997) pointed out that the concentrations of 2-butanone, 2-pentanone, toluene, hexanal, and 2-heptanal in milk (heated at 78 °C for 15 s) were significantly different from the UHT milk (heated at 144 °C for 3.5 s) and bottled milk (heated at 120 °C for 30 min). Vazquez-Landaverde, Velazquez, Torres, and Qian (2005) showed that the concentrations of 2-hexanone, 2-heptanone, 2-nonanone, 2-undecanone, 2-methyl aldehyde, 3-methyl butyral, heptanal, and decanal in UHT milk were higher than those in raw milk or in pasteurized milk. It also pointed out that levels of methyl ketones in UHT milk increased with the increase in fat content. Lloyd, Hess, and Drake (2009) reported that the concentrations of hexanal, 2-heptanone, heptanal, octanal, nonanal, 3-methyl aldehyde, and 2-methyl aldehyde in milk powder increased significantly after storage of 12 months. The related research also pointed that the oxidized off-odor occurred obviously as the concentration of hexanal was above 600 μg/kg or the concentration of nonanal was above 4 μg/kg in whole milk powder (Lloyd, Drake, & Gerard, 2009).
Hall pointed that lipid oxidation influenced the flavor during storage of milk powder and straight-chain aldehydes were the main flavor compounds, such as hexanal, heptanal, octanal, and nonanal Hall, Andersson, Lingnert, & Olofsson, 1985).
However, change in the oxidized flavor of heated milk was ignored although the flavor compounds would be quite different due to different heated intensity, which was essentially operated in the manufacture of low-pasteurized milk, high-pasteurized milk, or other dairy products. The natural fat globules in bovine milk are coated with a protective layer generally known as the milk fat globule membrane (MFGM). It markedly had the shielding effects on the milk lipid and could affect the lipid oxidation in milk (Mather, 2000). The study aimed at tracking the oxidized flavor compounds in different heated milk. It also explicated the destruction of milk fat as the main flavor precursor and the changes in the natural antioxidant capacity and oxidation susceptibility of different heated milk. It could provide the important theoretical and practical implications in the flavor quality of dairy products.

| Heating treatment
Raw milk (RM) obtained from a local dairy plant named XINGFU farm was added 0.02% (w/v) NaN 3 to prevent bacterial growth.

| Milk component analysis
Milk fat was determined by the Rose-Gottlieb method and total protein by the Kjeldahl technique with a factor of 6.38 as described by the previous study (Guinee, Auty, & Fenelon, 2000). Total solids of RM were calculated according to the weight loss by drying the samples at 105 ± 1°C (Almeida, Tamime, & Oliveira, 2009).

| Determination of volatiles
The volatiles in the headspace of the milk samples (raw milk, heated milk) were extracted and analyzed using SPME-GC-MS according to the previous studies (Li & Wang, 2016;Li et al., 2013). Heptanal, nonanal, 2-heptanone, and 2-nonanone, as selected oxidized volatiles, were identified by the NIST-02L GC-MS spectrum library and the retention time of their standard chemicals (Sigma, USA). The area of flavor compounds and internal standard could be provided by the GC-MS. The concentrations of individual compounds were calculated using the area ratio of flavor compounds and internal standard.
The odor activity value (OAV) of different volatiles was calculated by the ratio of the concentration and corresponding flavor threshold value. Only the volatile whose OAV was above 1.0 would be felt and contribute to the flavor (Qian & Reineccius, 2003).

| Isolation and heating the MFGM material
Raw milk was heated at 70, 80, and 90°C for 2 min, respectively.
After heating, the sample was cooled immediately to room temperature in the ice bath. Then milk sample was centrifuged at 15,000 g for 20 min at 20°C in a temperature-controlled centrifuge (Sigma 3-30k, Germany), and the cream layer was removed and stored at 4°C. Then, the supernatant cream was washed with simulated milk ultrafiltrate (SMUF). The SUMF solution was prepared according to the literature (Lee & Sherbin, 2002;Le, Van Camp, Rombaut, van Leeckwyck, & Dewettinck, 2009). The above steps were operated for three times.
The washed cream was resuspended in SMUF to yield a suspension with a fat content similar to that of the original milk.

| Determination of SDS-PAGE electrophoresis
The individual proteins in the washed cream were determined by SDS-PAGE with some modification of the reference (Ye, Singh, Oldfield, & Anema, 2004). The washed cream was dispersed (1:2 w/w) in 0.5 M Tris-HCl buffer, containing 10% glycerol, 2% (w/v) SDS, and 0.05% bromophenol blue. The gels were run in a Mini-Protean Tetra Electrophoresis System (Bio-Rad Laboratories, Hercules, CA, USA). The proteins of the MFGM were identified by comparing with the marker (Hercules, CA, USA).

| Distribution of milk fat globules
Size and specific surface area (SSA) of fat globules in milk samples were determined using a particle size analyzer (ZetaPLAS, Brookhaven Instruments Corporation, USA) at 40°C water bath. The milk sample containing 2% (w/v) SDS and 50 mM EDTA was dispersed (Lee & Sherbin, 2002;Ye et al., 2004;Ye, Singh, Taylor, & Anema, 2002). The ratio of refractive index of fat globule was 1.456. Value of obscuration ranged from 8% to 10%. Running time was 60 s. Globule size was expressed as d 32 , and volume surface average diameter in μm. Average fat globule diameters were calculated in duplicate.

| Peroxide value
Peroxide value (POV) was used to analyze the oxidation of milk. POV analysis was based on the procedure of Smet et al. (2009).

| Value of thiobarbituric acid reactive substances
Value of thiobarbituric acid reactive substances (TBARS) was determined based on the methods (Sun, 2013;Zhou, 2009).

| Statistical analysis
Analyses were duplicated for the SPME-GC-MS expressed as mean, and triplicated expressed as mean ± SD for others. One-way analysis of variance and univariate analysis of variance (ANOVA) for determining the effect of the processes were carried out using PASW Statistics 18.0 software (SPSS Inc, USA). Trend regression analysis was applied to analyze the parameters of oxidized volatiles by the intensity of heat treatment.
The flavor compounds in milk samples were quantified using the area ratio of volatiles and internal standard. Eighteen volatiles were identified from the samples, including three kinds of aldehydes, two kinds of ketones, two kinds of acids, five kinds of hydrocarbons, three kinds of benzenes, and three kinds of other compounds.

| Changes in oxidized flavor compounds
Levels of heptanal and nonanal in raw milk and heated milk are shown in Figure 2, which increased with increasing the heating intensity. Concentration of heptanal in the heated milk at 90°C for 5 min was 3.57 μg/kg and that was different in the heated milk at

| Dependency relationship of OFC and the heating intensity
The dependency relationship between the concentration of OFC and the regression function of heating temperature and time is shown in Table 1. A single parameter of oxidized flavor compound (z, including heptanal, nonanal, 2-heptanone, and 2-nonanone) was used as the dependent variable. Heating temperature (×°C) and heating time (y, min) and their quadratic term respectively were used as independent variables to get the dependency relationship.
The results showed that two quadratic polynomial could increase the determination coefficient R 2 . The stepwise regression method was used to determine the model, and F value was less than 0.05 as the standard. The coefficient of determination was 0.868-0.936, which indicated that the prediction could explain the relationship between heating intensity and the corresponding concentration of OFC in the heated milk. The increase of methyl ketone was related to the β-oxidation of β-hydroxyl fatty acids and to the decarboxylation of β-keto acids in the saturated fatty acids. The increase of aldehydes was related to the oxidation of unsaturated fatty acids (Nursten, 1997).   (Lopez, 2011;Mather, 2000). It

| Changes in the MFGM proteins
showed that the composition of MFGM protein changed gradually with the increase in heating intensity. The bands of β-lactoglobulin (β-LG) and α-lactalbumin (α-LA) appeared from the isolate of MFGM proteins, and the protein bands were the deepest at 90°C. That means the MFGM proteins could react with β-LG and α-LA as raw milk was heated above 80°C. The reaction probably resulted from the intermolecular and intramolecular thiol reactions among XDH/ XO, BTN, β-LG, α-LA in the milk system (Ye et al., 2002(Ye et al., , 2004. On the other hand, the polypeptides existed in MFGM protein contained a large number of cysteine residues, which might also participate in the reactions with whey protein as milk was heated (Kim & Jimenez-Flores, 1995). That could decrease the shielding effects of MFGM protein and cause the accumulation of fat particles.

| Sizes distribution of milk fat globules
The diameter of fat globules in raw milk ranged from 1.15 to 7.42 μm along with a slightly normal distribution (data was not shown). Values of skewness and kurtosis were 0.3991 and −0.3971, respectively.
Compared with the previous studies, it was indicated that the diameter of milk fat globule ranged from 0.1 to 10 μm and from 1.0 to 10 μm, respectively (Le et al., 2009;Lee & Sherbin, 2002). In this study, the diameter (d 3,2 ) of the milk fat globules in different heated milk is shown in Figure 5. It was not significantly different among the group of raw milk (d 3,2 , 3.04 μm) and the group of heated milk except the samples heated at 90°C for 10 min and for 20 min (p > 0.05). The specific surface area of milk fat globules among different milk samples is also shown in Figure 6. The SSA was from 19,815 to 19,959 cm 2 /ml after heating at 70°C for 0.5-20 min (p > 0.05). After heating at the relative high heating intensity, such as heating at 80°C or 90°C for 10 min or 20 min respectively, the SSA of milk fat globules in the heated milk decreased significantly (p < 0.05). Raikos, Kapolos, Farmakis, Koliadima, and Karaiskakis (2009) found that the average size of milk fat globule was 0.861 μm as the whole milk was homogenized at 50°C and the size was up to 0.901 μm after heating at 125°C (p < 0.05). It was contradicted by the previous result that heat treatment could not change the size distribution of milk fat (van Boekel & Folkerts, 1991). In this study, the diameter (d 3,2 ) of the milk fat globules increased significantly as the heating intensity raised to 90°C for 10 min (p < 0.05). The analysis of size distribution as affected by different heating treatments is F I G U R E 5 Difference in diameter (d 3,2 ) and specific surface area (SSA) of fat globules among heated milk. Note: Bars with different letters were significantly different from each other (p < 0.05) F I G U R E 6 DPPH˙ scavenging activity of different heated milk shown in Table 2. That also indicated the low intensity of heating treatment (heating temperature was below 70°C or heating time was below 5 min) could not change the diameter (d 3,2 ) and SSA of milk fat globules in milk (p > 0.05). As the raw milk was heated above 80°C, the heat treatment influenced the distribution of diameter (d 3,2 ) and SSA significantly and the p values were 0.046 and 0.002, respectively. On the other hand, it was effective as heating time prolonged above 10 min (p < 0.003). Figure 6 shows the DPPH˙ scavenging activity of raw milk and the heated milk. The DPPH˙ scavenging activity of raw milk was 54.74%, which reasoned from the antioxidants, such as fat-soluble vitamin A and vitamin E, transition metal ions, and water-soluble vitamin C and vitamin B (Soberon, Liu, & Cherney, 2012). The value of DPPH˙ scavenging activity in raw milk was close to the previous study, which

| Determination of the DPPH˙ scavenging activity in milk
showed that the ability of raw milk to remove DPPH was 53% (Liu, Chen, & Lin, 2005   As a result, the free fat is formed in milk and the values of POV and TBARS increased following the formation of oxidized flavor. These result indicated the distribution of milk fat during the course of heating could influence the formation of oxidation flavor.

ACK N OWLED G M ENTS
The authors acknowledge the financial support from the Natural Science Foundation of Zhejiang Province (Project Number

E TH I C A L S TATEM ENT
There is not any human or animal testing involved in this study.