Application of tea polyphenols as additives in brown fermented milk: Potential analysis of mitigating Maillard reaction products

Brown fermented milk (BFM) is favored by consumers in the dairy market for its unique burnt flavor and brown color. However, Maillard reaction products (MRP) from high-temperature baking are also noteworthy. In this study, tea polyphenols (TP) were initially developed as potential inhibitors of MRP formation in BFM. The results showed that the flavor profile of BFM did not change after adding 0.08% (wt/wt) of TP, and its inhibition rates on 5-hydroxymethyl-2-furalde-hyde (5-HMF), glyoxal (GO), methylglyoxal (MGO), N ε -carboxymethyl lysine (CML), and N ε -carboxyethyl lysine (CEL) were 60.8%, 27.12%, 23.44%, 57.7%, and 31.28%, respectively. After 21 d of storage, the levels of 5-HMF, GO, MGO, CML, and CEL in BFM with TP were 46.3%, 9.7%, 20.6%, 5.2%, and 24.7% lower than the control group, respectively. Moreover, a smaller change occurred in their color and the browning index was lower than that of the control group. The significance of this study was to develop TP as additives to inhibit the production of MRP in brown fermented yogurt without changing color and flavors, thereby making dairy products safer for consumers.


INTRODUCTION
Brown fermented milk (BFM), also known as brown yogurt, is a fermented beverage made of raw milk or milk powder, which is processed by a long-time hightemperature treatment (95-100°C, 2-3 h) to cause a Maillard reaction and then fermented by lactic acid bacteria (Han et al., 2019).Because of their unique burnt flavor and brown color, these dairy products are favored by consumers (Yu et al., 2020).
The Maillard reaction is widely found in the food industry and contributes significantly to flavor, odor, and color of foods.Studies have shown that the Maillard reaction can produce over 2,500 different flavor compounds, among which pyrazines and furans are considered as ideal flavors with roasted, coffee, nutty, caramel, and chocolate odor descriptions (Scalone et al., 2020).However, potentially unhealthy Maillard reaction products (MRP) can also be generated during heat treatment.The main reported undesirable MRP are 5-hydroxymethyl-2-furaldehyde (5-HMF), glyoxal (GO), methylglyoxal (MGO), N ε -carboxymethyl lysine (CML), and N ε -carboxyethyl lysine (CEL; Li et al., 2022c).Among these, 5-HMF is associated with cytotoxicity, indirect mutagenicity, carcinogenicity, hepatotoxicity, and nephrotoxicity (Capuano and Fogliano, 2011;Zhang et al., 2021); GO and MGO react with nucleophilic sites of proteins to form advanced glycosylation end products (AGE) related to neurological diseases, diabetes, and other diseases (Hellwig and Henle, 2014), and CEL and CML are frequently used as indicators of AGE (Han et al., 2019).Therefore, the development of a new food additive to more effectively reduce the accumulation of MRP in BFM would be a promising research direction.
"Tea polyphenols" (TP) is a general term for the phenolic substances in tea leaves, consisting of more than 30 kinds of phenolic substances, the main components of which are catechins and their ramification.Tea polyphenols have physiological antioxidant, antiaging, hypoglycemic, antibacterial, and enzyme inhibitory effects (Wu et al., 2020;Luo et al., 2022).Studies have demonstrated that polyphenols can successfully prevent and modify the Maillard reaction, thus reducing the production of harmful MRP (Culetu et al., 2016).For example, TP reduced the production of francolin by inhibiting the glycosylation reaction between ovalbumin and glucose (Cömert et al., 2017), and inhibited the formation of furan during sterilization of canned coffee (Bi et al., 2017).Moreover, catechins showed significant hypoglycemic effects by trapping α-dicarbonyl compounds (α-DC) and blocking the formation of AGE (Luo et al., 2022).Green tea extract enriched with epigallocatechin gallate was reported to have a 95% reduction in Strecker aldehyde inhibition in lactose hydrolyzed milk.Meanwhile, the release of free amino acids caused by proteolysis in milk during storage was reduced by the addition of green tea extract either before or after UHT treatment (Jansson et al., 2017).However, the inhibitory effect of TP on the MRP in BFM has not been elucidated.
In this study, we focused on the inhibition effect of TP on 5 kinds of MRP, verified the flavor of BFM containing TP, and finally examined the changes of MRP and colors during 21 d of storage at 4°C.These experimental results provide theoretical data to support the quality control of BFM products.

MATERIALS AND METHODS
Because no human or animal subjects were used, this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.

Materials
Mixed raw milk (Holsten Friesian) from different lactation periods was obtained from Zhongcheng Trading Co. Ltd. (Tangshan, China).Skim milk powder and butter for standardization were provided by Fonterra Co-Operative Group Limited (Auckland, New Zealand).Tea polyphenols (≥80%; catechins account for 74.2% of polyphenols, and epigallocatechin gallate accounts for more than 30% of catechins) were supplied by Yuanye Biotechnology Co. Ltd. (Shanghai, China).Commercial starter (a mixture of Lactobacillus delbrueckii ssp.bulgaricus and Streptococcus salivarius ssp.thermophilus) was obtained from Chr. Hansen Co. Ltd. (Beijing, China).A CML-AGE ELISA Kit and a CEL-AGE ELISA Kit were obtained from Shanghai Enzyme-linked Biotechnology Co. Ltd. (Shanghai, China).All other chemicals were of analytical grade.

Preparation of Brown Fermented Milk
To avoid biological differences between cows in different lactations, raw milk was standardized.The protein, fat, and carbohydrate contents of standard milk were 3.0%, 3.6%, and 4.6%, respectively.The standardized milk was then pasteurized (65°C, 30 min) and prepared for use.Different contents (0.01%, 0.02%, 0.04%, 0.08%, 0.1%, and 0.2%, wt/wt) of TP and glucose (4%, wt/wt) were blended with milk base and homogenized (TW-Basic, Wardy Company, Shanghai, China) at 20 MPa for 10 min.The mixture was then browned at 115°C for 15 min using an autoclave (YXQ-LS-50SII, Bosun Industrial Co., Shanghai, China) and immediately cooled to 42°C with cold water.Commercial starter (0.04%, wt/wt) was inoculated into the baked milk bases and fermented at 42°C until pH reached 4.6.After fermentation, the baked milk was quickly stirred to crush the fermented yogurt curd, which was subsequently cooled in cold water and finally put into the refrigerator at 4℃ for storage.Brown fermented milks with or without TP were designated TP-BFM and NTP-BFM, respectively.Samples were taken on d 0, 7, 14, and 28 of the storage period for testing.All samples were analyzed in triplicates.

Determination of Maillard Reaction Products
Maillard reaction products were determined according to the methods previously established in our laboratory (Li et al., 2022a).Quantification of 5-HMF, GO, and MGO was performed using an Agilent-1260 liquid chromatographic system (Agilent, Palo Alto, CA) coupled with a 1260 Bin Pump VL (Agilent).The separation column was an Agilent Zorbax Eclipse xdb-C18 column (250 mm × 4.6 mm, 5 µm), and the column temperature was maintained at 30°C for 5-HMF and 25°C for GO and MGO.The injection volume was 5 µL, and the detection wavelength was set to 284 nm for 5-HMF and 314 nm for GO and MGO.For 5-HMF, the mobile phase was 10% acetonitrile-water solution (vol/vol), and the flow rate was 0.6 mL/min.For GO and MGO, mobile phase A was 0.1% (vol/vol) formic acid aqueous solution, and mobile phase B was 100% acetonitrile.The flow rate was 0.6 mL/min.The elution gradient was carried as follows: 25% B from 0 to 10 min, 25-80% B from 10 to 15 min, and 80-25% from 15 to 17 min.ChemStation analysis software, installed with the Agilent-1260 instrument, was used for data acquisition.
The concentrations of CML and CEL were determined using an ELISA Kit following the manufacturer's instructions.The detection wavelength was set to 450 nm.A standard curve of concentration (µg/kg) versus optical density values was plotted using the standards provided in the kit, and the contents of CML or CEL in BFM samples were calculated.

Volatile Compound Analysis
The volatile components of BFM were examined by headspace solid-phase microextraction and gas chromatography-mass spectrometry (GCMS-QP2010, Shimadzu, Japan; Chen et al., 2020).Briefly, a 5-g sample was transferred into a 15-mL glass threaded vial containing 1.2 g of sodium chloride and subsequently incubated in a water bath at 45°C for 30 min.Then, the desorbed volatile chemicals were added back into the GC inlet after being extracted using fibers for 10 min.The GC program was conducted as follows: the initial oven temperature was maintained at 40°C for 3 min, and subsequently a linear increase of temperature was started from 5 to 180°C at a rate of 5°C per minute (hold for 1 min), and from 180°C to 250°C at a rate of 7°C per minute (hold for 5 min).The purge flow was increased to 2.0 mL/min after the injection, which took place at 230°C in the transmission line.The following are the MS symptoms: the ionization mode is an electron impact ion source with an electron energy of 70 eV, and the scanning range is 30.00 to 500.00 m/z.The ion source temperature and interface temperature were each 220°C.The solvent delay, emission current, and detecting voltage were 1.5 min, 100 A, and 1.4 kV, respectively.

Color Analysis
Color parameters (L*, a*, and b*) of the samples were measured using a CM-26 dG chromatic meter (Shenzhen 3NH Technology Co. Ltd.; Roldan et al., 2015).The total chromatic aberration (ΔE*) and browning index (BI) were calculated by the following formulas (all analyses were performed in triplicate):

Statistical Analysis
All experiments were conducted in triplicate, and results were presented as mean ± standard deviation of 3 replicates.Statistical analysis was performed by one-way ANOVA using SPSS 24 (IBM Corp., Armonk, NY).The significant difference (P < 0.05) was determined using the Duncan multiple range test.The raw files of the mass spectrometry tests were searched with Max Quant (version 1.5.5.1, https: / / www .maxquant.org/maxquant/ ) for the corresponding database, and quantitative analysis results were obtained.

Inhibitory Effect of Tea Polyphenols on Maillard Reaction Products in Brown Fermented Milk
According to the different stages of the Maillard reaction, MRP can be divided into initial, intermediate, and advanced glycosylation products (Sunds et al., 2018).Furfural, GO, and MGO are products of the intermediate stage of the Maillard reaction, where 5-HMF can be used to evaluate milk heat treatment damage (Li et al., 2022b).Both CML and CEL are usually considered the products of the advanced stage of Maillard reaction (Erbersdobler and Somoza, 2007).In a previous study, we detected these 5 MRP in BFM (Li et al., 2022a).Because TP could mitigate the formation of harmful MRP, we selected TP as a food additive for BFM.
The inhibitory effect of different concentrations of TP (0.01%, 0.02%, 0.04%, 0.08%, 0.1%, and 0.2%, wt/ wt) on MRP in BFM were evaluated and are shown in Figure 1.All levels of addition of TP could effectively inhibit the production of MRP, but the best inhibition was achieved at the concentration of 0.08%.The contents of 5-HMF, GO, MGO, CML, and CEL in BFM without TP were 3.78 ± 0.44 mg/kg, 1.23 ± 0.05 mg/ kg, 7.37 ± 0.67 mg/kg, 95.47 ± 1.10 µg/kg, and 74.85 ± 1.36 µg/kg, respectively.The content of 5-HMF decreased with the increase of TP addition, reaching the lowest value of 1.48 ± 0.18 mg/kg at the level of 0.08%, when the inhibition rate was 60.8% (Figure 1a).This was due to the direct reaction of TP with C2, C3, and C4 of the sugar molecule in the early stage of the Maillard reaction leading to the formation of tea polyphenol-glycan fragment adducts, which block the binding of sugar to amino acids (Totlani and Peterson, 2005).As shown in the Figure 1b and Figure 1c, when the addition amount was 0.08%, the contents of GO and MGO were 0.90 ± 0.08 mg/kg and 5.62 ± 0.18 mg/kg, and the inhibition rates were 27.1% and 23.5%, respectively.It was generally believed that TP can trap α-DC (such as 3-deoxyglucuronide) by aromatic electrophilic substitution, thus preventing their further reaction (Totlani and Peterson, 2006).In addition, with the reduction of α-DC production, the CML and CEL synthesis pathways were blocked, which ultimately inhibited AGE formation (Zhu et al., 2020).In addition, the dehydration of 3-deoxyglucuronide also led to the formation of 5-HMF.For CML (Figure 1d) and CEL (Figure 1e), the maximal inhibition rates were 57.7% (40.33 ± 1.01 µg/kg) and 31.3%(51.41 ± 1.09 µg/ kg), respectively.We observed a positive correlation with the inhibition rate when the TP concentration ranged from 0.01% to 0.08%.However, when the addition amount was in the range of 0.08% to 0.2%, the inhibition rate of TP on the MRP showed a decreasing trend, except for GO.This might be due to excessive TP binding to the substrate, which in turn reduced the capture efficiency of TP.The trapping mechanism of TP played an important role in its inhibition of MRP, especially for the intermediate products 5-HMF and α-DC, and the formation of adducts further suppressed the content of MRP, which was consistent with previous studies (Cheng, 2010).

Volatile Compounds Analysis in Brown Fermented Milk with Tea Polyphenols
Flavor is one of the most critical qualities of food products and is a key factor in determining consumer choice and acceptance.The unique flavor of BFM mainly comes from lactic acid produced by milk fermentation and aromatic compounds generated from the Maillard reaction (Liu et al., 2014).
The volatile compounds in BFM were analyzed to study the effect of TP on the flavor of yogurt.As shown in Table 1, a total of 40 flavor compounds were detected in the 2 samples, of which 35 were common to BFM.Acetone in these compounds has a sweet, fruity aroma and is known to affect the aroma and flavor qualities of yogurt.3-Hydroxybutanone (acetoin) is an important aroma compound associated with the light creamy flavor of yogurt.It is formed by the fermentation of the citrate in milk (Papaioannou et al., 2021).In TP-BFM, the content of 2-nonanone was found to be 4.86 times higher than in NTP-BFM, which had a positive effect on the spicy, cinnamon, and fruity taste in brown fermented milk (Zhao et al., 2020).The increased content of other ketones in TP-BFM, such as 2,3-pentanedione and 2-undecanone, gave TP-BFM more vanilla, milder scent as well as rose and herbaceous flavors.
Acids are important compounds produced by the fermentation of lactic acid bacteria, among which acetic Different letters in the same column indicate significant differences at P < 0.05. 1 Results are expressed as mean ± SE. Brown fermented milk with or without tea polyphenols (TP) are respectively designated TP-BFM and NTP-BFM.ND indicates that the volatile compound was undetected in the product.
acid gives yogurt a vinegary flavor; butyric acid gives a cheese flavor, and capric acid has a floral aroma.Most esters have fruity and floral aromas that can promote aroma and flavor by reducing the sharpness and bitterness associated with fatty acids and amines (Güler, 2007).The content of diethylhexyl adipate was found to be higher in NTP-BFM (5.09%).However, the content of acetaldehyde in TP-BFM was 1.69 times higher than that in control group.Acetaldehyde, which produces a fresh, pungent flavor, is one of the substances that constitute the unique flavor of yogurt, and high-temperature treatment of milk is conducive to the production of acetaldehyde (Saint-Eve et al., 2008).

Content and Evolution of Maillard Reaction Products in Brown Fermented Milk During Storage
Both NTP-BFM and TP-BFM were stored at 4°C for 21 d, and MRP, including 5-HMF, GO, MGO, CML, and CEL, were detected on d 1, 7, 14, and 21.As shown in Figure 2, MRP in both samples increased with the prolongation of storage time, but the content of MRP in TP-BFM was significantly lower than that of NTP-BFM, indicating that TP can effectively reduce the production of MRP during the storage periods, thus ensuring the quality of BFM.

Content and Evolution of 5-Hydroxymethyl-2-Furaldehyde
An intermediate of the Maillard reaction, 5-HMF is produced by degradation of Amadori rearrangement products under acidic conditions or direct dehydration of sugars in the product (Ramírez-Jiménez et al., 2000).As shown in Figure 2a, the contents of 5-HMF in NTP-BFM and TP-BFM on d 1 of storage were 1.63 ± 0.37 mg/kg and 1.61 ± 0.29 mg/kg, respectively.The lactic acid bacteria in BFM typically continue to ferment while being stored at 4°C, causing post-acidification and a decrease in pH (pH ≤4.6).Moreover, studies have shown that low pH is conducive to the production of 5-HMF (Zhang et al., 2019).Therefore, the content of 5-HMF continued to increase, eventually growing to 3.56 ± 0.1 mg/kg at the end of storage in the NTP-BFM.Similarly, the content of 5-HMF in TP-BFM also increased, but the trend of was more moderate, with no significant difference during 21-d storage.The contents of 5-HMF in NTP-BFM were 1.01, 1.33, 1.69, and 1.86 times higher than in TP-BFM on d 0, 7, 14, and 21 of the storage period, respectively.Meanwhile, the inhibition rate of TP to 5-HMF also gradually increased, reaching the highest inhibition rate of 46.6% on d 21, which indicated that the capture ability of TP to 5-HMF was increased with time.Consistently, the presence of EC and HMF dimer adducts in fried potato chips after 28 d of storage has also been confirmed (Qi et al., 2018).

Content and Evolution of α-Dicarbonyl Compounds
α-Dicarbonyl compounds are yellow compounds with low molecular weight, formed by sugar fragmentation during caramelization and non-enzymatic browning (Liu et al., 2017).Moreover, they are also formed during oxidative degradation as intermediates of the Maillard reaction.The pH value significantly affects the nucleophilic reactivity of amino acids, thus changing the production of α-DC (Yu et al., 2017).When pH is 7 or above, the 1-deoxygenation pathway is dominated by 2,3 enolization to produce GO, MGO, and other fission products; that is, higher-pH conditions are more favorable for the production of α-DC (Martins et al., 2000).The pH of BFM was approximately 4.6, so BFM was not conducive to α-DC production.As shown in Figure 2b and Figure 2c, TP-BFM had lower levels of GO and MGO than did the NTP-BFM at d 0, which did not change significantly from 7 to 14 d, but a significant increase in GO and MGO occurred at the end of storage (d 21; P < 0.05).The final contents of GO and MGO in TP-BFM were 1.76 ± 0.08 mg/kg and 10.5 ± 0.59 mg/kg, respectively, which were significantly lower than those of NTP-BFM (1.95 ± 0.04 mg/kg and 13.3 ± 0.89 mg/kg; P < 0.05).This might be related to the post-acidification of BFM in early storage, and the stabilization of the pH in late storage.The inhibition rates of GO by TP on d 0, 7, 14, and 21 of the storage period were 13.8%, 26.6%, 22.2%, and 11.7%, and the inhibition rates of MGO were 46.1%, 28.5%, 24.5%, and 20.7%, respectively.This indicates that the inhibition effect of TP on GO was more effective in the mid-storage period, whereas the inhibition efficiency on MGO gradually decreased throughout the storage period.However, the inhibition rate of MGO was always higher than that of GO.This result showed that the capture efficiency of MGO by TP was higher than that of GO, which was consistent with the previous studies (Luo et al., 2022).At pH 7.4 and 37°C, epigallocatechin gallate could capture more than 90% of MGO within 5 min and 70% of GO within 1 h (Sang et al., 2007).

Content and Evolution of Advanced Glycosylation End Products
We selected CEL and CML to reflect the content of AGE.As shown in Figure 2d and Figure 2e, the contents of CML and CEL in both samples increased during the storage period, with a higher increase rate of NTP-BFM than TP-BFM.This suggested that TP did not interrupt the process of the Maillard reaction, but indirectly reduced the formation of end products through the capture of intermediate products.Similar to previous studies, resveratrol mainly inhibited the formation of early MRP, and achieved inhibition by inhibiting autocatalytic lipid oxidation that produces α-DC, without directly acting on α-DC (Yu et al., 2020).The contents of CML and CEL in NTP-BFM were always higher than in TP-BFM during the storage.Levels of CEL increased from 26.7 ± 1.36 µg/kg (NTP-BFM) and 18.47 ± 1.34 µg/kg (TP-BFM) on d 1 to 219.5 ± 29.6 µg/kg and 165.2 ± 12.7 µg/kg by d 21, respectively.The inhibition rates of both CML and CEL by TP reached the maximum at d 14 with 50.2% and 38.8%, respectively.However, the contents of CML in TP-BFM and NTP-BFM were not significantly different at d 21, which might also be related to the increased production of GO at the end of the storage period, since CML could be formed by the oxidation of GO or by reaction with lysine (Troise et al., 2015).

Color Changes in Brown Fermented Milk with Tea Polyphenols During Storage
In general, the color of sample browning was positively correlated with the accumulation of MRP.The values of the color parameters (a*, b*, L*, ∆E*, and BI) of BFM were measured and are presented in Table 2.The brightness of TP-BFM (73.04 ± 0.82) was lower than that of NTP-BFM (76.39 ± 1.02) on the first day, which might be influenced by the addition of tea polyphenols.During the storage process, the L* values kept decreasing while the a* and b* values of the samples gradually increased, indicating a gradual darkening of BFM.This was related to the accumulation of MRP, resulting in more red and yellow colors of the sample.The values of ∆E* and BI were calculated according to the equations and were found to increase gradually during storage of BFM, but the ∆E* and BI values were lower than those of NTP-BFM.After 21 d of storage, the ∆E* values of the 2 samples were 6.49 ± 0.02 (NTP-BFM) and 6.17 ± 0.02 (TP-BFM), respectively, and BI were 13.80 ± 0.05 (NTP-BFM) and 13.16 ± 0.01 (TP-BFM), respectively.Therefore, the addition of TP did not negatively affect the color of BFM and even slowed down the color change during storage, which facilitated the maintenance of the attractive color of yogurt.

CONCLUSIONS
In this study, the inhibition effect of TP on 5 MRP in BFM was investigated.When the addition amount of TP was 0.08%, we found a better inhibitory effect on the production of MRP.Because most flavor compounds were detected simultaneously in both BFM formulations, the overall flavor profile of BFM with 0.08% TP did not change significantly.During storage at 4°C for 21 d, the 5 MRP (5-HMF, GO, MGO, CEL, and CML) in BFM with 0.08% TP increased, but their contents were lower than those in the group without TP.Considering the high inhibition of TP on 5-HMF (from 1.61 to 1.91 mg/kg), GO (from 1.50 to 1.76 mg/kg), and MGO (from 4.15 to 10.51 mg/kg), we speculate that TP mainly inhibits the Maillard reaction by capturing the intermediate products.Moreover, the color and BI of BFM hardly changed after the addition of 0.08% TP.Based on these results, a promising strategy to inhibit the formation of MRP in BFM was established.Moreover, this study further improves the safety quality of brown fermented products and can increase their competitiveness on the market.
Li et al.: TEA POLYPHENOLS IN BROWN FERMENTED MILK

Table 1 .
Li et al.: TEA POLYPHENOLS IN BROWN FERMENTED MILK Summary of volatile compounds detected in brown fermented milk (BFM) 1

Table 2 .
Li et al.: TEA POLYPHENOLS IN BROWN FERMENTED MILK Color analysis of brown fermented milk during 21-d storage 1 Different letters in the same column indicate significant differences at P < 0.05.Results are expressed as mean ± SE. Brown fermented milk (BFM) with or without tea polyphenols (TP) are respectively designated TP-BFM and NTP-BFM.L* = lightness; a* = greenness/redness; b* = blueness/yellowness; ∆E* = total color difference; BI = browning index. 1