short communication Ital. J. Food Sci., vol. 27 - 2015 1 - Keywords: aminoreductone, Maillard Reaction, optimum condition, Box-Behnken design, response surfaces - OPTIMIZATION OF THE AMINOREDUCTONE FORMATION IN THE MAILLARD REACTION VU THU TRANg1*, NgUyEN THI HAO1, VU HONg SON1, HO PHU HA1, HIROyUkI UkEDA2 and TOMOkO SHIMAMURA2 1School of Biotechnology and Food Technology, Hanoi University of Science and Technology, N. 1 Dai Co Viet, Hanoi, Vietnam 2Faculty of Agriculture, Kochi University, B 200 Monobe, Nankoku, Kochi 783-8502, Japan *Corresponding author: Tel. +84 43 8680119, Fax +84 43 8682470, email: trang.vuthu@hust.edu.vn; thutrangvu1981@yahoo.com AbstrAct the optimization of process parameters for the production of aminoreductone (Ar), a bioac- tive product formed in the initial stage of Maillard reaction was investigated using response sur- face methodology (rsM) and box-behnken design technique. the optimum process conditions were determined by analyzing the response surface of three-dimensional surface plot and solv- ing the regression model equation with the Design Expert software. the optimum conditions in- clude: heating time of 15 min, temperature of 112.85°c, pH of 8.33 and buffer concentration of 0.53 which were used to obtain the maximum Ar yield (76.6 mM) in the model solution of lactose (0.3 M) and butylamine (0.3 M). 2 Ital. J. Food Sci., vol. 27 - 2015 INtrODUctION Maillard reaction products are responsible for the change of color, taste, flavor and the nu- tritional value of food products (rAMONAItYtE et al., 2009). therefore, the Maillard reaction is the most important influence on food quali- ty and acceptance (JAEGEr et al., 2010). How- ever, the evaluation of the extent of the Mail- lard reaction is difficult with many parallel and consecutive reactions (MOrALEs and JIMEN- EZ-PErEZ, 1998). several studies have report- ed detection methods for estimating the extent of Maillard reaction by the detection of an in- termediate product as hydroxymethylfurfural (sHIMAMUrA et al., 2004) or the final polymer- ized products such as melanoidins (bOEKEL, 1998). In the early stage of the Maillard reac- tion, Aminoreductone (Ar) is formed (bOEKEL, 1998). therefore, the detection of Ar would be more effective in the indication of the Maillard reaction and heat treatment of food than other methods (sHIMAMUrA et al., 2004). thus, the role and characteristics of Ar is, of great inter- est to food scientists. A number of functionalities of Ar such as an antioxidative activity, have been found (PIs- cHEtsrIEDEr et al., 1998), a protective abil- ity on photo-degradation of riboflavin in milk (trANG et al., 2008), and antimicrobial activities against pathogenic bacteria: Helicobacter pylori (trANG et al., 2009), Pseudomonas aeruginosa (PA), multi-drug resistant Pseudomonas aerugi- nosa (MDrP), Escherichia coli (Ec), methicillin- susceptible Staphylococcus aureus (MssA) and methicillin-resistant S. aureus (MrsA) (trANG et al., 2011, 2013). In the field of food technol- ogy, food scientists and producers always con- sider many factors that can contribute to a good and healthy product. As Ar presents the poten- tial properties in medical practices (trANG et al., 2009, 2011) and contributes to food quality (KAtsUNO et al., 2013), it can be used as a func- tional additive ingredient in food to improve the quality of food. As a product of the Maillard reaction, the formation of Ar depends on multiple parame- ters such as, the heating time, the heating tem- perature, the pH and the buffer concentration (bOEKEL, 1998). conventionally, the formation of Ar might be optimized using a single factor optimization to evaluate the optimum produc- ing condition, which is relatively simple and does not require statistical analysis. However, the single variable optimization strategy is not only tedious, but can also lead to misinterpre- tation of results, especially since the interaction between different factors are overlooked (MAN- NAN et al., 2007). the response surface meth- odology (rsM) as a combination of mathemati- cal and statistical techniques was employed to overcome this major problem in the optimization study (LI et al., 2008). In this method, statisti- cally designed experiments used a small set of carefully planned experiments, to build models, evaluate the effects of factors and find the op- timum conditions for desirable responses (LI et al., 2002). It can simultaneously study several variables with a small number of observations, less time consumed and cost effects (DEEPAK et al., 2008). For this reason, the purpose of this study is to optimize the technological conditions favoring the production of Ar in a model sys- tem using a statistical approach: response sur- face methodology. MAtErIALs AND MEtHODs Reagents X t t ( 2 , 3 - b i s [ 2 - m e t h o x y - 4 - n i t r o - 5-sulfophenyl]-2H-tetrezolium-5-carboxanil- ide) was purchased from sigma chemical co. (st Louis, MO, UsA). Lactose monohydrate was purchased from Nacalai tesque, Inc. (Kyoto, Japan). n-butylamine was obtained from Wako Pure chemical Industries (Osaka, Japan). All other reagents were of the highest commercial grade available. Milli-Q water was used in all procedures. Model solutions the solutions containing lactose and butyl- amine were used as a model system of by Ar production in the Maillard reaction (model so- lution). the solutions were prepared according to the previous report (sHIMAMUrA et al., 2004; trANG et al., 2011). Lactose monohydrate (0.3 M) and butylamine (0.3 M) were dissolved in phosphate buffer. One milliliter of the model so- table 1 - Factors in actual and coded levels for the box-behnken design. No Factors Symbols Coded and actual level -1 0 +1 1 Heating temperature (°C) A 90 110 130 2 Heating time (min) B 5 15 25 3 pH C 7 8 9 4 Buffer concentration (M) D 0.3 0.5 0.7 Ital. J. Food Sci., vol. 27 - 2015 3 lutions was heated under the indicated condi- tion. Immediately after heating, the heated so- lutions were cooled in ice and used for the de- termination of Ar formation. Determination of aminoreductone formation the formations of Ar in the heated model so- lutions were determined using a Xtt assay, per- formed in a 96-well microtiter plate according to the method described by shimamura et al. (2011). Each well contained 60 μL of 0.5 mM Xtt prepared with 0.2 M potassium phosphate buff- er (pH 7.0) saturated with menadione. A sample (40 μL) was added to the well and after mixing in a microplate shaker at a speed of 500 rpm for 15 s, the difference in the absorbance between 492 nm and 600 nm was measured using a micro- plate reader (MPr A4i, tosoh, tokyo, Japan) as the absorbance at 0 min. After 20 min at room temperature, the difference in absorbance was again measured and the increase in the absorb- ance was recorded as the ability of a sample to reduce Xtt (Xtt reducibility). the concentration of Ar was estimated by the following equation: y = 0.606 x + 0.046, where x and y represent the concentration of Ar (mM) and the reducibility of Xtt, respectively (trANG et al., 2008). Experiment design and procedure box-behnken design with three levels (low, medium, and high, coded as -1, 0, and +1) is more efficient and easier to arrange and to in- terpret when compared with the others, such as the Plackett-burman design, the central com- posite design and the Graeco-Latin square de- sign (FrANcIs et al., 2003). this statistical tech- nique was therefore used in this study. A total of 27 runs was used to optimize the producing parameters namely: pH, buffer con- centration, heating temperature and heating time (bOEKEL, 1998). Factors in actual and cod- ed levels considered in this study are listed in table 1. the experiments were designed accord- ing to the box-behnken design using 24 axial points and three central points as shown in ta- ble 2. Individual experiments were carried out in random order. the average of two replicated values of each run was taken as dependent var- iables or responses. Design-Expert 7.1 (stat-Easse, Inc., Minne- apolis, MN, UsA) was used for the experimental design, data analysis, quadratic model building, graph (three-dimensional response surface and contour) plotting and to optimize by desirabili- ty methodology. table 2 - Experimental design and results of the box-behnken design. Run Factor 1 Factor 2 Factor 3 Factor 4 AR concentration A (°C) B (min) C D (M) (mM) 1 90 5 8 0.5 0.032 2 130 5 8 0.5 34.855 3 90 25 8 0.5 26.244 4 130 25 8 0.5 15.270 5 110 15 7 0.3 45.056 6 110 15 9 0.3 64.033 7 110 15 7 0.7 49.940 8 110 15 9 0.7 74.594 9 90 15 8 0.3 8.802 10 130 15 8 0.3 35.402 11 90 15 8 0.7 19.148 12 130 15 8 0.7 36.247 13 110 5 7 0.5 10.947 14 110 25 7 0.5 33.1 15 110 5 9 0.5 25.3 16 110 25 9 0.5 43.9 17 90 15 7 0.5 10.811 18 130 15 7 0.5 34.91 19 90 15 9 0.5 24.954 20 130 15 9 0.5 46.294 21 110 5 8 0.3 20.89 22 110 25 8 0.3 44.396 23 110 5 8 0.7 35.508 24 110 25 8 0.7 44.478 25a 110 15 8 0.5 76.657 26a 110 15 8 0.5 76.657 27a 110 15 8 0.5 69.672 a Center points 4 Ital. J. Food Sci., vol. 27 - 2015 rEsULts AND DIscUssIONs Effects of individual factors on aminoreduc- tone formation. Variables and factor levels of the experimental design to optimize the variables of Ar formation, the key factor affecting the Ar formation as well as the range of experimental values must be deter- mined. the mechanism of Ar formation in the Maillard reaction in the model solution of lactose and butylamine has been investigated (trANG et al., 2011). based on the results of those exper- iments, the solution of lactose and butylamine at 1:1 in concentration ratio (0.3:0.3 (M)) was the best model system for the formation of Ar and was used in our previous studies (trANG et al., 2011). thus, we used it in the production of Ar in this study. the effects of individual fac- tors in the formation of Ar while keeping oth- er variables constant were shown in Fig. 1. In the model solution consisting lactose and but- ylamine, the extent of the Ar formation strong- ly relied on the heating temperature and heat- ing time (Fig. 1a and 1b). this phenomenon was also similar to our previous studies (sHIMAMU- rA et al., 2004; trANG et al., 2011). As soon as the maximum amount of Ar was obtained, Xtt value decreased. these might be referred to as the progress of the advanced reactions of Ar that commonly takes place in the complicated sequences of the Maillard reaction during heat- ing (trANG et al., 2011) or by the competition of isomerisation⁄degradation reaction to lactose at the high heating temperature (more than 100°c) leading to the reduction of lactose for the Mail- lard reaction (bOEKEL, 1998). the maximum val- ue of Ar obtained in Fig. 1a and 1b might also change depending on the change of heating time and heating temperature, respectively. thus, the heating time (with a range of 5-25 min) and heating temperature (with a range of 90-130°c) were chosen as the main factors and their lev- els for the experimental design on the response surface methodology. boekel (1998) suggested that changes in pH can have an effect on reaction rates (bOEKEL, 1998). At the same, heating time and tempera- ture, reactivity of sugar and amino group are also influenced by pH (MArtINs et al., 2001). similar results were also found in this study (Fig. 1c). In the Maillard reaction of lactose and butylamine Fig. 1 - Effects of individual factors in the formation of aminoreductone. the effect of heating temperature (Fig. 1a), heating time (Fig. 1b), pH (Fing. 1c), buffer concentration (Fig. 1d) on the Ar for- mation while other factors were controlled (heating time of 15 min; heating temperature of 110°c; buffer concentration of 0.2 M and pH of 7). Ital. J. Food Sci., vol. 27 - 2015 5 at 110°c, 15 min of heating time, buffer con- centration of 0.2M, the formation of Ar rapid- ly reached maximum values from pH 5 to 8 and did not change much at higher pH values. the pH value that was higher, was used. the more complicated equipment design and handling for Ar producing was required. thus, the range of pH value from 7 to 9 for Ar production was cho- sen for experimental design. Limited data exist on the effects these buff- ers have on the Maillard reaction and the for- mation of Ar. besides, buffer agents were add- ed in diverse foods to control the pH of the sys- tem (bELL, 1997). to find out the suitable con- ditions and establish the good model solution for the production of Ar, the effect of phosphate buffer concentration in the formation of Ar was also investigated. As shown in Fig. 1d, the rates of Ar formation increased with the increasing phosphate buffer concentration from 0.1 to 0.5 and slightly decreased with higher concentra- tions. similar observations of the increase in the Maillard reaction rate with the increasing buff- er concentration was also presented in the mod- el system of glycine and glucose (bELL, 1997). the results obtained in our study indicated that phosphate anion should be used as a catalytic compound for the production of Ar. the range of buffer concentration from 0.3 to 0.7 which con- tains an optimized buffer concentration for Ar production was chosen for experimental design. the parameters of optimization for Ar produc- tion were investigated using a box-behnken de- sign under rsM. Parameters such as, tempera- ture of 110oc, time of 15 min, pH of 8 and buff- er concentration of 0.5 were chosen as center points from the above pre-screening on the ef- fect of individual factors for the formation of Ar. Evaluation of aminoreductone formation the design matrix of the factors is shown in table 2, along with the experimental response values. Using the software Design Expert, the results of the experiment of Ar formation were used to calculate the coefficients of the quad- ratic polynomial equations, which were used to predict the formation of Ar. the statistical model was checked by F-test, and the analysis of variance (ANOVA) for the re- sponse surface quadratic model was summa- rized (table 3). As shown in table 3, the Mod- el’s F-value of 10.27 and the p value of 0.0001 (< α = 0.05) implied that the model was highly significant in which A, b, c, Ab, A2, b2, c2, D2 (p < 0.05) are significant model terms (table 3). because D2 had a significant effect, the corre- sponding main effect of D is included in the re- gression model. After excluding these insignifi- cant effects from the model and rerunning the software Design Expert, the model for Ar pro- duction might be expressed by: Y = + 74.33 + 9.42*A + 6.65b + 7.86*c + 3.44*D - 11.45*A*b- 33.26*A2 - 28.01*b2 - 11.90*c2 - 10.07*D2 where Y (yield) is the yield of Ar (mM); A, b, c, D is the coded values of the heating temperature, the heating time, the pH and the buffer concentration, respectively; R2 = 0.9155; adjusted-R2= 0.8708. the goodness of the model can be checked by the determination coefficient r2 and the ad- justed-r2. the determination coefficient r2 of table 3 - results of the regression analysis of the box-behnken design for Aminoreductone production. Factor Variable Regression coefficient F-value P-value (Probability) > F Model 10.27 0.0001 b o + 74.33 Linear A b 1 + 9.42 14.01 0.0028* B b 2 + 6.65 7.00 0.0214* C b 3 + 7.86 9.76 0.0088* D b 4 + 3.44 1.87 0.1960 Interaction AB b 12 - 11.45 6.90 0.0221* AC b 13 - 0.69 0.025 0.8769 AD b 14 - 2.38 0.30 0.5957 BC b 23 - 0.89 0.042 0.8419 BD b 24 - 3.63 0.70 0.4206 CD b 34 + 1.42 0.11 0.7503 Quadric A2 b 11 - 33.26 77.65 < 0.0001* B2 b 22 - 28.01 55.09 < 0.0001* C2 b 33 - 11.90 9.94 0.0083* D2 b 44 - 10.07 7.11 0.0205* *, insignificant model terms. 6 Ital. J. Food Sci., vol. 27 - 2015 0.9155 indicated that the model could explain 91.55% of the variance (WANG and LU, 2005). the value of adjusted-r2 closed to r2 and 1, showed the good correlation between the exper- imental and predicted values. thus, this mod- el can be used to predict the formation of Ar in the Maillard reaction. Analysis of response surfaces three-dimensional response surfaces were plotted on the basis of the model equation by the Design Expert program to investigate the interac- tion among the variables and to determine the op- timum condition of each factor for maximum Ar Fig. 2 -response surface plot for Aminoreductone production. the interaction between (a) heating time and heating temperature, (b) pH and heating temperature, (c) buffer concentration and heating temperature, (d) pH and heating time, (e) buffer concentration and heating time, (f) buffer concentration and pH. Ital. J. Food Sci., vol. 27 - 2015 7 Fig. 3 - Perturbation graph showing the effect of each independent factors on Aminoreductone production while keeping oth- er factors at their respective midpoint levels. (A) heating temperature, (b) heating time, (c) pH, (D) buffer concentration. production in the model solution of lactose and butylamine (Fig. 1). by keeping other variables at their center point values, three dimensional plots of two factors versus the Ar formation were drawn (Fig. 2). Perturbation graph showed the effect of each independent factors on Ar production while keeping other factors at their respective midpoint levels (Fig. 3). From the response surface (Fig. 2) and perturbation plot (Fig. 3), it is obvious that heating temperature and time had a significant effect on Ar production compared with other vari- ables. Although, pH was reported to influence the Maillard reaction (MArtINs et al., 2001), the re- sults of this study indicated an unimportant ef- fect of pH on the formation of Ar. the perturba- tion graph clearly showed that the two variables (buffer concentration and pH) did play any sig- nificant role in the Ar production. Optimization of conditions for aminoreductone formation based on the analysis of the response surface of the regression equation, the optimum process parameters were found to be 112.85oc for heat- ing temperature 15 min for heating time with pH 8.33 and buffer concentration 0.53 M resulting in the predicted maximum Ar formation which was 76.6 mM. Validation of the models the trail experiments were conducted under optimized process conditions with temperature of 112.8oc, heating time of 15 min, pH of 8.3 and buffer concentration of 0.5 M. the results of Ar formation were founded to be 75.76 ± 0.02 mM, which was very close to the predicted Ar forma- tion obtained from the regression equation (76.6 mM). thus, the model could be used to predict the Ar content formed in the food during heat- ing; and to find the suitable heating, condition (temperature and time) that favored the forma- tion of aminoreductone in the specific food sys- tem which the pH and food components were clarified. the chemistry underlying the Maillard reac- tion is very complex. It encompasses not only one reaction pathway, but a whole network of various reactions with thousand products (MArtINs et al., 2001). by applying the rsM, the optimum process parameters for Ar pro- duction were found. these were both helpful for studying further the application of Ar in the health and medical fields, as well as pro- viding provide useful information for identi- fying the technological conditions that favors the formation of Ar as a functional ingredi- ent in food. 8 Ital. J. Food Sci., vol. 27 - 2015 cONcLUsIONs In this study, rsM, box-behnken design was used to model and establish a regression equa- tion between the response (Ar formation) and four statistically significant factors: the heating time, the heating temperature, the pH, the buff- er concentration. In four variables, the two fac- tors of heating time and heating temperature showed the effect on the Ar formation. Finally, the optimal solutions were sought on the basis of the influence of the four parameters in the formation of Ar in the Maillard reaction. Pre- dicted values obtained using the model equa- tions were in agreement with the observed val- ues. Heating time of 15 min, heating tempera- ture of 112.85oc, pH of 8.33 and buffer concen- tration of 0.53 has been determined as optimum levels of the process parameters to achieve the maximum amount of Ar formed in the Maillard reaction of lactose and butylamine. these opti- mum conditions were used to evaluate the trail experiment and the maximum yield of Ar for- mation was recorded as 75.76 mM. the results indicated that optimization using rsM can be useful to control and predict the production of Ar in the Maillard reaction. the results obtained from this study would be applied in Ar produc- tion for therapeutic application as an efficient source of antimicrobial agents, an antioxidant compound in functional food. AcKNOWLDGMENts this research was funded by the Vietnam National Foun- dation for science and technology Development (NAFOst- ED, 106.99-2011.22.). rEFErENcEs bell L.N. 1997. Maillard reaction as influenced by buffer type and concentration. Food chem. 59: 143. boekel, M.A.J.s.V. 1998. Effect of heating on Maillard reac- tions in milk. Food chem. 62: 403. Deepak V., Kalishwaralal K., ramkumarpandian s., babu V.s., senthilkumar, s.K. and sangiliyandi, G. 2008. Op- timization of media composition for Nattokinase produc- tion by Bacillus subtilis using response surface method- ology. biores. technol., 99: 8170. Design-Expert. 2008. Design-Expert 7.1 User’s Guide, re- lease 7.1, Minneapolis, MN. Francis F., sabu, A. and Nampoothiri, K.M. 2003. Use of re- sponse surface methodology for optimizing process pa- rameters for the production of a-amylase by Aspergillus oryzae. biochem. Engineer. J. 15: 107. Katsuno s., shimamura t., Kashiwagi t., Izawa N. and Ukeda H. 2013. Effects of dissolved oxygen on the Mail- lard reaction during heat treatment of milk. Int.Dairy J. 33: 34. Li c., bai J. and cai Z. 2002. Optimization of a cultur- al medium for bacteriocin production by Lactococcus lactis using response surface methodology. J. biotech- nol. 93: 27. Li X., Xu t., Ma X., Guo K., Kai L., Zhao Y., Jia X. and Mad Y. 2008. Optimization of culture conditions for produc- tion of cis-epoxysuccinic acid hydrolase using response surface methodology. biores. technol. 99 (13): 5391. Mannan s., Fakhru’l-razi’ A. and Alam M. 2007. Optimiza- tion of process parameters for the bioconversion of acti- vated sludge by Penicillium corylophilum using response surface methodology. J. Environmental sci. 19: 23. Martins s.I.s., Jongen W.M.F. and boekel M.A.J.s.V. 2001. A review of Maillard reaction in food and implications on kinetic modeling. trends Food sci. technol. 11: 364. Morales F. J. and Jimenez-Perez s. (1998) study of hydrox- ymethylfurfural formation from acid degradation of the Amadori product in milk-resembling systems. J. Agric. Food chem. 46: 3885-3890. Pischetsrieder M., schoetter c. and severin t. (1998). For- mation of an aminoreductone during the Maillard reac- tion of lactose with Nα-acetyllysine or proteins. J. Agric. Food chem. 46: 928-931. ramonaityte D. t., Kersiene M., Adams A. and tehrani K. A. (2009) the interaction of metal ions with Maillard re- action products in a lactose-glycine model system. Food res. Int. 42: 331-336. Jaeger H., Janositz A. and Knorr D. (2010) the Maillard re- action and its control during food processing. the po- tential of emerging technologies. Pathologie biologie, 58: 207-213. shimamura t., Ukeda H. and sawamura M. 2004. rela- tionship between the Xtt reducibility and aminoreduc- tone formed during the Maillard reaction of lactose: the detection of aminoreductone by HPLc. Food sci. tech- nol. res. 10: 6. shimamura t., Kurogi Y., Katsuno s., Kashiwagi t. and Ukeda H. 2011. Demonstration of the presence of ami- noreductone formed during the Maillard reaction in milk. Food chem. 129: 1088. trang V.t., Kurogi Y., Katsuno s., shimamura t. and Uke- da H. 2008. Protective effect of aminoreductone on pho- to-degradation of riboflavin. Int. Dairy J. 18: 344. trang, V. t., takeuchi, H., Kudo, H., Aoki, A., Katsuno, s., shimamura, t., sugiura, t. and Ukeda, H. (2009). Anti- microbial activity of aminoreductone against Helicobacter pylori. J. Agric. Food chem., 57, 11343-11348. trang V.t., shimamura t., Kashiwagi t., Ukeda H. and Kat- suno s. (2011). Elucidation of mechanism of aminoreduc- tone formation in the Maillard reaction of lactose. Int. J. Dairy technol. 64 (2): 188. trang V.t., son V. H., thanh L. X., sarter s., shimamura t., Ukeda H. and takeuchi H. (2013). Functional prop- erties of Maillard reaction products in food: Antimicro- bial activity of Amnireductone against pathogenic bacte- ria. Food sci. technol. res. 19 (5): 833. Wang Y.X. and Lu Z.X. 2005. Optimization of processing pa- rameters for the mycelial growth and extracellular poly- saccharide production by Boletus spp. Accc 50328. Pro- cess biochem. 40: 104 Paper Received July 23, 2014 Accepted September 19, 2014