2018 Volume 24 Issue 2 Pages 283-287
To maximize the potential of the Maillard reaction for the development of novel time-temperature integrators/indicators (TTIs), we investigated various color change patterns using different combinations of reactant concentrations and temperatures for the Maillard reaction with xylose, glycine and disodium hydrogen phosphate. We analyzed digital photo images for quantitative evaluation of the color changes in the RGB color space. The results revealed that the sigmoidal RGB value decreased under all conditions, and the reaction was accelerated by increases in temperature and reactant concentration. Thus, the reaction completion time, which was 18–61, 7–21, 4–14, 2–10, 1–4, and 0.5–2 days at 0, 5, 10, 15, 20, and 25°C, respectively, could be tuned by controlling the above variables. In addition, we developed a mathematical model that predicted the color change over time with high accuracy (R2 > 0.99) for a wide range of temperatures.
The rise in quality of life has increased consumer demands for safe, fresh and high-quality foods. The development of sophisticated distribution systems for ensuring food safety and freshness has therefore been required. Temperature is the most important environmental factor influencing food quality (Vaikousi et al. 2008). Consequently, a method for recording and estimating temperature history during food distribution and storage would enable prediction of the changes in food quality and microbial growth and/or survival (Taoukis and Labuza 2003). Among the devices addressing these requirements, time-temperature integrators/indicators (TTIs) have attracted attention.
TTIs based on different mechanisms of color development have been proposed, featuring active ingredients such as anthocyanin (Maciel et al. 2012), bromophenol blue (Kuswandi et al. 2013) and tests have been conducted to assess the freshness of guava (Psidium guajava L., CO2 (Nopwinyuwong et al. 2010; Rukchon et al. 2014), food dyes (Murakami and Isshiki 2011; Ohta et al. 2008; Suto et al. 2013), isopropyl palmitate (Kim et al. 2016), lactic acid (Wanihsuksombat et al. 2010) and microbes (Kim et al. 2012; Vaikousi et al. 2008). In addition, Yamamoto and Isshiki (2012) developed a Maillard reaction-based TTI for alerting the growth of Listeria monocytogenes during low temperature distribution. In the Maillard reaction, reducing sugars react with amino acids to form melanoidins (Ames, 1998). Although the reaction mostly results in browning, it can also produce blue, green and yellow colors at certain reaction temperatures. Yamamoto and Isshiki (2012) applied this feature in their model Maillard reaction-based TTI.
The TTI developed by Yamamoto and Isshiki (2012) successfully illustrated the usefulness of chilled temperature distribution and has validated the accuracy of color change for alerting the growth of Listeria monocytogenes (Rokugawa and Fujikawa 2015) Although the TTI was developed for low temperature management, numerous color variations can be realized by the Maillard reaction by adjusting reactant combinations and concentrations under various temperatures (Bell 1997; Benzing-Purdie et al. 1985). This flexibility would benefit various issues for temperature management such as food distribution, long-term storage, judgment of fruit ripening, and harvesting times.
Under these circumstances, we aimed to maximize the applicability of the Maillard reaction-based TTI. We used various combinations of xylose and glycine (similar to that used previously by Yamamoto and Isshiki (2012)) at different temperatures to realize the numerous color change patterns of Maillard reaction. Furthermore, we quantitatively evaluated the color change by image analysis and developed a predictive model for determining the time required for the desired color changes as a function of temperature and reactant concentrations.
Reactants D-xylose and glycine were used as reactants for the Maillard reaction due to their excellent reactivity and pronounced color changes (Ellis 1959; Laroque et al. 2008). Disodium hydrogen phosphate (Na2HPO4) was used as a reaction accelerator. The following reactant concentrations were used: D-xylose: 1.0, 1.5, 2.0, and 2.5 M; glycine: 1.0, 1.5, and 2.0 M; and Na2HPO4: 0.2, 0.3, and 0.4 M. Reaction solutions were prepared by mixing the D-xylose and glycine solutions in a 1:1 volume ratio. During this process, D-xylose and glycine solutions were diluted by aqueous Na2HPO4 solutions of different concentrations. The mixed reaction solution was dispensed (0.2 mL) into a well of a 96-well microplate and the dispensed plates were placed in an incubator (CN-25C, MEE, Tokyo & SU240, ESPEC, Osaka, Japan) set at 0, 5, 10, 15, 20, and 25°C until the solution turned dark brown or black.
Analysis of color change The color variation in solution was evaluated in the RGB color space (R (red), G (green), and B (blue)) by imaging with a digital camera (iPhone 5S, Apple Corp., CA) at 12-h intervals in most reaction conditions. In contrast, fast reactions at 20 and 25°C were imaged at 5-min intervals automatically using time-lapse photography. Digital images were acquired at a resolution of 1920 × 1080 pixels, f/2.2 aperture, and 4.15-mm focal length, with a box blocking ambient light for minimizing the effect by lighting. For better performance, optimized exposure conditions were employed, i.e., the camera was used at an automatically set white balance and sensitivity (ISO 32). Captured images were analyzed using the R (ver. 3.3.0) statistical environment running in the R studio (ver. 0.99.491).
To compare reaction completion times under different conditions, a threshold value was required as a completion index. Herein, 0.20, 0.15, and 0.15 were used as threshold values for red, green, and blue, respectively, with colors exhibiting lower RGB values appearing as dark brown or black. Additionally, these values could also be used for timing purposes in the fabricated TTIs.
Color variation expressed by RGB value changes was described by a sigmoid function (Eq. 1):
 |
where Δcolor value, rmax, time and Tinf denote RGB parameter changes, inflection point slope, reaction time, and inflection point time, respectively. Parameters for each condition were estimated using the non-linear least square method (R, nls package).
The temperature dependence of all reaction conditions was expressed by the Arrhenius function (Eq. 2):
 |
where k is rmax (inflection point slope) of sigmoid function, Ea the activation energy, A is the pre-exponential factor, and R and T are ideal gas constant and absolute temperature (K), respectively. The slope (Ea/R) and pre-exponential factor were estimated using linear regression (Kim et al. 2012).
Predictive modeling of color change kinetics Each parameter obtained by fitting the experimental data with Eq. 1 was described as a function of reactant concentration and reaction temperature with a multiple regression analysis (Eq. 3):
 |
where ai are coefficients, T is the reaction temperature (absolute temperature), and Xy, Gl, and DH are concentrations of xylose, glycine, and Na2HPO4, respectively. All parameters (µmax, Tinf, or Δcolor value), reaction temperature (T), and reactant concentration (Xy, Gl, DH) were transformed to a natural logarithmic value with the highest value in coefficient of determination of the regression analysis results. Additionally, data pertaining to 0, 5, 15 and 25°C were used for predictive modeling. Data obtained at 10 and 20°C were used to validate the predictive model.
Characterization of color change under different conditions The Maillard reaction between D-xylose, glycine, and Na2HPO4 showed a pronounced color change (Fig. 1) from colorless to black via transitions of light blue, green, and brown, and a broadly adjustable reaction completion time, which marked it suitable for TTI development. The changes in color developed during the reaction enable the visual recognition of the temperature history. Therefore, each step in the color changes of TTI expresses the current state for temperature accumulation and enable user to the rough prediction of quality of food during storage or distribution. Nonetheless, a measuring system is required for quantitatively evaluating the color change for different food types. Furthermore, TTI color changes have to be correlated to food quality and/or safety changes to provide additional information on food distribution under the selected temperature conditions. Thus, quantitative evaluation of the color change is indispensable for development of flexible TTIs.
Color changes observed for the Maillard reaction between 2.0 M D-xylose, 2.0 M glycine, and 0.2 M Na2HPO4.
Quantitative evaluation of color changes revealed that RGB values decreased sigmoidally under all conditions. Figure 2 demonstrates representative observed/fitted color variations at 10°C for xylose (1.0 M), glycine (1.5 M), and Na2HPO4 (0.3 M). The observed color variations were successfully described using Eq. 1, resulting in root mean square errors (RMSEs) of < 0.02 for all investigated conditions.
Representative changes in R (red), G (green), and B (blue) values at 10°C for 1.0 M D-xylose, 1.5 M glycine, and 0.3 M Na2HPO4, with solid, dashed, and dotted lines representing the respective fitting curves.
A variety of numerical color expression systems are available, e.g., sRGB, CIELAB, and CIEXYZ. Although the CIELAB color system is frequently used in food science and technology, it requires a dedicated device for color value measurement, which limits its application range. In contrast, the RGB color space is commonly utilized in digital cameras, color TVs, and smartphones, being most suited for easy recognition and analysis of Maillard reaction-induced TTI color changes with compact digital devices. Furthermore, a corresponding smartphone application can be developed if the captured TTI images (color change) can be associated with target food quality and/or safety. Therefore, the above color changes were recorded using a digital camera and analyzed in the RGB color space.
Effect of reactant concentration and temperature on color change The observed color change kinetics was highly influenced by reaction temperature (Fig. 3). The reaction rate increased with substrate concentration for all reactants. However, the reaction rate did not obviously change above 2, 1.5, and 0.3 M for glycine, xylose, and Na2HPO4, respectively. In contrast, the reaction rate increased markedly with temperature, with observable color changes detected even at temperatures as low as 5 and 0°C. This flexibility allowed the reaction completion time (and thus the time required for the color change) to be adjusted according to the storage condition for the target food (18–61, 7–21, 4–14, 2–10, 1–4, and 0.5–2 days at 0, 5, 10, 15, 20 and 25°C, respectively) by choosing appropriate combinations of reactants and their concentrations.
A representative B (blue) value change as a function of reaction time and temperature.
The reaction rates at all conditions showed temperature dependency according to the Arrhenius relationship. A relationship between reaction rate (rmax) and temperature (1/T) expressed by the Arrhenius function showed high association (R2 > 0.96). Figure 4 shows a representative Arrhenius plot for concentration conditions of xylose (2.0 M), glycine (2.0 M), and Na2HPO4 (0.3 M). The reaction rate increased with temperature and reactant concentration. In addition, the range of Ea at each color channel (red, green, and blue) was 80–104 kJ/M, 94–112 kJ/M, and 90–106 kJ/M, respectively.
A representative Arrhenius plot of the Maillard reaction induced by 2.0 M D-xylose, 2.0 M glycine, and 0.3 M Na2HPO4.
These Ea values are similar to the reported Ea values for changes in food quality. The typical Ea values for lipid oxidation, nutrient loss, and microbial growth were 42–105, 84–125, and 84–251 kJ/M, respectively (Labuza 1982). The Eα values for growth of Brochothrix thermosphacta, Enterobacteriaceae, and lactic acid bacteria in modified atmosphere packed beef were 86, 93, and 106 kJ/M, respectively (Vaikousi et al. 2009). The typical Ea value for respiration rates in fruits and vegetables was 29–79 kJ/M (Poças et al. 2008). Thus, these similar Ea values between the measured color change and the reported food quality losses show that the color change in Maillard reaction-based TTI can be used to show the temperature history during food quality change such as lipid oxidation, nutrient loss, and microbial growth.
Development of a model for predicting color variation The fitted parameters (rmax, Tinf, and Δcolor change) of Eq. 1 were successfully described as functions of reactant concentrations and temperatures by Eq. 3 (R2 > 0.96). A combination of Eq. 1 and 3 allowed comprehensive and accurate prediction of Maillard reaction-induced color changes as shown in Fig. 6 (R2 = 0.99). The average differences between observed and predicted completion times were 42, 16, 9, 9, 5, and 3 h at 0, 5, 10, 15, 20, and 25°C, respectively.
The color variation rate of the Maillard reaction described herein allows the reaction time to be adjusted from 12 h to 2 months by controlling reaction temperature and reactant concentrations. Thus, Maillard reaction-based TTIs can not only potentially indicate the shelf life of various food products, including those stored chilled or at ambient temperature, but also establish the optimal ripening times of fruits and/or vegetables by the color changes of the TTI. Furthermore, the harvest time of fruits and/or vegetables during cultivation periods can also be optimized and predicted based on the accumulation temperature. Both the primary model (Eq. (1)) and the secondary model for the reaction rate describing Arrhenius equation (Eq. (2)) as shown in Fig. 4 will enable us to calculate and predict the color change even under fluctuating temperature conditions. Needless to say, although there are still some issues to realize the TTIs mentioned above, for example, synchronizing the color change with target quality change, the present study has demonstrated the great potential of the Maillard reaction-based TTI. We plan to develop some TTIs for practical use in the near future.
Acknowledgements This study was partially supported by a grant from the Tojuro Iijima Foundation for Food Science and Technology in 2016.
Accuracy of the developed predictive model for estimating the Maillard reaction completion time at 0, 5, 10, 15, 20, and 25°C. Each point represents a single reaction condition.
