Evaluation of eight kinds of flavor enhancer of umami taste by an electronic tongue

Abstract The umami intensity of single flavor enhancer was detected by an electronic tongue and human sensory. The linear fit was performed to unravel the concentration–response values correlations of eight flavor enhancers. The electronic tongue response data were then analyzed using principal component analysis (PCA) and discriminant factor analysis (DFA) method. It was found that the umami response value of the electronic tongue and the flavor enhancer concentration showed a semi‐logarithmic function. Moreover, the PCA and DFA could successfully distinguish the variety and concentration of flavor enhancer. The umami intensities were also assessed by human sensory and showed consistency with those of electronic tongue tests. This implies the electronic tongue has a great potential as an alternative for biological tongue on sensing intensity of flavor enhancer.


| INTRODUC TI ON
Taste of foodstuff is a decisive factor for purchasing and consumers' acceptance of the food. Umami is among the most important tastes such as sourness, sweetness, bitterness, and saltiness (Stone & Herbert, 2012;Yamaguchi & Ninomiya, 2000). Umami substances are found naturally in various foods, including sea food (e.g., fish, seaweed, clam, and oyster), cheese, and vegetables (e.g., edible fungi, soybean, and carrot) (Shah et al., 2010). They could also be used as additives, with a well-known commercial name-flavor enhancer. The traditional flavor enhancers can be categorized into three main groups: amino acids and their sodium salts, nucleotides and their sodium salts, organic acids and their sodium salts. Of which, monosodium glutamate has been widely used for many years (Flavor Enhancers & Potentiators, 2006). It enables to enhance persistence, richness, thickness, and mouthfulness of foodstuff (Mccabe & Rolls, 2010;Yamaguchi, 1979) without introducing its own taste.
The quality of foods has been evaluated using many different analytical tools to identify their physical, chemical, and sensory characteristics. Sensory evaluation by panelists is among the most commonly used methods, which can reflect the overall acceptance of food products (Stone & Herbert, 2012). When eating and tasting, the stimulation of a substance to human tongue varies with its concentrations. A taste curve can be established to unreal the correlation between the concentration of the substance and the intensity of the taste (Tian et al., 2015). However, sensory evaluation was affected by various external factors and physical and psychological conditions of the panelist, which requires large number of participants in order to acquire meaningful data (Yang et al., 2013). Therefore, an objective and rapid taste evaluation method is of vital importance for industries who intend to quickly narrow down the range of flavor enhancer content in food products.
In recent years, the electronic tongue has been considered as an valuable tool to evaluate the food stuffs in industries to discriminate and quantify the compounds of the basic tastes (Yang et al., 2013).
The electronic tongue tested the samples by simulating the human tongue concept, and meanwhile, it has higher sensitivity, better repeatability, and shorter determination time compared to the human tongue (Vlasov et al., 2002). Because of these advantages, the electronic tongue was applied in mineral waters (Labrador et al., 2009), tea (Chen et al., 2008), honey (Major et al., 2011), wines (Giorgio et al., 2007), beverages (Peres et al., 2009), and pharmaceuticals for the quality analysis and taste analysis. The electronic tongue enables to distinguish the intensity of bitterness and detect the tastemasking effect (Zheng & Keeney, 2006). For evaluating umami taste, it was mainly used to rank the umami intensity of monosodium glutamate, disodium inosinate, and guanylate (Yang et al., 2013) and in quantification of umami in tomato (Katrien et al., 2008).
In this study, the relationship between the concentration and intensity of a single umami taste on the electronic tongue was examined. The potential of using the electronic tongue to compare and evaluate the different flavor enhancers was also investigated. To prove the accuracy and reliability of the electronic tongue, human sensory evaluation was conducted concomitantly. From the current study, we found the electronic tongue has great potential as a replacement for some human sensory evaluation without comprising the sensory results.

| Reagents and materials
The eight different commercial umami substances (food grade) including monosodium glutamate, sodium aspartate, disodium inosinate, disodium guanylate, disodiumcytidylate, disodium adenylate, disodium uridylate, and disodium succinate (food grade) were purchased from Tianfeng Food Technology Co., Ltd. in a form of white

| Preparation of umami substance samples
Eight flavor enhancers were weighted into water to a certain concentration. For each flavor enhancer, a serious of concentrations were prepared (Table 1).

| Electronic tongue measurements
All samples were measured using the ASTREE electronic tongue To achieve the best performance of the electronic tongue, the sensors were conditioned by a conditioning, calibration, and diagnostic process before analyzing each sample. 0.01 mol/L of sodium chloride (NaCl), sodium L-glutamate (MSG), and hydrochloric acid (HCl) were used for the conditioning, diagnosis, and calibration processes (Yang et al., 2013). The E-tongue detecting conditions were set the same as the sensory evaluation method. Distilled water was used for cleaning during the testing. The conditioning test was used for regenerating the activity of sensor coating. The calibration was applied to standardize the sensor values in each analysis and ensure the results of all the samples were consistently and comparably (Campos et al., 2012;Ciosek & Wróblewski, 2007). A diagnostic process was performed to measure the sensitivity and discriminating capability of the sensor.
All samples were analyzed at ambient temperature with a 10 s measurement time (20 ml volume) and 120 s rinse time for the reference electrode and seven sensors. A washable cycle was performed to clean the sensors before the next analysis (Tian et al., 2013). Each sample was measured six times by the sensors, and three stable equilibrium data points were recorded. All the experiments were

| Sensory evaluation
The flavor enhancers of various concentrations were prepared as described previously. Notably, only single flavor enhancer, rather than their mixtures were tested to reduce the difficulty of human sensory evaluation. In this way, the panelists can easily distinguish the concentrations of single flavor enhancer and score their intensity. The scale of the intensity was from 1 to 10, where 1 represents no umami intensity and 10 represents the highest umami intensity.
The values given by ten panelists for the umami intensity were used for multivariate statistical analysis.

| Statistics and data analysis
The data obtained from the electronic tongue were analyzed by PCA and DFA using the α-Astree software. The PCA produces a score plot to visualize the differences among experiments and identify the main variables called components (Berrueta et al., 2007). The principal component (C1) and the second principal component (C2) were chosen in the present work. DFA was used to estimate the possibility of separating different groups (Winquist, 2008). The data used for the curve were calculated using a Macro Arithmetic Processor (commercial information). The response values were transformed into the value from 0 to 12.
The sensory data were calculated and statistically tested using the statistical analysis system (SAS 8.2) software (SAS Institute Inc.).

| Response pattern of electronic tongue to flavor enhancer of various concentrations
The determination results for single taste of umami using the electronic tongue are shown in Figure 1. It can be seen that the umami  (Tian et al., 2015). The coefficients of semi-logarithmic functions were different for different flavor substances even though they had similar curve shape. Disodium adenylate had the highest coefficient, followed by disodium succinate and disodium uridylate.
Monosodium glutamate had the lowest coefficient. These coefficients reflected the sensitivity of eight flavor enhancers toward the metal oxide membrane of sensors of electronic tongue. Neither molecular weight nor their chemical/physical properties (Table 2) showed direct correlation with the sensitivity. Most likely the interaction between the flavor substance and the metal oxide membranes is the decisive factor, which requires further study. In a previous study (Keast & Breslin, 2003), the tastes of sour, sweet, bitter, and salty were investigated at low, medium, and high concentrations. In

| Electronic tongue analysis of the flavor enhancer of same concentration gradients
The discriminative ability of electronic tongue system in distinguishing the umami taste of eight different flavor enhancers was examined. The PCA was performed on the raw data from the electronic tongue, and the principal component score vectors were extracted F I G U R E 2 Principal component analysis (PCA) score plot of the eight flavor enhancer solutions. The samples are the same as those in Table 2. AMP1, 0.5 g/L disodium adenylate; AMP2, 1.0 g/L disodium adenylate; A1, 0.5 g/L sodium aspartate; A2, 1.0 g/L sodium aspartate; CMP1, 0.5 g/L disodium cytidylate; CMP2, 1.0 g/L disodium cytidylate; GMP1, 0.5 g/L disodium guanylate; GMP2, 1.0 g/L disodium guanylate; S1, 0.5 g/L disodium succinate; S2, 1.0 g/L disodium succinate; IMP1, 0.5 g/L disodium inosinate; IMP2, 1.0 g/L disodium inosinate; MSG1, 0.5 g/L monosodium glutamate; MSG2, 1.0 g/L monosodium glutamate; MSG3, 2.0 g/L monosodium glutamate; UMP1, 0.5 g/L disodium uridylate; UMP2, 1.0 g/L disodium uridylate F I G U R E 3 Discriminant factor analysis (DFA) score plot of the eight flavor enhancer solutions. The samples are the same as those in Table 2. A1, 0.5 g/L sodium aspartate; A2, 1.0 g/L sodium aspartate; AMP1, 0.5 g/L disodium adenylate; AMP2, 1.0 g/L disodium adenylate; CMP1, 0.5 g/L disodium cytidylate; CMP2,1.0 g/L disodiumcytidylate; GMP1, 0.5 g/L disodium guanylate; GMP2, 1.0 g/L disodium guanylate; IMP1, 0.5 g/L disodium inosinate; IMP2,1.0 g/L disodium inosinate; MSG1, 0.5 g/L monosodium glutamate; MSG2, 1.0 g/L monosodium glutamate; MSG3, 2.0 g/L monosodium glutamate; S1, 0.5 g/L disodium succinate; S2, 1.0 g/L disodium succinate; UMP1, 0.5 g/Ldisodium uridylate; UMP2, 1.0 g/Ldisodium uridylate as the input of the pattern recognition. Therefore, it can extract the useful information by elimination of overlapped data. The responses of the sensor array were used to create a matrix with 51 rows (17 samples × 3 batches) and seven columns (7 sensor outputs). The PCA was conducted on the matrix, and the results were shown in The DFA was also performed on the taste sensor data from the electronic tongue ( Figure 3). The cumulative contribution percentage of two discriminant factors was 95.41%. It was found that the distribution maps of the samples from the same flavor substance were much smaller than those of PCA. But for different flavor substances, DFA had better distinguishable capacity better than those of PCA. Our finding agrees well with that of Wei (Wei et al., 2013) who also analyzed by principal component analysis (PCA) and discriminant function analysis (DFA) for category classification.

| Sensory evaluation of the umami intensity of flavor enhancers
The umami intensity of each sample obtained by the electronic tongue analysis was also assessed by sensory, and the result was shown in Table 3. The umami score of the flavor substances from the human sensory evaluation ranged from 1.5 (sodium aspartate) to 8.6 (disodium adenylate). The umami score of the flavor substance at 1.0 g/L was higher than those at 0.5 g/L. No significant difference was found on monosodium glutamate, sodium aspartate, and disodium uridylate regarding the umami score. The umami intensity The higher umami taste score has the higher E-tongue sensory value.
The result suggested that human sensory evaluation and electronic tongue detection were consistent in the evaluation of the umami taste intensity.

| CON CLUS ION
In this study, the electronic tongue analysis was able to identify the tongue is a valuable tool to rapidly analyze the flavor as a partial replacement for human sensory.

ACK N OWLED G M ENTS
The authors gratefully acknowledge the support provided by the funding from The Establishment of the Tabocco

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

E TH I C A L A PPROVA L
This study does not involve any human or animal testing.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.