Skim yoghurt with microbial transglutaminase: evaluation of consumer acceptance

ABSTRACT The decrease in the fat content of yoghurt causes sensory modifications. Microbial transglutaminase (TG) has been proposed as an alternative for reducing the problems caused by the fat reduction. This work deals with the evaluation of the sensory profile, acceptance and preferences of skim yoghurt produced with TG at pilot scale. A group of 124 consumers of yoghurts were randomly selected. A lower firmness has been observed in yoghurt without TG (NoTG). TG Yoghurt has shown a lower whey odor and less acid taste than NoTG yoghurt. Free whey was not observed in the TG yoghurt. Despite of those differences, consumers only observed minor textural differences. In terms of overall acceptance and preference, no statistically significant differences were observed. The use of TG avoids the addition of milk protein or other texture additives into the yoghurt, decreasing production cost with the same overall acceptance and preference by consumers.


Introduction
Microbial transglutaminase (TG) is an enzyme that form bonds between protein molecules (inter-and intramolecular crosslinks between γ-carboxylamide groups of glutamine residues and ε-amino groups of lysine residues). TG isolated from Streptomyces mobaraensis has been available at commercial scale several years ago. TG is widely used in food industry due to its capacity to improve functional properties of proteins like firmness, elasticity, viscosity, heat stability and waterholding capacity. Several studies have shown that the casein is an excellent substrate for TG (Dube, Schäfer, Neidhart, & Carle, 2007;Kuraishi, Yamazaki, & Susa, 2001;Motoki & Seguro, 1998;Routray & Mishra, 2011).
The lactose fermentation produces lactic acid, which decreases pH and denatures milk protein to form a coagulated gel. It gives its texture and its characteristic tang to the yoghurts. By-products are also produced during this process being responsible of specific aroma and flavor of yoghurt (Tamine & Robinson, 1999). Yoghurts can be classified into two types: (a) set-style yoghurt is made in the sale packaging with a continuous gel structure in the final product and (b) stirred yoghurt where the gel is disrupted by stirring before being packaged.
Yoghurt is a popular product worldwide. It has gained widespread consumer acceptance as a healthy food (Adolfsson, Meydani, & Russell, 2004;Fisberg & Machado, 2015;Glanville, Brown, Shamir, Szajewska, & Eales, 2015;McKinley, 2005). Today consumers continually search for new and unique food products while trying to maintain healthy eating habits (Karagül-Yüceer, Coggins, Wilson, & White, 1999;Zhi et al., 2017). Around 16 kg of fermented milks are consumed per capita annually in Europe (https://ec.europa.eu/agriculture/ milk/origin-labelling/com-2015-205_es.pdf). Skim set-type yoghurt is the most consumed despite the great variety of yoghurts offered by the market in order to satisfy all palates and meal occasions (textures, flavors or fat content).
Analytical sensory evaluation and consumer testing allow identifying the key sensory attributes that affect consumer preferences for food products providing powerful information for product development. The sensory properties of products are generally evaluated using descriptive methods, such as conventional profiling (Zhao, Feng, Ren, & Mao, 2018). In parallel, determining hedonic responses to products involves asking consumers to identify their preferences (Dabija, Codină, & Gâtlan, 2018). To establish relationships between sensory data and consumer preferences is the main interest of these studies to reduce the gap that may exist between product development and consumer expectations. Consequently, it reduces the number of failed new products.
As a result, improving the identification of drivers of liking appears to be a necessary part of the development of new yoghurts (Guinard & Mazzucchelli, 1996;Masson, Saint-Eve, Delarue, & Blumenthal, 2016). To satisfy increasing consumer demand for low-fat yoghurt, manufactures increasingly propose yoghurts with low-fat content. However, the consumers' demand for skim yoghurt also included a similar sensory quality as fat yoghurts (Sahan, Yasar, & Hayaloglu, 2008).
Fat is well known to have an impact on food texture, aroma and taste. In order to improve texture and physical properties, fat is replaced by milk proteins or gums but changes in the sensory properties are difficult or impossible to avoid. Thus it seems that more research is needed to maintain the fat yoghurt sensory characteristics on reduced-fat or skim yoghurts (Routray & Mishra, 2011). Skim milk powder is traditionally used for decreasing syneresis and improving gel texture in yoghurt production (Akalin et al., 2012). However, fortification ingredients on low or skim yoghurts can negatively affects sensory quality and consequently decreases in acceptance (Isleten & Karagul-Yuceer, 2006;Matumoto-Pintro, Rabiey, Robitaille, & Britten, 2011;Ünal & Akalin, 2013). Therefore, it is necessary to find alternatives.
Enzymatic cross-linking may be an alternative method to the addition of protein or gums in the production of skim yoghurt. Problems such as low strength and syneresis in skim yoghurt can be solved by a TG treatment (Faergemand, Otte, & Qvist, 1998;Kuraishi et al., 2001;Motoki & Seguro, 1998). Consequently, TG treatment can be possibly a useful method in production of skim and low-solid content yoghurts without using additivities and texturally similar to yoghurt with regularfat and so-called SNF (all the solids in milk except fat) content (Yüksel & Erdem, 2010).
TG treatment in the yoghurt production can be performed by two methods: acting TG prior to the fermentation step (Lorenzen, Neve, Mautner, & Schlimme, 2002) or simultaneous addition of TG and the starter culture (Yüksel & Erdem, 2010). Simultaneous addition does not require additional processing step like enzyme inactivation. TG is gradually inactivated by acidification of the media. The physicochemical and sensory characteristics of both have been analyzed by our group (García-Gómez, Romero-Rodríguez, Vázquez-Odériz, Muñoz-Ferreiro, & Vázquez, 2018, 2019. However, no studies on acceptance and preference of TG yoghurt by consumer were found. Therefore, the aim of this work was to study the acceptance and preference of a set-style skim yoghurt manufactured with a microbial TG produced in our laboratory.

Materials and methods
Skim milk was measured before use, obtaining the following composition: fat, 0.04%; protein, 3.30%; and total dry matter, 9.35%. Microbial TG was produced in our laboratory following the manufacture process described in our Spanish patent (Vázquez & Guerra-Rodriguez, 2012). The activity measured before use was 757 U/g. A colorimetric procedure was used to determine TG activity before use. Briefly, N-α-CBZ-gln-gly (Sigma-Aldrich Corp, St. Louis, MO, USA) was used as substrate. A calibration curve was made using L-glutamic acid γmonohydroxamate (Sigma-Aldrich Corp, St. Louis, MO, USA). One unit of TG is defined as the formation of 1 μmol l-glutamic acid γ-monohydroxamate in 1 min at 37°C (Grossowicz, Wainfan, Borek, & Waelsch, 1950).

Preparation of yoghurt/samples
The manufacturing of yoghurts was performed in a pilot plant (Aula de productos lácteos, USC, Lugo, Spain). A setstyle skim yoghurt with milk powder was used as control (NoTG yoghurt). The skim milk (120 L) was split into two batches, one of 60 L for TG yoghurts and the other of 60 L for the control yoghurts (NoTG yoghurts). For the control batch, the milk was standardized for achieving an increase in dry matter without increase in fat prior to the heat treatment. It was blended with skim milk powder to achieve the following values: 0.05% fat; 3.80% protein; and 10.45% total dry matter. For TG batch, it was used a dose of 0.76 U/g of milk protein. The dose was calculated after conducting preliminary studies (García-Gómez et al., 2018, 2019. Culture starter of 0.2 g/L of milk was used. Inoculation of microbial TG was simultaneous with the starter culture. The milk used of the two batches was pasteurized at 95°C for 5 min and homogenized (200 + 50) bar. Then they were cooled to incubation temperature at 43°C. Freeze-dried lactic culture (LyoCulture Dairy, BDF Natural Ingredients, Girona, Spain) at 0.2 g/L of milk was inoculated and blended into the milk. The mixture was poured into 125 g plastic cups and incubated at 43°C. Each batch was formed by 358 yoghurts. The coagulation of milk was monitored by pH change during the incubation period until a pH of 4.6 was attained. TG was gradually inactivated with the decrease of pH. TG activity was not detected at the end of the fermentation. Then, yoghurts were moved to a cool room and stored at 4°C. Samples were evaluated by the consumers and the trained panel after 5 days.

Descriptive sensory analysis
The samples were evaluated by a trained panel performed by 17 assessors with previous experience stablishing yoghurt's sensory profiles. The assessments were carried out at a sensory laboratory room. The samples were randomly presented to the judges coded with three-digit in 125 g plastic cups. Sensory tests were performed 4 days after the yoghurt production. The serving temperature of the samples was 8°C. Tap water and crackers were available. Descriptors were quantified using 10-cm unstructured intensity scale line anchored by appropriate references for each of the tested attributes. Definition and evaluation procedure for each attribute are shown in Table 1.

Consumer panel analysis
Simultaneously to the trained panel analysis, the hedonic test was conducted in Lugo (NW Spain). A group of 124 consumers of yoghurts were randomly selected and invited to take part in the test. Regard to gender, 64.5% were women and 35.5% were men. In terms of age, 37.9% were between 18 and 39 years old, 39.5% were between 40 and 59 years old and 22.6% were over 60 years old.
The two samples were randomly presented to the consumers (such as each sample was scored individually) at 8°C, served in 125 g covered plastic cups and coded with three-digit numbers using a balanced complete block design. Participants were instructed to drink water between samples to clean the palate.
Participants were asked about purchasing habits and frequency and motivation for yoghurt consumption. In the acceptance test, appearance, odor, aroma, texture and overall acceptability have been assessed using a 9-point hedonic scale: 1dislike extremely, 2dislike very much, 3dislike, 4dislike slightly, 5neither like nor dislike, 6like slightly, 7like, 8like very much and 9like extremely (Majchrzak, Lahm, & Dürrschmid, 2010). Respondents were also asked about preference, if they would consume any of the products or if they would totally discard some of them, as well as the reasons for choosing or discarding the samples.

Statistical analysis
The t-Student test was applied to examine the differences between trained judges' scores for TG and NoTG yoghurts. Previously to compare the difference between quantitative variables means, the distribution of data was studied by the Shapiro-Wilk test. A normal distribution was assumed when the p-value was significant (p-value < 0.05). For those that did not comply with the assumption of normality, the Wilcoxon test with Bonferroni correction was applied.
The Pearson chi-squared test of independence was applied in order to analyze the relation between the consumer profile and the frequency or the type of yoghurt consumed. Wilcoxon test for comparing the hedonic score of the two yoghurts was proposed in terms of hedonic attributes acceptance. Then, Spearman test was performed to study the correlations between the acceptance of hedonic attributes and the overall acceptance for both yoghurts.
Respondents were grouped through cluster analysis applying the conglomeration of ward and a biplot with the groups and the hedonic attributes correlated with overall acceptance were drawn. Finally, multiple correspondence analysis (MCA) was performed in order to visualize the associations between the groups of consumers depending on acceptance, the preference and the motivation for the yoghurt consumption. Statistical calculations were performed using Statistics 20 software for Windows (IBM, Armonk, NY, USA) and R (R Core Team, 2018).

Descriptive sensory analysis
The yoghurts were evaluated by a trained panel. The sensory profiles obtained were different for each kind of yoghurt as can be seen in Figure 1. The trained panel perceived more syneresis in NoTG yoghurt than in TG yoghurt. Significant differences (t-Student test) were observed for several attributes. Creaminess (p-value = 0.0005), whey odor (p-value = 0.0011), density (p-value = 0.0147) and acid taste (p-value = 0.0138) were perceived higher in NoTG yoghurt by the trained panel. Meanwhile firmness (p-value = 0.0051) was perceived higher in NoTG yoghurt. Concerning the acidity, other studies did not find noticeable differences in acidity between camel's milk yoghurt treated with TG and a control without TG (Abou-Soliman, Sakr, & Awad, 2017). Other authors also found that TG yoghurt was slightly less acid taste and more firmness than no treated yoghurt. Although contrary to our results, they considered it with less odor whey and slightly more creaminess (Faergemand et al., 1998).

Consumer profile
Most of the respondents share the responsibility for purchasing (64.6%), a 30.6% were solely responsible for purchasing food at home and only 4.8% never made it. Overall, 93.5% Table 1. Definition and evaluation procedure for the sensory attributes considered in this work. like to try novel foods. Yoghurt consumption was high since 49.2% of the respondents consumed yoghurt daily and 25.8% do it 3-5 times per week. This agrees with other studies where 42% of the panelists consumed yoghurt once a day, 47% consumed yoghurt once a week and 11% consumed yoghurt once a month (Hekmat & Reid, 2006). In our study, significant differences were detected by the chi-square test of independence on consumer frequency related with age (p-value = 0.01). Daily consumers were observed in the elderly and weekly in the younger consumers. This does not agree with a previous study where yoghurt consumption was higher for younger respondents (Valli & Traill, 2005). Yoghurt most frequently consumed was low and regular-fat set type and only 1.6% of the respondents were consumers of stirred yoghurt. Yoghurt type consumed was not related with gender and age, except in lowfat yoghurt with fruit pieces since were consumed mainly by women (p-value = 0.024). This agrees with previous studies where low-fat was an important factor for women (Valli & Traill, 2005). Respondents were asked about the motivation for yoghurt consumption. The main reason cited was the aroma (80.6%), the 54.8% of the respondents pointed the texture, for the 35.5% health was the motivation and others were the price, calories and without additives with 24.2%, 21.8% and 0.8%, respectively.

Consumer acceptance
There was a high share of respondents scoring both yoghurts with "like" (Table 2). Thus, in terms of consumers' scores, both TG and NoTG yoghurts were scored similar. Wilcoxon test showed only statistically significant differences among yoghurts in terms of texture (Z = −3.273, p-value = 0.001) because the consumers gave higher scores to the NoTG yoghurt. Some authors have found that yoghurt treated with was better scored than control without TG according to texture and aroma. It was although found in yoghurt control with any other additive or component used (Aprodu, Gurau, Ionescu, & Banu, 2011). The regular fat yoghurt obtained using 75 U/L and adding TG simultaneously with the starter culture was better scored than the control yoghurt in appearance and texture. No differences were detected in aroma (Mahmood & Sebo, 2012). In contraposition, negative effects in TG addition in the aroma and odor evaluation were found. It was probably due to the adverse effect caused by TG addition on yoghurt acidity and acetaldehyde production. This is the main compound responsible for the aroma (Ozer, Avni Kirmaci, Oztekin, Hayaloglu, & Atamer, 2007).
The aroma and the texture are the hedonic attributes that are more related with overall acceptance for both yoghurts. When the Spearman test between the scores for acceptance was performed, it was observed that a correlation for the Figura 1. Perfiles sensoriales obtenidos por el panel entrenado para el yogur elaborado con transglutaminasa microbiana (TG) y el yogur control (NoTG) elaborado con leche en polvo desnatada. Significación estadística: *p-value < 0,05, **p-value < 0,01 and ***p-value < 0,001. Table 2. Percentage of acceptance scores obtained for the yoghurts produced with microbial transglutaminase (TG) and control yoghurt (NoTG) performed with powdered skim milk.
Tabla 2. Porcentaje de puntuaciones de aceptación obtenidos para los yogures producidos con transglutaminase microbiana (TG) y para el yogur control (NoTG) elaborado con leche desnatada en polvo. aroma and overall acceptability of TG and NoTG yoghurts was r = 0.874 and r = 0.796, respectively. It was also detected that a correlation for texture and global acceptance was r = 0.774 and r = 0.641, respectively, for TG and NoTG and p-value < 0.0001.
Aroma, texture and overall acceptability were used to perform the cluster analysis. Three clusters of respondents were identified as shown in Figure 2. A biplot is shown for illustrated data acceptance in each cluster (Figure 3). Cluster 1 comprised 50.81% of respondents who pointed TG yoghurt with high scores. Second cluster (28.23% of respondents) comprised those respondents who showed a high acceptance toward NoTG yoghurt. Cluster 3, which comprised 20.97% of respondents, groups those respondents with a low acceptance for both yoghurts. A relationship among gender, age or the affinity to taste novel food was not established in the consumers of each cluster. The Pearson chi-squared test of independence was not significant.

Consumer preference
The main motivation toward set-type skim yoghurt produced with powdered skim milk (NoTG yoghurt) or TG yoghurt was the taste. More than 70% of the respondents would consume one of the two proposed yoghurts. The main reasons are that they liked taste (80%), texture (33%) and for health (12%). Health is an important driver for skim yoghurt consumers (Valli & Traill, 2005). NoTG yoghurt was chosen as preferred by 55.6% of the respondents. However, the difference was not statistically significant (p-value = 0.2087). An 86.8% of those consumers who preferred TG yoghurt would consume it due to its aroma and 13.2% preferred it because they like its texture. In those who have chosen NoTG as a preferred yoghurt, 75.5% would consume it because of the aroma and 49.0% due to its texture. Despite of preferring yoghurts (TG or NoTG), only 9.7% and 12.1% of consumers have discarded the consuming of TG and NoTG yoghurts, respectively.
MCA factor map was performed with the aim of relating acceptance, preference and yoghurt consume motivation. Results are shown in Figure 4. The first group is the consumers with high scores in aroma, texture and overall acceptance toward TG yoghurt. Consequently, they preferred TG yoghurt. Regard the consume motivation, health was pointed as important. The sensory properties (aroma and texture) were not pointed as important. The second group are those that shown more acceptance toward NoTG yoghurt and predictably preferred it. The last group was comprised of consumers that shown a low acceptance for both yoghurts. They did not opt for either of the two yoghurts. They pointed as motivation of consume the aroma and texture.

Relation between trained panel and consumers
The small significant differences in acid taste found by the trained panel are valued positively by consumers even though differences in acceptability are not statistically significant. Differences in odor whey were detected by trained panel but they were not detected by the consumers since both yoghurts were similar pointed in appearance. Creaminess perception (thickness and smoothness) was correlated to consumer acceptance in a wide range of dairy products, including yoghurts (de Wijk, Terpstra, Janssen, & Prinz, 2006;Guinard & Mazzucchelli, 1996;Janhøj, Bom Frøst, & Ipsen, 2008;Janhøj, Petersen, Frøst, & Ipsen, 2006). Oppositely, it was suggested that sensory texture properties related with the creaminess are not significantly correlated with consumer acceptability of natural yoghurts (Jaworska & Hoffmann, 2008).
For US consumers, the relatively high extent of sourness along with the intensity of acetaldehyde (the key volatile compound of yoghurt) has resulted in low consumer acceptance but cannot be used for predicting consumer preferences for unflavored yoghurt . Other findings identified that many plain yoghurts were simply too sour for many consumers . 1 3 2 Figure 2. Cluster dendrogram of aroma, texture and overall acceptance. Cluster 1 = high acceptance for TG yoghurt; Cluster 2 = high acceptance for NoTG yoghurt; Cluster 3 = low acceptance for both yoghurts.
Figura 3. Representación biplot de la textura, al aroma y de la aceptación global con los gupos obtenidos del análisis cluster. Cluster 1 = aceptación alta para el yogur con TG; Cluster 2 = aceptación alta para el yogur NoTG; Cluster 3 = aceptación baja para ambos yogures. Figure 4. Multiple correspondence analysis (MCA) factor map (1 is cluster 1, 2 is cluster 2, 3 is cluster 3, TG represents those consumers who have preferred TG yoghurt, NoTG represents those consumers who have preferred NoTG yoghurt; aroma_Y when aroma has been a consume motivation, aroma_N when aroma has not been a consume motivation, texture_Y when texture has been a consume motivation, texture_N when texture has not been a consume motivation, health_Y when health has been a consume motivation, health_N when health has not been a consume motivation).
Sensory characteristic of yoghurt has a large effect on consumer acceptability (Delikanli & Ozcan, 2014;Grygorczyk, Lesschaeve, Corredig, & Duizer, 2013;Pereira et al., 2003). Although not all sensory texture properties and aroma attributes were correlated with yoghurt consumer acceptability (Jaworska & Hoffmann, 2008), absence of free whey was determinant for overall acceptance of yoghurts (Lee & Lucey, 2010). Statistically significant differences were observed among yoghurt samples only in terms of texture because the consumers gave higher scores to the yoghurt produced with powdered skim milk (NoTG). The consumers valued better high density and creaminess and less firmness of NoTG yoghurt.

Conclusions
Sensory profiles performed by the trained panel were different for both yoghurts. TG yoghurt was firmer than NoTG yoghurt. This fact has decreased the creaminess perception. TG yoghurt exhibits lower odor whey and acid taste. Despite these differences, consumers only perceived textural differences. TG yoghurt presented lower textural acceptance than NoTG yoghurt. Consumers prefer yoghurt with a creamier, dense texture and low firmness. However, in terms of overall acceptance and preference, no significant differences were observed. Therefore, the use of TG avoids the addition of milk protein or other texture additives, decreasing production cost and the same overall acceptance and preference by consumers.

Disclosure statement
No potential conflict of interest was reported by the authors.