Retracted: Preparation and optimization of soy (Katul cultivar) protein isolate cold‐set gels induced by CaCl2 and transglutaminase

Preparation and optimization of soy (Katul cultivar) protein isolate cold‐set gels induced by CaCl2 and transglutaminase. Food Science & Nutrition, https://doi.org/10.1002/fsn3.3158. The above article, published online on December 8, 2022 in Wiley Online Library (https://wileyonlinelibrary.com), has been retracted by agreement between the journal Editor in Chief Y. Martin Lo, and Wiley Periodicals LLC. The retraction has been agreed upon due to an error in which the incorrect version of the article was published.


| INTRODUC TI ON
Protein gelling is an important functional property because it contributes greatly to the sensory and textural properties of foods. Most recently, the gelation of globular proteins at room temperature, known as "cold gelation", has received a lot of attention (Ingrassia et al., 2019).
As they induce gelation at lower protein concentrations and lower temperatures than other methods, therefore permitting the incorporation of thermolabile compounds into formulations with lower losses (Zheng et al., 2019). This is feasible because, in these techniques, denaturation and aggregation arise throughout a preliminary preheating step, separate from the gelation process, which can be induced, for example, through salt addition (Kuhn et al., 2010). Calcium ions, which are the most widely used cations in foods, interact with proteins, and as a result, influence cross-linking within a gel and alter its mechanical behavior (Sikorski et al., 2007). The interest in introducing plant proteins in food formulations has been growing due to the increase in the world's population in recent years, there has been an important change the properties of gelatin films produced from tilapia scales (Weng & Zheng, 2015). They found that when TGase (1%) was used, the strength of the gelatin films improved. In the dairy industry, TGase treatment has been successfully applied to prohibit syneresis and improve gel homogeneity in yogurt (Ray & Rosell, 2017). It was reported that crosslinking of milk whey by TGase led to a decrease in the degree of syneresis (Gauche et al., 2008). TGase can catalyze the acyl-transfer reaction among the γ-carboxamide groups of glutamine residues and the ε-amino groups of lysine residues in proteins, leading to inter-or intramolecular cross-linking (Romeih & Walker, 2017). Soy proteins are widely used in food products for their nutritional value and ability to improve texture (Tang, Wu, Yu, et al., 2006). They consist of two major components: glycinin and β-conglycinin. In food applications, their gelforming ability is of particular importance, and the heat-induced gelforming ability of soy proteins or their constituents has been widely investigated (Renkema & van Vliet, 2002). Soybean (Katul cultivar) is one of the most important cultivars cultivated in Iran. This cultivar is 173 cm in height, and drought tolerant. Katul cultivar has the highest amount of globulin protein (18.75 mg/g) and it is cultivated in Gorgan and Mazandaran provinces, North Iran (Arefrad et al., 2020). In this study, for the first time, soybean (Katul cultivar) protein isolate was used. There are few published data on the effect of TGase and CaCl 2 levels on the gelation of soy protein isolate (SPI) or the characteristics of the derived gel. Owing to the fact that the gelation properties of processed products are considerably protein-attribute dependent, understanding the gelation properties of SPI is necessary for the investigation of their potential application in food systems. Therefore, it was aimed to find the optimal gelation conditions concerning turbidity, syneresis, and textural and rheological characteristics.

| Materials
The soybean seeds (Katul cultivar) used in the present study were obtained from the Oilseeds Research Institute in Gorgan, a city in the east north of Iran. At first, the seeds were manually cleaned to remove all foreign materials, then milled using an electrical miller (model M20IKA). Full-fat soybean flour (FFSF) was defatted using hexane by mixing at a ratio of 1:5 (w/v) and continued stirring (300 rpm) at room temperature for 6 h. Hexane was removed using a Buchner vacuum funnel, and the flour was air-dried at room temperature. Defatted soybean flour (DFSF) was sieved using a 40-mesh sieve, was packed in polyethylene bags, and stored at 4°C until further use. All the chemicals were analytical grade and purchased from Sigma-Aldrich Co. LLC (USA). Microbial transglutaminase (TGase, Activa TI; 100 units of enzyme activity per gram of powder) was purchased from Ajinomoto Foods Europe SAS, France.

| Preparation of soy protein isolate (SPI)
SPI was prepared from defatted soybean flour based on Yancheshmeh et al. (2022) with slight modifications. In brief, 50-g defatted flour was stirred at 25°C for 1 h at pH 9.5, and then centrifuged at 5000× g for 20 min. The supernatant was collected and the pH was adjusted to the isoelectric point for soy protein (pH 4.5).
The mixture was centrifuged at 5000× g for 20 min to precipitate the protein. The precipitate was washed twice with deionized water and its pH was adjusted to 7 using 1 M NaOH. All extraction processes were done at room temperature (25°C). Finally, the protein isolate was freeze-dried and stored at 4°C for further analysis.

| Chemical composition of soy flour and SPI
The protein content of FFSF, DFSF, and SPI was measured based on the Kjeldahl method and assuming a (N × 6.25) conversion rate between nitrogen and crude protein. The fat content was determined according to AOAC 922.06 method based on the Soxhlet extraction method (Extraction system B-811, Buchi Switzerland). Also, the moisture and ash contents were measured using the analytical techniques of AOAC 925.1 and 923.03, respectively (AOAC, 1990).
All proximate composition data are reported on a dry weight basis (d.b.). The difference in mass was employed to calculate the total carbohydrate content in the sample.

| Experimental design using response surface methodology (RSM) to optimize soy protein gel preparation
The statistical design of the study was based on a face central composite design (FCCD) and analyses were conducted using RSM statistical package (Design-Expert version 6.01, Statease Inc., Minneapolis, USA). The effects of three independent variables (SPI concentration x 1 ; TGase concentration x 2 ; and CaCl 2 concentration x 3 ) at three levels on stiffness (N/m), syneresis (%), turbidity (Å), and tan δ were investigated. According to FCCD, a set of 20 experimental runs which consisted of 2 3 factorial runs (at ±1 level), 6 axial points (at α = 1), and 6 replications at the center point of the domain were performed.
In brief, protein isolate dispersions (8%-12% w/v) were prepared by heating in test tubes (at 90°C for 15 min). The dispersions were then stirred for 15 min and poured into plates, followed by the addition of TGase (10-50 U g −1 ) and CaCl 2 (0-0.60 M). The samples were incubated at 48°C for 2 h and then left in the refrigerator at 4°C for 24 h before each experiment.
To keep the effects of external variables to a minimum, the experiments were randomized. Equation (1) illustrates the proposed model for the response.
where β 0 is the constant term, β 1 , β 2 , and β 3 are the linear effects, β 1 β 1 , β 2 β 2 , and β 3 β 3 represent quadratic effects, and β 1 β 2 , β 1 β 3 , and β 2 β 3 introduce interactions. For each coefficient, significance was evaluated using a p-value of .05. Model analysis, lack-of-fit test, coefficient of determination (R 2 ), and adjusted-R 2 analysis were employed to analyze the adequacy of models. The responses obtained from laboratory experiments were compared with the predictions made by the models. The numerical optimization feature of Design-Expert was used to simultaneously optimize several responses. We chose the desired goal based on the highest stiffness, viscoelasticity of gels, and the lowest syneresis and turbidity.

| Rheological properties of SPI gels
Dynamic rheological measurements were carried out using a Physica MCR301 controlled stress/strain rheometer (Anton Paar GmbH, Germany) with a parallel plate geometry (diameter = 50 mm, gap = 1.0 mm). The samples were prepared beforehand using circular molds to obtain gelled disks with a diameter of 55 mm and a height of 1.50 mm. A small amount of mineral oil was added to the exposed edges of samples to conserve moisture. The temperature was monitored and maintained using a Peltier system (Viscotherm VT2, Phar Physica). The frequency varied between 0.1 and 100 Hz while the samples were kept at 25°C and within the identified linear viscoelastic region (at 0.1% strain) to disturb network formation as little as possible.
The values of dynamic rheological parameters, including elastic modulus (G′, Pa), viscous modulus (G″, Pa), and tan δ, were extracted and analyzed by the Physica Rheometer Data Analysis software (Rheoplus/32, version V3.40). The tests were conducted at least in duplicate.

| | Texture analysis
Penetration tests were performed on cylindrical gels (25 mm height × 25 mm diameter) using a texture analyzer at room temperature (TA. XT PLUS; Canners Ltd., Ontario, Canada). The plunger velocity was 1 mm s −1 . For the penetration test, a probe with a diameter of 4 mm was allowed to penetrate the samples (50% of the sample length). Force/deformation was evaluated for four samples from each treatment, and the mean initial slope (initial tangent modulus) of the force/deformation curves was recorded as gel stiffness (N/m).

| Syneresis
Syneresis of samples was determined according to the method of Aichinger et al. (2003) with some alterations. Gel samples were centrifuged at 1000 g for 20 min. The separated serum was removed with a Pasteur pipette, and syneresis (%) was calculated as the mass of released serum divided by the total mass of the gel before centrifugation, multiplied by 100. Three gel samples were analyzed for each treatment.

| Measurement of turbidity
A quantity of 0.5 ml of distilled water was added to 2.0 ml of each SPI gel in a special test tube (internal diameter of 2.5 mm, length of 6 mm) to measure turbidity (Hatta et al., 1986). Test tubes containing the sample and distilled water were placed directly in the cell holder of a spectrometer (Shimadzu UV-160A). Turbidity was recorded as the sample's absorbance at 600 nm (Hatta et al., 1986).

| Cryo-scanning electron microscopy (cryo-SEM)
The gel microstructure was analyzed by cryo-SEM. Briefly, 3μl samples were pipetted into a copper brass rivet, which was mounted onto the transfer rod of a Quorum PP2000 Cryo System. The sample was frozen to preserve the structure of the samples and allow the structure to remain stable during the microscopy analysis (Sriamornsak et al., 2008). The frozen sample was then transferred to the Quorum preparation chamber attached to the FEI Quanta 200F FEG ESEM. In the preparation chamber, the sample was frozen, fractured, and sublimated for 10 min at −90°C. Finally, the sample was transferred to the SEM chamber and imaged in a high vacuum at −140°C, by monitoring the secondary electrons generated by an electron beam of spot 3.5 accelerated to 5 keV with the ETD detector.

| Chemical composition
The proximate composition of whole soy flour (WSF), defatted soy flour (DFSF), and soy protein isolate (SPI) is summarized in Table 1. The protein content of WSF was 40.33% whereas after protein extraction, it raised to 90.75%. Moreover, the fat, ash, and moisture contents of SPI were distinctly lower than the flours due to performed extraction/ precipitation stages (Shokrollahi Yancheshmeh et al., 2022).

| Fitting the models
To analyze the empirical findings on syneresis, stiffness, tan δ, and turbidity for SPI gel, a multiple regression model was fitted to the (1) data using a polynomial quadratic equation which included linear, quadratic, and interaction terms for SPI, TGase, and CaCl 2 . Table 2 presents the sequential sum of squares and summary statistics for the models, as well as the results of lack-of-fit (LOF) tests.
The appropriate regression models (linear, interactive, quadratic, and cubic) were selected accordingly. According to the model summaries, the quadratic model presented the best fit for all responses, and the cubic model was found to be aliased. In addition, LOF tests indicated nonsignificant results for the quadratic models for all responses (Table 2). Quadratic models which had greater R-squared (R 2 ), adjusted R-squared (Radj 2 ), predicted R-squared (Rpred 2 ), and smaller standard deviation (Std. Dev.) were chosen for all the cases.
The results of ANOVA for the FCCD design matrix are summarized in Table 3. The selected models were highly significant for all responses (p < .05) and LOF results were insignificant (p > .05). Thus, the models are applicable for optimizing the conditions. Table 3 presents R 2 , adjusted R 2 , p-value, and the results of LOF tests for all dependent variables. Model adequacy was evaluated using the LOF test and R 2 . The LOF test failed to produce significant results (p > .05), which shows that the model fits the empirical data well and that the data points are all reasonably close to the surface predicted by the model. Small coefficients of variation (CV) attest to the high precision of the experiments.
The negligible difference between R 2 and Radj 2 values reveals the accuracy of the polynomial models (Rodrigo et al., 2012). In addition, adequate precision values for all four responses were greater than 4, which indicates that the quadratic models could navigate the design space.

| Gel stiffness
The response surfaces for gel stiffness are shown in Figure 1a-c.
With increasing CaCl 2 concentration and protein content, gel stiffness increased (Figure 1a). The effect of CaCl 2 on gel stiffness was more pronounced when higher protein content was used.  (Lee et al., 2017) and β-lactoglobulin gels (Mulvihill & Kinsella, 1988). Salt ions increased the strength of protein gels in other cases as well, such as soy protein isolate gel and whey protein gel (Hsia et al., 2016;Otte et al., 1999). Moderate concentration of salt ions induced a denser gel network of arachin-basil seed gum composite gels with higher G′, G″ (G′ > G″) and gel strength via the enhanced electrostatic, hydrophobic, and hydrogen interactions.
However, a higher concentration of salt ions by the salting-out effect caused excessive aggregation of arachin, and composite gels were destroyed, which further resulted in a loose structure of the gel with poor hydration properties (Yang et al., 2021). Figure 1b shows the interaction between TGase content and protein concentration. It can be seen that when TGase content increased from 10 to 50 U g −1 , the stiffness of SPI gel increased. This observation can be explained through the covalent bonding of proteins' lysine and glutamine residues, which enhances texture (Bönisch et al., 2007). At low protein concentrations, TGase caused a nonsignificant change in gel stiffness due to decreased substrate concentration (Ustunol, 2014).  Note: Results are expressed as means ± SD for three replications based on a dry weight basis except for the moisture content which is based on a wet basis.

| Turbidity
The structure of protein gels depends on the nature of the medium, protein concentration, pH, and the type and ionic strength of ions present in the medium (Clark & Lee-Tuffnell, 1998;Gosal & Ross-Murphy, 2000). The influence of protein content, CaCl 2 , and TGase on the turbidity of SPI gel is illustrated in Figure 2a-c.
By increasing both CaCl 2 and protein concentration, gel turbidity increased. The effect of protein content on this property was more pronounced compared to that of CaCl 2 concentration. Gels with 12% protein had the highest degree of turbidity. The positive effect of CaCl 2 concentration on gel turbidity may be due to the formation of structures that are large enough to scatter light (Bryant & McClements, 2000). Furthermore, Kitabatake  Zhu and Damodaran investigated the influence of CaCl 2 concentration on the turbidity of native whey protein isolate (Zhu & Damodaran, 1994) and reported maximum turbidity after treatment with CaCl 2 (40 mM) followed by incubation for 24 hours.
They stated that this effect may be associated with salting-in at low and high CaCl 2 concentrations and salting-out at 40 mM CaCl 2 . Conversely, Hongsprabhas and Barbut observed a positive relationship between the opacity of whey protein isolate gels and CaCl 2 concentration, and an inverse relationship between the opacity of whey protein isolate gels and protein concentration (Hongsprabhas & Barbut, 1997).
As seen in Figure 2b, when TGase concentration increased from 10 to 50 U g −1 and protein concentrations from 8 to 12 U g −1 at constant CaCl 2 concentration (0.3 M), the turbidity of SPI gel increased to 0.88. The turbidity of proteins is commonly determined by the colloidal characteristics of the proteins and the coagulum formed (Inouye et al., 2002). Hence, the observed increase in the turbidity of SPI gel induced by TGase can be attributed to the formation of insoluble aggregates or coagula. This result is consistent with those reported for the TGase-induced coagulation of soy protein isolate Tang, Wu, Yu, et al., 2006). In this research, the degree of turbidity varied between 0.5 and 0.88 under various gelation conditions. Minimum turbidity was found at the lowest TGase, CaCl 2 , and protein concentrations.

| Syneresis
Syneresis is defined as the loss of water during the aging of gels and reflects the instability of the gel network. SPI gels with higher protein concentrations demonstrated less syneresis compared F I G U R E 1 Response surface for the effect of CaCl 2 and protein concentration (a, TG = 30 U g −1 ), TG and protein concentration (b, CaCl 2 = 0.30 M), and CaCl 2 and TG (c, protein concentration = 10%) on the stiffness of SPI gel.
to those having lower protein concentrations. By increasing protein concentration from 8% to 12%, syneresis decreases ( Figure 3a) since gels with a higher protein concentration retain more water due to a denser network with greater capillary forces (Remondetto et al., 2002). It has also been stated that more elastic protein gels have higher water retention capacity (Kinekawa & Kitabatake, 1995). CaCl 2 concentration significantly influences the fundamental structure and mechanical properties of the SPI-WPI gel . According to the results of tan δ measurements, tan δ decreased and elastic behavior enhanced with increasing protein concentration (Figure 4a), leading to less severe syneresis. By increasing CaCl 2 concentration from 0.3 to 0.60 M, syneresis decreased (Figure 3a). The degree of syneresis is determined by gel density and how the gel forms a three-dimensional network (Moritaka et al., 2003). In the presence of CaCl 2 , the carboxyl groups present in the amino acid structure can act as binding sites for Ca 2+ . The intermolecular ionic interaction between carboxyl groups and Ca 2+ ions leads to an increase in the shrinkage of the network structure (Moritaka et al., 1995). Therefore, it is expected that the added CaCl 2 should reduce syneresis. A similar effect was observed in whey protein and SPI cold-set gels by increasing the concentration of Na + from 75 to 400 mM and CaCl 2 concentration from 10 to 20 mM, respectively (Barbut & Drake, 1997;Maltais et al., 2005), which suggests that the number of available sites interacted with water in the gel also play an important role in retaining water. The number of available sites increases in tandem with the increase in the concentration of monovalent cations, which attracts water molecules from the hydration layer on the surface of aggregated particles, and reduces syneresis (Kuhn et al., 2011).
Conversely, the syneresis increased at CaCl 2 concentrations below 0.30 M, possibly due to higher porosity (Remondetto et al., 2002;Roff & Foegeding, 1996). Theoretically, water-holding capacity and F I G U R E 2 Response surface for the effect of CaCl 2 and protein concentration (a, TG = 30 U g −1 ), TG and protein concentration (b, CaCl 2 = 0.30 M), and CaCl 2 and TG (c, protein concentration = 10%) on the turbidity of SPI gel.
syneresis are mainly determined by porosity and hydrophilic sites in the gel .
The use of TGase has proven to be an appropriate method for altering the technological characteristics of raw products (de Góes-Favoni & Bueno, 2014). For example, this enzyme forms cross-links between protein molecules and as a result changes protein properties, gelation ability, water-holding capacity, and thermal stability (Kuraishi et al., 2001). As can be seen in Figure 3b, increasing TGase concentration reduced gel syneresis. This effect can be attributed to the influence of TGase on α-amine groups of lysine residues in proteins, which resulted in an increase in firmness and viscosity, and improved water-holding capacity as a result of reduced syneresis (Kuraishi et al., 2001;Zhu et al., 1995). The proteins cross-linked with transglutaminase showed significantly improved water-holding capacity (5.2-5.6 g/g protein) compared with the control pea protein isolate (2.8 g/g) due to catalyzing covalent cross-linking between lysine and glutamine residues in forming inter-or intra molecular ε-(γ-Glu)-Lys polymers, which results in larger protein molecules and more intensive protein aggregation, favoring water-holding capacity followed by lower syneresis (Shen et al., 2022).

| Loss tangent (tan δ)
Viscoelastic behavior can be described by the loss tangent (tan δ). A similar effect was observed with increasing TGase concentration (Figure 4b,c). The low value of tan δ in the presence of TGase is due to cross-linking between Gln-Lys and other molecular interactions which result in ordered heteropolymers or aggregates. Such interactions and cross-linking produce gels with great storage moduli (Ramırez-Suárez & Xiong, 2003). This effect was more pronounced in the presence of high amounts of protein, possibly due to the greater number of cross-links, a higher level of polymerization, and greater viscosity (Gauche et al., 2008). With increasing transglutaminase concentration in faba bean isolate protein, tan δ value reduced to 0.04 which is caused by higher G′ (Nivala et al., 2021).
According to the results, the smallest loss tangent (0.14) was

| Optimization of the variables
Optimization was employed to find the best condition for protein gelation. Since the purpose of this study was to apply this protein F I G U R E 4 Response surface for the effect of CaCl 2 and protein concentration (a, TG = 30 U g −1 ), TG and protein concentration (b, CaCl 2 = 0.30 M), and CaCl 2 and TG (c, protein concentration = 10%) on tan δ of SPI gel.
in different food industries, different conditions were tested. Also, because reducing production costs is an important part of any industry, we tried to minimize enzyme use (due to its relatively high price) and protein content to reduce the cost of the finished product, while maintaining the desired properties of the final gel. The responses including syneresis, tan , and turbidity were set to be minimum and the gel's stiffness was set to be maximum.
A weight factor of 1 was selected for all individual desirability values in this design. The most value of 3, displayed by "+++", was chosen for all the responses. Considering these aspects, optimum gelation conditions were evaluated under different conditions (Table 4): 1. When protein content and TGase concentration were set to "is in range" and CaCl 2 concentration was set to "is equal to 0".
2. When protein content and CaCl 2 concentration were set to "is in range" and CaCl 2 concentration was set to "is equal to 0".
3. When all the variables were set to "is in range".
4. When protein content was set to "is in range" and other variables were defined as "minimize".

| Microstructure
Cryo-scanning electron microscopy (cryo-SEM) was used to investigate the microstructure of the gel (Zhou et al., 2017). The surface structures of freeze-fractured SPI gels under optimum gelation conditions are presented in Figure 5a,b.  (Gauche et al., 2008). These findings are in line with those published by Murekatete et al. (2014) and Barbut (1995), who reported that a denser structure was obtained at high ionic strengths.
Visualizing the gels' microstructure by cryo-SEM helped to clarify the impact of TGase and CaCl 2 on the gels' functional properties.

| CON CLUS ION
In this research, different levels of protein, TGase, and CaCl 2 produced gels with different levels of strength, syneresis, and turbidity. Numerical optimization determined the optimum preparation of soy protein isolate cold-set gels conditions based on the highest stiffness of gels, and the lowest syneresis, turbidity, and loss tangent as being protein content of 10.10%, CaCl 2 of 0.6 M, and TGase concentration of 46 U g −1 . At this optimum point, stiffness, turbidity, syneresis, and loss tangent were found to be 218.85 (g), 0.76, 0.34 (%), and 0.2 (%), respectively. Cryo-SEM analysis revealed that the three-dimensional structures of SPI gels made in the presence of both TGase and CaCl 2 were more compact than those formed with TGase or salt alone. Generally, SPI could be considered an efficient ingredient to make elastic gels. Therefore, new gelled foods with various textures and sensory specifications can be expanded.

ACK N OWLED G M ENT
The authors greatly thank Ferdowsi University of Mashhad for financial and laboratory support (grant number: 3/41123).

CO N FLI C T O F I NTE R E S T
All authors have declared that they have no conflict of interest in publishing this research.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated are available from the corresponding author upon reasonable request.

E TH I C S S TATEM ENT
This study does not involve any human or animal testing.

F I G U R E 5
The microstructure of SPI gels observed by cryo-SEM. Sample 1 contains 8.72% protein and 10 U g −1 TGase. Sample 2 contains 9.06% protein, 0.6 M CaCl 2 , and no TGase. Sample 3 contains 10.39% protein, 0.6 M CaCl 2 , and 41.73 U g −1 TGase. Sample 4 contains 9.92% protein and 10.02 U g −1 TGase. The (a) micrographs show the same areas as presented in the (b) micrographs but at greater magnification.