Microbe Selection and Optimizing Process Parameters for Degradation of Glucosinolates in Rapeseed Meal by Box – Behnken Design

The present study applied Aspergillus oryzae, Aspergillus niger, Penicillium purpurogenum, Trichoderma sp. MAB-2010b and Saccharomyces cerevisiae in solid state fermentation to degrade the glucosinolates in rapeseed meal. In addition, SDS-PAGE was used to determine the effect of hydrolysis of those five microbes on peptide size in rapeseed meal. The results indicated that the solid state fermentation with S. cerevisiae degraded the glucosinolates more than those with other microbes. The peptides were hydrolyzed by S. cerevisiae to a greater extent than others. Thus the following procedure was just focused on the solid state fermentation with S. cerevisiae. Box-Behnken design of response surface methodology was applied to optimize the substrate to water ratio, inoculum amount, and duration. The glucosinolate level in rapeseed meal was as the response. The optimal conditions derived from response surface methodology for S. cerevisiae fermentation were: 1.0 of substrate to water ratio, 1.5 mL (equal to 5%) of inoculum amount, and 48 h of duration. The minimum content of glucosinolates was 0.46 μmol/g dry matter. S. cerevisiae used in the present study thus exhibit the potential use in large scale solid state fermentation for increasing nutrition quality of rapeseed meal.

Solid state fermentation (SSF) has been reported to improve the nutritional value of rapeseed meal by breakdown of antinutrients through production of exogenous enzymes, as well as using the antinutrients as substrate for microbial growth (Bau et al., 1994;El-Batal & Karem, 2001; Nair & Duvnjak, 1991;Rozan et al., 1996;Vig & Walia, 2001).SSF also can increase protein content duo to the synthesis of single cell protein (Frias, Song, Martinez-Villaluenga, De Mejia, & Vidal-Valverde, 2008; Hong, Lee, & Kim, 2004) and change the essential amino acid profile of the substrate (Yigzaw, Gorton, Solomon, & Akalu, 2004).Besides, the influence of SSF on protein peptide characteristics has been proved.Hong et al. (2004) reported that SSF with Aspergillus oryzae GB-107 reduced peptide size in soybeans and soybean meals.In order to make the SSF efficient, we used some molds from the different sources.They were Aspergillus oryzae, Aspergillus niger, Penicillium purpurogenum, Trichoderma sp.MAB-2010b.In addition, SSF with yeast has been tested for increasing nutritional quality of soybean product (Rashad, Mahmoud, Abdou, & Nooman, 2011).Thus, yeast could be used to increase the nutritional quality of rapeseed meal, and a commercial yeast product (Saccharomyces cerevisiae) was evaluated in current experiment.
Response surface methodology is a collection of statistical techniques for designing experiments, building models, evaluating the effects of factors, and analyzing optimum conditions of factors for desirable responses.It has been accepted as a convenient method for analyzing optimum conditions due to the minimized cost and time compared with the traditional single variable experiments (Vohra & Satyanarayana, 2002).The Box-Behnken is a design of response surface methodology, and is an independent quadratic design in that it does not contain an embedded factorial or fractional factorial design.In this design the treatment combinations are at the midpoints of edges of the process space and at the center.These designs are rotatable (or near rotatable) and require three levels of each factor (Gong et al., 2007).
The aim of the present study thus was to compare the effect of A. oryzae, A. niger, P. purpurogenum, Trichoderma sp.MAB-2010b and S. cerevisiae on degradation of glucosinolates and change of peptide characteristics in rapeseed meal.The most effient one was selected for further opimizing SSF conditions in terms of degradation of glucosinolates, including substrate to water ratio, inoculum amount and duration, by employing three-level and three-variable Box-Behnken design.

Processing of Rapeseed Meal
Rapeseed meal was purchased from Ningbo Chia Tai Agriculture Co., Ltd (Ningbo, China), and was ground using a FW100 grinder (Tianjin Taisite Instrument Co., Ltd, Tianjin, China) equipped with a 0.4 mm screen.Rapeseed meal was dried carefully (30 min at 60°C) before it was stored in a drier containing silica until use.
~ 3 ~ MAB-2010b (HQ829121) were isolated by our laboratory.S. cerevisiae (product No.: Q/YB.J02.07) was purchased from Angel Yeast Co. Ltd (Yichang, China).A. oryzae, A. niger, P. purpurogenum and Trichoderma sp.MAB-2010b were transferred to Potato-Dextrose Agar (PDA) medium (Xinran Biological, Shanghai, China).The strain then was incubated at 28 • C for 7 days.The spores on the surface of PDA medium were scratched by an inoculation loop and washed carefully by 1-mL sterile water.The washed spores were further served as the inoculum (CFU was adjusted to be 1×10 7 by adding sterile water) for SSF. S. cerevisiae preparation was carried out according to the product specification.In general, the dry powder (5 g) was mixed with distilled water (50 mL) and dextrose (1 g) in a 100 mL beaker, and was incubated at 40 • C for 1h.The CFU was adjusted to be 1×10 7 by adding sterile water.

Solid State Fermentation
Rapeseed meal (30 g) mixed with distilled water (30 mL) was transferred to a 150 mL incubation flask (ZP9-150, Shanghai Jiafeng Co. Ltd, Shanghai, China) equipped with ventilated plug as substrate.The inoculum amount was 1 mL for A. oryzae, A. niger, P. purpurogenum, Trichoderma sp.MAB-2010b, and S. cerevisiae.The incubation was performed at 28°C in a QYC-211 incubator (Shanghai Fuma Laboratory Equipment Co. LTD, Shanghai, China) for 72 h in triplicate.The fermented rapeseed meal was freeze-dried before it was ground and maintained at 4 • C until further chemical analysis.

Optimization of SSF
The most efficient microbe in degradation of glucosinolates and change of peptide characteristics in rapeseed meal was selected for opimizing SSF conditions.The Box-Behnken design with three independent variables (substrate to water ratio, inoculum amount and duration) of SSF and three levels, including five replicates at the central point, was used for fitting a second-order response surface.The range and levels of the independent variables in the Box-Behnken design are indicated in Table 1.A total of 15 runs were used to optimize the three independent variables.Each run was conducted in triplicate.The glucosinolate content in rapeseed meal after each run was taken as the dependent variable, and the mean value of three observations was used.A quadratic polynomial regression model was assumed for the dependent variable.The model proposed for each response of Y was: Where Y is the dependent variable, x i and x j are independent variables, represents the intercept term, represents the linear effect, represents the squared effect and shows the interaction effect.The Design Expert 7.1.3(trial version, Stat Ease Inc., Minneapolis, USA) computer program was used for estimating the coefficients of Equation ( 1) by regression analysis of the experimental data.

Determination of Glucosinolates
The determination of glucosinolates was conducted in accordance with the palladium oxide method developed by Wang (2009).In brief, 100 g rapeseed meal was placed in a 100 mL centrifuge tube, and treated with water bath (100°C,10 min).Afterwards, 8 mL distilled water (90°C) was added to the centrifuge tube, which was then placed in a water bath at 100°C for 30 min.The tube was stirred twice during incubation.After cooling, 2 mL distilled water was added to the centrifuge tube, and mixed with the treated rapeseed meal.The filtrate (0.5 mL) was collected to a 10 mL cuvette with plug.Sodium carboxymethyl cellulose solution (2 mL, 0.1%) and palladium chloride solution (1 mL, 0.004M) was added to the the cuvette in sequence.After being shaken, the filtrate was placed in an incubator (24°C, 3 h).The spectrophotomete was used to measure the absorption of the samples at 540 nm.Sodium carboxymethyl cellulose solution and palladium chloride solution were served as the reference solutions.The content of glucosinolates calculated as: 185.2×absorption value+0.2, in which 185.2 was obtained by the standard curve, and 0.2 was the adjust value.

Electrophoresis
Proteins in rapeseed meal and fermented rapeseed meal were extracted by a modified method based on Aluko and McIntosh (2001).Samples were finely ground with the A11 basic Analytical mill.Ground samples (125 mg) were mixed with 2.5 mL of 50 mM Tris-HCl (pH 8.0) containing 1% SDS, 5 mM DTT and 5 µg/mL protease inhibitor.Samples were then incubated on ice for 3 h, followed by 8 consecutive ultrasonications with ice, each lasting 1 min.Samples were kept on ice during ultrasonication.The samples were thereafter transferred to a 1.5-mL microtube containing glass beads (Qiagen, Hilden, Germany) and further disrupted in a FastPrep machine (FP120, BIO101, Savant) at 6,500 g for 45 sec, 5 time repeats with 1 min interval, on ice.The samples were then treated in an ultrasonication water bath (2 cycles of 5 min) prior to centrifugation at 16,000 g for 30 min at 4°C (Centrifuge 5415 R, Eppendorf, Hamburg, Germany).The supernatants were transferred to Eppendorf tubes and kept at -20°C overnight.The samples were concentrated from 380 to 40 µl using Nanosep® Centrifugal Devices (Pall Life Sciences, Ann Arbor, USA) before mixing with 40 µl SDS-PAGE loading buffer, and incubation at 65 °C for 5 min (Thermomixer comfort, Eppendorf, Hamburg, Germany).Mixed samples were then centrifuged at 16,000 g for 15 min at 4°C.The polyacrylamide separating gel (12%) and stacking gel were prepared by using P0012A gel parparation kit (Beyotime, Jiansu, China).SDS-PAGE was conducted in a Bio-Rad Electrophoresis system (Bio-Rad, Hercules, CA) with polyacrylamide separating gels in 20x dilution of XT MOPS running buffer.Ten µl of samples were loaded respectively and run at 120 V for 1.5 h.After electrophoresis, the gel was fixed in a solution of 50% methanol, 15% acetic acid and 35% water with shaking (Heidolph Unimax 2010, Heidolph, Schwabach, Germany).The gel was further stained for 1h and destained until the background was clean before taking photos.

Statistics
Statistical analyses of glucosinolates in rapeseed meal after SSF with different microbes were carried out by using general linear models procedure in SAS (SAS, 1999).The results were subjected to one-way analysis of varinace, with microbe as the class variable.Significant difference (P<0.05) were ranked by Duncan's multiple range method.The results are presented as mean±standard error of mean (SEM).

Effect of SSF on Glucosinolates and Peptide
The content of glucosinolates in rapeseed meal after SSF with five microbes is presented in Table 2.The results showed that rapeseed meal fermented by S. cerevisiae resulted in lowest content of glucosinolates (32.8 mg/g, dry matter basis), which was significantly lower than the ones fermented by A. oryzae and Trichoderma sp.MAB-2010b.The highest content was in rapeseed meal fermented by the Trichoderma sp.MAB-2010b (67.5 mg/g, dry matter basis), followed by the A. oryzae (47.6 mg/g, dry matter basis).The peptides in rapeseed meal before and after SSF are presented in Figure 1.The results showed that SSFs with A. oryzae, A. niger, P. purpurogenum and Trichoderma sp.MAB-2010b did not affect the peptide size of rapeseed meal (lane B to E).However, SSF with S. cerevisiae affected the peptide size.The peptide with size of around 20 kDa was clearly increased in rapeseed meal fermented with S. cerevisiae (lane F).The same size peptide was not increased in rapeseed meal fermented by other microbes.

Optimization of SSF with S. cerevisiae
In general, SSF with S. cerevisiae showed the high level in degradation of glucosinolates and influence of peptide size.Thus, the optimization process was just carried out for SSF with S. cerevisiae.The design matrix of the variables in coded units is depicted in Table 3 along with the predicted and experimental values of the dependent variable.The quadratic model, representing the content of glucosinolates (Y) as a response of substrate to water ratio (A), inoculum amount (B) and duration (C), was expressed by the Equation ( 2).In order to obtain a better fit model, some of the statistically insignificant terms (P > 0.05), including B 2 and C 2 , were taken away from the model.
Y=14.59-37.40A-1.76B+11.22C+7.35AB-20.07AC+0.76BC+39.49A 2  (2) The statistical significance of Equation ( 2) was determined by Fisher's F-test.The ANOVA of the quadratic regression model (Table 4) demonstrated that Equation ( 2) was an extremely significant model at 99% confidence level based on P value (P<0.01).The substrate to water ratio (P<0.01),duration (P<0.05) and their interaction (P<0.05)significantly affected the content of glucosinolates in rapeseed meal.However, the inoculum amount, as well as the interaction between the inoculum amount and other two factors did not influence the content of glucosinolates.The graphs, representing the regression of Eq. ( 2), were obtained using the Design Expert (Figure 2).The graphs were generated for the pair-wise combination of the three variables, while inoculum amount was kept at the optimal level (1.5 mL) for the lowest content of glucosinolates.The optimal values of the three factors during fermentation for obtaining the predicted minimum content of glucosinolates (0.22 μmol/g, dry matter basis) derived from response surface were delineated as the followings: 1.0 of substrate to water ratio, 1.5 mL of inoculum amount and 48 of duration.

Validation of Optimal Conditions of SSF
An additional SSF experiment was performed with the optimal substrate to water ratio, inoculum amount and duration to confirm the optimal conditions.The experiment was performed in triplicate.The mean content of glucosinolates in fermented rapeseed meal was 0.46 μmol/g dry matter, which was close to the estimated value (0.22 μmol/g dry matter basis).The result thus corroborated the validity of the response surface model.

Discussion
The total glucosinolates in rapeseed meal were efficiently reduced by SSF with five selected microbes, in keeping with previous findings (Bau et al., 1994;Smits, Knol, & Bol, 1993;Vig & Walia, 2001;Wang et al., 2012).The degradation of glucosinolates in fermentation may be due to utilization of glucose and sulphur moieties of these compounds by microbial enzymes (Tripathi & Mishra, 2007).In the present study, the molds reduced the content of glucosinolates from 133.7 to 39.2, 42.6, 47.6 and 67.4 μmol/g (dry matter basis).Yeast (S. cerevisiae) exhibited higher efficiency in degradation of glucosinolates, and reduced the content of glucosinolates from 133.7 to 32.8 μmol/g (dry matter basis).However, the results from Wang et al. (2012) showed that yeast had a lower ability to reduce glucosinolates compared to molds in general.Some yeast fermentation even elevated the content of glucosinolates in rapeseed meal.The disagreement between that and current experiment may be a result of the different types of yeast and mold used.
The fermentation with S. cerevisiae increased the amount of small peptide with size of around 20 kDa in rapeseed meal.The increase may be due to partial digestion of peptides with size larger than 20 kDa by proteases secreted by the yeast.The types of peptide with 20kDa generated in the present experiment were unclear.It will be interesting if we could identify them in the following experiment.Secretion of proteases in many types of yeast has been proved, including S. cerevisiae (Ogrydziak, 1993).However, some previous studies also concluded that the molds had the capability of secreting proteases.The study conducted by Sardjono, Zhu, and Knol (1998) showed that A. oryzae obtained highest protease activity after a 48-hour fermentation.In addition, Hong et al. (2004) demonstrated that A. oryzae fermentation reduced the peptide size of soybean and soybean meal.Sandhya, Sumantha, Szakacs, and Pandey (2005) compared effect of SSF and submerged fermentation on production of proteases in A. oryzae, and proved the superiority of SSF over submerged fermentation.The capability in protease production of A. niger also was confirmed (Lopes et al., 2011).In contrast, Rodarte, Dias, Vilela, and Schwan (2011) demonstrated that P. purpurogenum did not show protease activities by hydrolysis of casein.There was no study concerning the protease production of Trichoderma sp.MAB-2010b.However, Kredics et al. (2005) reviewed the extracellular proteases of Trichoderma species.It may indicate that most of Trichoderma species secret extracellular proteases, and thus, may be able to digest peptide with large size to some extent.S. cerevisiae showed the higher capability in digestion of peptides compared to the other microbes in the present study,.This may be ascribed to that the efficiency of protease secretion and protease activity was different in SSF.Unfortunately, the protease activity was not detected in the present study.Thus, the protease activity of those microbes should be detected in the following experiment.In addition, the protease activity was affected by the different environmental conditions.Braaksma, Smilde, van der Werf, and Punt (2009) concluded that culture pH and nitrogen concentration strongly affected extracellular protease activities in fermentation of A. niger.The initial condition of SSF was set to be the same in the present study.However, the production of organic acids may be occurred during fermentation.Therefore, the pH was varied subsequently during fermentation, and was not the same amongst fermentations different microbes.
The optimal substrate to water ratio was 1.0 in SSF with S. cerevisiae.The moisture content of untreated rapeseed meal used in current work was 10%.Thus, the 1.0 of substrate to water ratio was equal to 55% of moisture in the solid matrix.This level was lower than the suggested moisture (60-72%) for yeast (Rodriguez-Leon, Soccol, Pandey, & Rodriguez, 2007).This variance may be caused by different water activity, which is an important factor for solid matrix.The optimal inoculum amount was 1.5 mL, which was equal to 5% in basis of that the amount of rapeseed meal was 30g in fermentation.To our knowledge, the studies concerning rapeseed fermentation by yeast were limited.However, the molds have been tried in SSF with rapeseed meal as substrate.Gu et al. (2010) attempted to increase the quality of by mixed SSF with A. oryzae and T. viride.The optimal inoculum amount was 30% in that study, which was higher than that in the current study.The different inoculum amount may be due to the different type of microbes.In principle, the hyphae of A. oryzae and T. viride can grow on particle surfaces and penetrate intrate into the inter-particular spaces, thereby colonizing solid substrates rapidly than yeast (dos Santos, da Rosa, Dal'Boit, Mitchell, & Krieger, 2004).Therefore, the inoculum amount would be lower in fermentation with A. oryzae and T. viride than with S. cerevisiae.The duration significantly affected the content of glucosinolates.A previous experiment conducted by Gao, Wang, Zhu, and Qian (2013) also showed the significant effect of duration in SSF.The optimal duration was 48 h, which was the minimum in the setting range.It indicated that fermentation may reduce other components more rapidly than glucosinolates of rapeseed meal in the setting range.Thus, the relative content of glucosinolates was showed to be increased.However, the total amount of glucosinolates was not increased indeed.
In conclusion, SSF with S. cerevisiae degraded the glucosinolates more than SSF with other four microbes.In addition, the peptides were hydrolyzed by S. cerevisiae to a greater extent than other four strains.The optimal conditions derived from response surface methodology for S. cerevisiae SSF were: 1.0 of substrate to water ratio, 1 mL of inoculum amount, and 48 h of duration.The minimum content of glucosinolates was 0.46 mg/g dry matter.S. cerevisiae employed in the present study thus exhibit the potential use in large scale SSF for improving the nutritional quality of rapeseed meal.

Figure 1 .
Figure 1.SDS-PAGE of rapeseed meal before and after solid state fermentation with different microbes Lane A represents rapeseed meal without solid state fermentation; Lane B represents rapeseed meal fermented by Aspergillus oryzae; Lane C represents rapeseed meal fermented by Aspergillus niger; Lane D represents rapeseed meal fermented by Penicillium purpurogenum; Lane E represents rapeseed meal fermented by Trichoderma sp.MAB-2010b; Lane F represents rapeseed meal fermented by Saccharomyces cerevisiae.

Figure 2 .
Figure 2.Response surface 3D plot of glucosinolates in rapeseed meal during solid state fermentation with Saccharomyces cerevisiae, showing the effects of substrate to water ratio and duration (Inoculum amount was 1 mL that was the optimal level)

Table 1 .
Level and code of independent variables chosen for Box-Behnken design

Table 2 .
Content of glucosinolates in rapeseed meal after solid state fermentation with different microbes cGlucosinolates in raw rapeseed meal was 133.7 μmol/g dry matter; Values are expressed as mean ± SEM.Different letters indicate significant differences (P < 0.05).

Table 3 .
Box-Behnken design matrix with the experimental and predicted values of gluocsinolates content in Saccharomyces cerevisiae fermented rapeseed mealValues are expressed as mean ± SEM.