Optimisation of beef tenderisation treated with bromelain using response surface methodology ( RSM )

The purpose of this study is to determine the optimum condition for the tenderization of beef by bromelain using Response Surface Methodology (RSM). Initially, bromelain powder was produced from pineapple crown of variety N36. Production of the bromelain powder involves several process steps such as extraction, purification, desalting and freeze drying. The cube size beef of round part was treated with bromelain at different pHs of beef, immersion temperatures, bromelain solution concentrations, and immersion times according to the experimental design which was recommended by RSM of MINITAB software version 15. Beef tenderness was then measured by Texture Analyser. The MINITAB software Version 15 was used to optimise the tenderisation of beef by bromelain. The determination coefficient R was 99.97% meaning that the experimental data were acceptable. It was found that beef could be optimize tenderised 89.907% at the optimum condition at pH of beef of 5.4, immersion temperature of 60°C, bromelain solution concentration of 0.1682% and immersion time of 10 minutes. The verification value of beef tenderisation at the feasible optimum condition which was determined by experiment was 89.571%. Since the difference between the verification and predicted values was less than 5%, therefore, the optimum condition for the tenderisation of beef predicted by MINITAB software Version 15 could be accepted.


INTRODUCTION
Beef from Brahman breed is known for its toughness [1].Thus, it is unsatisfactory to the consumer, if the meat is so tough that it is difficult to eat [2].Therefore, numerous attempts have been made to encounter this problem and tenderising beef using bromelain becomes one of the widely used methods to improve the beef tenderness.This is because bromelain showed hydrolytic activity on the connective tissue leading to the better tenderisation of the tough meat [3].
Bromelain is a proteolytic enzyme found in pineapple plant [4][5] where it is accumulated in the entire part with different extents and properties depending on its source [6].Production of bromelain powder involves extraction, purification, desalting and drying processes which can be performed using varieties of procedure.In the present study, production of the bromelain powder from pineapple crown of variety N36 was carried out according to the method by [7] with slight modification.
The action of bromelain in tenderising beef is affected by pHs of beef [8][9] immersion temperatures, bromelain solution concentrations and immersion times [10,11].Research on the optimisation of beef tenderisation treated with bromelain has not been carried.Therefore, this study was conducted to determine the optimum condition for the tenderisation of beef by bromelain using Response Surface Methodology (RSM).
RSM is a collection of statistical techniques for designing experiments, building models, evaluating the effects of factors on the response and searching for the optimum conditions [12].The principal advantage of using RSM is the number of experimental runs required to evaluate multiple factors and their interactions can be reduced [13][14][15][16].Consequently, less time-consuming used to optimise a process compared to other approaches [17].

Materials
Pineapple crown of variety N36 from maturity index 2 was obtained from Peninsula Plantations Sdn Bhd at Simpang Renggam, Johor, Malaysia.Beef of round part from three years old bull of Brahman breed was purchased from a butcher at Kota Kemuning, Shah Alam, Selangor, Malaysia.Analytical grade of methanol and acetonitrile and food grade of acetic acid, sodium hydroxide, hydrochloric acid and sodium chloride were purchased from Merck Sdn Bhd (Petaling Jaya, Selangor, Malaysia).All other chemicals of analytical grade including standard bromelain were purchased from Sigma Technologies Sdn Bhd (Petaling Jaya, Selangor, Malaysia).Custom-made cation exchange resin and diafiltrator were purchased from IT Tech Research (M) Sdn Bhd (Subang Jaya, Malaysia) and Isetake Enterprise (Kajang, Selangor, Malaysia), respectively.

Production of Bromelain Powder
The pineapple crowns were extracted by crushing their small pieces using fruit juice processor with ratio of pineapple crown to purified water 1:1.The extract was filtered through a muslin cloth.Then, the pineapple crown extract was centrifuged at 360 x g for 10 min at 4℃.The clear supernatant was collected followed by purification process by Preparative High Performance Liquid Chromatography (HPLC) using cation exchange resin column of 21.2 mm internal diameter and 250 mm length.The eluents used were acetate buffer (25 mm, pH 4.0) and 1M sodium chloride (NaCl) solution.Removal of salt in the purified bromelain samples was carried out by continuous diafiltrator with hollow fiber membrane using purified water for exchange.Finally, the desalted bromelain solution sample was converted to powder form using freeze dryer (Christ alpha 1-4LD Plus model).

Optimisation of Beef Tenderisation Treated
with Bromelain Using RSM

Tenderisation of Beef Using Bromelain
Beef of round part was cut into cube size of approximately 2 cm 3 .Then, the beef cube was treated with bromelain at different pHs of beef, immersion temperatures, bromelain solution concentrations, and immersion times according to the experimental design which was recommended by RSM of MINITAB software Version 15 as shown in Table 2. Non-immersed or untreated beef cube was served as a control.

Texture measurement
Beef tenderness was measured by Texture Analyser TAX-T2i (Stable Micro Systems, Ltd., England, UK) using P2N needle probe with a load cell of 10 kg cross head speed equipped with a 0.5 in diameter.The depth of beef was 5 mm.

Optimisation using RSM of MINITAB
Software Version 15 The range of four selected factors namely pH of beef, immersion temperature, bromelain solution concentration and immersion time is as shown in Table 1.
Response surface regression analysis was performed to obtain a second-order polynomial equation or model.Statistical analysis of the model was represented in the form of Analysis of Variance (ANOVA).The MINITAB software Version 15 was also used to generate response contour and surface plots.

RESULTS AND DISCUSSION
Results in Table 3 show that the highest actual and predicted responses were 89.907% and 89.813%, respectively at factors whereby pH of beef was 5.6, immersion temperature at 60℃, bromelain solution concentration of 0.15% and immersion time of 10 minutes.The lowest actual and predicted responses were 58.267% and 58.608%, respectively at factors whereby pH of beef was 5.4, immersion temperature at 65℃, bromelain solution concentration of 0.20% and immersion time of 15 minutes.Where: X 1 = pH of beef, X 2 = immersion temperature (℃), X 3 = bromelain solution concentration (w/v%), X 4 = immersion time (min).Where: X 1 = pH of beef, X 2 = immersion temperature (℃), X 3 = bromelain solution concentration (w/v%), X 4 = immersion time (min), SE = standard error, t = student test, p = probability, R 2 = R -squared, R 2 (adj) = adjusted R -squared.
Response surface regression analysis was performed and results of estimated regression coefficients of second-order polynomial model for optimisation of beef tenderisation treated with bromelain are as shown in Table 4.
Based on Table 4, the second-order polynomial model equation for optimisation of beef tenderisation treated with bromelain is as given in equation 2: Where: X 1 = pH of beef, X 2 = immersion temperature (℃), X 3 = bromelain solution concentration (w/v%), X 4 = immersion time (min) The significant second-order polynomial model equation at the 5% level for the optimisation of beef tenderisation treated with bromelain is same as in equation (1).By referring to Table 4, it was found that linear factors such as pH of beef (X 1 ), immersion temperature (X 2 ) and immersion time (X 4 ) showed negative coefficients, respectively while bromelain solution concentration (X 3 ) showed positive coefficient.Square factors such as pH of beef (X 1 X 1 ), immersion temperature (X 2 X 2 ) and immersion time (X 4 X 4 ) showed positive coefficients, respectively while bromelain solution concentration (X 3 X 3 ) showed negative coefficient.Quadratic or interaction factors such as pH of beef and bromelain solution concentration (X 1 X 3 ), pH of beef and immersion time (X 1 X 4 ), immersion temperature and bromelain solution concentration (X 2 X 3 ), immersion temperature and immersion time (X 2 X 4 ) and bromelain solution concentration and immersion time (X 3 X 4 ) showed negative coefficients, respectively.
Student t and p tests were performed and it was found that linear effect factors namely X 1 , X 2 and X 4 showed negative t values, respectively while X 3 showed positive t value.p value for X 1 , X 2 , X 3 and X 4 were 0.000, respectively.Square effect factors namely X 1 X 1 , X 2 X 2 and X 4 X 4 showed positive t values, respectively while X 3 X 3 showed negative t value.p value for X 1 X 1 , X 2 X 2, X 3 X 3 and X 4 X 4 were 0.000, respectively.Quadratic or interaction effect factors namely X 1 X 3 , X 1 X 4 , X 2 X 3 , X 2 X 4 and X 3 X 4 showed negative t values, respectively and the respective p values were 0.000.It means that all linear, square and quadratic or interaction factors gave significant (p  0.05) effect on beef tenderisation.
Student t test was used to determine the significance of the estimated coefficient of the regression model equation (equation 1).The student t test value can be obtained by dividing each coefficient by its SE [18].p values were used as a tool to evaluate the significance and contribution of each factor and the statistical polynomial model equation [19].[20,21] reported that the larger the magnitude of the t value and the smaller the p value, the more significant is the corresponding coefficient.Based on those reported by [20,21], results of the present study showed that linear factors (X 1 , X 2 , X 3 , X 4 ), square factors (X 1 2 , X 2 2 , X 3 2 , X 4 2 ) and quadratic or interaction factors (X 1 X 2 , X 1 X 3 , X 1 X 4 , X 2 X 3 , X 2 X 4 , X 3 X 4 ) terms were highly significant.
The goodness of fit of the regression model was determined by determination coefficient R 2 which provides a measure of how much variability in the observed response values can be explained by the experimental factors and their interactions [22].Results in Table 4 showed that R 2 value was 99.97% which signified 99.97% of variability in the observed response values could be explained by the model while only 0.03% of variability in the observed response values cannot be explained by the model.The remaining R 2 value of 0.03% of the total variations would be due to other factors which were not included in the model.
The adjusted R 2 was a corrected value for R 2 after the elimination of unnecessary model terms.The adjusted R 2 would be remarkably smaller than the R 2 if there were many non-significant terms have been included in the model [23].In this study, it was found that the adjusted R 2 was high and very close to the R 2 and the respective values of adjusted R 2 and R 2 were 99.95% and 99.97%.The high adjusted R 2 value was attributed to the absence of non-significant terms in the model.The high adjusted R 2 and R 2 values thus, indicated a high dependence and correlation between the observed and predicted value responses.This is based on [24], who also found the high adjusted R 2 (98.81) and R 2 (97.96) values in their study on the optimisation of enzymatic detection of cadmium in aqueous solution using immobilized urease from vegetable waste.
ANOVA was performed to test for the significance and adequacy of the second-order polynomial model [25].The results are as summarised in Table 5.
The significance of regression was evaluated by f and p values using Fischer's and null-hypothesis tests.The f value predicts the quality of the entire model considering all design factors at a time.The p value is the probability of the factors having very little or insignificant effect on the response.Larger f value signifies better fit of the RSM model to the experimental data [26].[27] stated that f value with low p value indicates the high significance of the regression model.However, the p value should be lower than 0.05 for the model to be statistically significant [28].Based on those reported by [26][27][28], the regression model found in this study was highly significant as denoted by the large f and low p values with 4185.84 and 0.000, respectively.The linear, square and quadratic factors were also highly significant as denoted by the large f values of 1073.06,10142.46 and 1016.48,respectively and low p values of 0.000, respectively.
Lack of fit test was also performed.It describes the variation in the data around the fitted model [29].[28] testified that insignificant lack of fit indicates a good model.Insignificant lack of fit is desired as significant lack of fit indicates that there might be contributions in the regresses-response relationship that are not accounted for by the model.The f value for the lack of fit can be Where: DF = degree of freedom, Seq SS = sequential sum of square, Adj SS = adjusted sum of square, Adj MS = adjusted mean square, f = fischer, p = probability.
Openly accessible at http://www.scirp.org/journal/as/obtained by dividing the lack of fit mean square by its pure error mean square.Results of the lack of fit are shown in Table 5 and it was found that the f and p values for the lack of fit were 1.16 and 0.445, respectively.The insignificant p value thus indicates that the model was good and fitted well to the experimental data.
Response optimiser was performed and the results at optimum condition for target, maximum and minimum goals are shown in Figures 1, 2 and 3, respectively.
The feasibility of experiment for target, maximum and minimum goals was determined from the overlaid contour plot and the results are shown in Figures 4, 5 and 6, respectively.
Results of optimum conditions for different goals of actual and predicted responses and the feasibility of experiments obtained from response optimiser of MINI-TAB software Version 15 are shown in Table 6.
Based on Table 6, it was found that optimum conditions of target goal with pH of beef of 5.6, immersion temperature of 60℃, bromelain solution concentration of 0.1864% and immersion time of 10 minutes, and maximum goal with pH of beef of 5.6, immersion temperature of 60 ℃, bromelain solution concentration of 0.1682% and   Where: X 1 = pH of beef, X 2 = immersion temperature (℃), X 3 = bromelain solution concentration (w/v%), X 4 = immersion time (min), Y = actual response (%), FITS = predicted response (%), F = feasible, NF = not feasible.
immersion time of 10 minutes were feasible to be carried out.Meanwhile, optimum condition for minimum goal with pH of beef of 5.5, immersion temperature of 68℃, bromelain solution concentration of 0.25% and immersion time of 16 minutes was not feasible to be carried out.This is because according to the overlaid contour plots for target and maximum goals as shown in Figures 4 and  5, respectively, the optimum conditions of target and maximum goals located in white or feasible region.Meanwhile, the overlaid contour plot of minimum goal as shown in Figure 6 located in grey or not-feasible region.However, optimum condition from maximum goal was chosen because the target and FITS values was closer.Contour and surface plots for beef tenderisation treated with bromelain at feasible optimum condition are shown in Figures 7 and 8, respectively.
2D contour and 3D surface plots are the graphical representatives of the regression equation illustrating the function of two factors at a time while holding other factors at a fixed level [27].The 2D contour and 3D surface plots as demonstrated in Figures 7 and 8, respectively showed the effect of pH of beef, immersion temperature (℃), bromelain solution concentration (%) and immersion time (min) on beef tenderisation.The plots illustrating the values for pH of beef and immersion temperature while holding the values of bromelain solution concentration and immersion time at 0.1682% and 10 minutes, respectively.Where: X 1 = pH of beef, X 2 = immersion temperature (℃), X 3 = bromelain solution concentration (w/v%), X 4 = immersion time (min), V = verification value, P = predicted value.
Circular or elliptical shapes of contour plot indicate whether the reciprocal interactions between the factors are significant or not.Circular contour plot indicates the interactions between corresponding factors are negligible, while elliptical contour plot indicates the interactions between corresponding factors are significant [30] [20].Results of the present study showed that the contour plot was elliptical shape thus indicates significant interaction effect between pH of beef and immersion temperature on beef tenderisation.
The surface plot showed that the beef tenderisation increased at the lower and higher levels of pH of beef and immersion temperatures while at the middle level of pH of beef and immersion temperature, the beef tenderisation decreased.
Verification of beef tenderisation treated with bromelain at the feasible optimum condition was performed and the result is shown in Table 7.
The verification value of the beef tenderisation treated with bromelain at the feasible optimum condition was 89.571% which was very close to the predicted value with 89.813%.Since the difference between the verification and predicted values was less than 5%, therefore the feasible optimum condition of the beef tenderisation predicted by MINITAB Software Version 15 was acceptable.

CONCLUSIONS
The determination coefficient R 2 (99.97%) was high, thus the experimental data was acceptable.Optimum condition for the tenderisation of beef by bromelain using RSM had been determined.It was found that beef could be optimise tenderized 89.907% at the optimum condition pH of beef of 5, immersion temperature of 60℃, bromelain solution concentration of 0.1682% and immersion time of 10 minutes.It aws also found that the difference between the verification and predicted values was less than 5%, therefore, the optimum condition for the beef tenderisation predicted by MINITAB Software Version 15 could be accepted.

Figure 4 .Figure 5 .
Figure 4. Overlaid contour plot at optimum condition for target goal; pH of beef of 5.6, immersion temperature of 60˚C, bromelain solution concentration of 0.1864% and immersion time of 10 minutes.

Figure 7 .
Figure 7. Contour plot of beef tenderisation treated with bromelain at feasible optimum condition; pH 5.6, immersion temperature of 60℃, bromelain solution concentration of 0.1682% and immersion time of 10 minutes (holding value:bromelain solution concentration and immersion time fixed at 0.1682% and 10 minutes, respectively).

Figure 8 .
Figure 8. Surface plot of beef tenderisation treated with bromelain at feasible optimum condition; pH 5.6, immersion temperature of 60℃, bromelain solution concentration of 0.1682% and immersion time of 10 minutes (holding value : bromelain solution concentration and immersion time fixed at 0.1682% and 10 minutes, respectively).

Table 1 .
Coded and uncoded factors for the design of experiment.

Table 2 .
Experimental design recommended by MINITAB software Version 15.

Table 3 .
Factors and comparison between actual (Y) and predicted (FITS) responses.

Table 4 .
Estimated regression coefficients of second-order polynomial model for optimisation of beef tenderisation treated with bromelain.

Table 5 .
ANOVA for optimisation of beef tenderisation treated with bromelain.

Table 6 .
Comparison values of target and predicted responses for different optimum conditions and feasibilities of experiment.

Table 7 .
Comparison of verification and predicted values of beef tenderisation treated with bromelain at feasible optimum condition.