Carbon sources were screened for the production of PDC and biosynthesis of PAC. Cane molasses was the best carbon source as it produced 2.22 g/L PAC whereas other sources produced lesser amounts. That is pure cheese whey (whey without suspended proteins) 1.11, whole cheese whey (whey with suspended proteins) produced 0.88, glucose 0.99, galactose 0.44, fructose 0.66, maltose 0.77 and Carbon sulphite liquor 0.33 (Fig. 1). Researchers reported cane molasses a rich source of sugars, vitamins, minerals and a number of other nutrients and an excellent carbon source for growth of yeasts and production of different enzymes and other metabolites (Aguilar et al 2002; Sughra et al. 2013). Moreover this industrial waste is easily available round the year (Darvishi and Moghaddami, 2019).
3.1. Response Surface based optimization
Several researchers have used RSM for many novel research projects like synthesis of nanoparticles (Othman et al. 2017), Bioethanol (Darvishi and Moghaddami, 2019) and pretreatment of lignocellulosic substrates (Asadi and Zilouei, 2016) due to its robust action and sound statistical studies.
3.2. Plackett Burman Burman Model (PBM)
By using the Design-Expert version 10.1.6 (Stat-Ease, Inc., Minneapolis, MN, USA) Sugar Conc. (%, v/v), Incubation Time (Hrs) and Temperature of fermentation (oC) were selected as significant factors through PBM. Other factors were non-significant as R2 was greater than 1 and P value was less than 0.05. Linear regression analysis of eleven factors through four responses (Table 1A) was done using Plackett Burman Model. Standard error of design (Fig.2) was smaller and similar within all types of co-efficients. VIF was 1.0 and hence satisfactory. Ri2 for all factors was 0.00. All these aspects showed terms are not co-related and Model is significant.
As per ANOVA, P-values for initial sugar (%, v/v), Time (Hrs) and Temperature (o C) were less than 0.05 therefore these were significant factors. Central Composite Design was designed using selected factors to determine and optimize their co-relation for higher productions. Asif et al. (2012) designed experiments for proteases using the similar approaches.
3.3. Central Composite Designe and evaluation
Sugar Conc. (%, v/v), Incubation Time (Hrs) and Temperature of fermentation (oC) were used as input factors for CCD (Table 1B.) and studied through four responses as PDC activity (mmol/L) final pH, sugar consumed (%, v/v) and PAC (g/L) produced. The fitted models in terms of the coded values of Sugar Conc. (A), Time (B) and Temperature (C) are given below;
3.4. Evaluation of RSM Designe
Final Equations in Terms of Coded Factors:-
Y pH = 2.23 + 0.008 * A + -0.045* B + 0.010*C + -0.017*AB + -0.0169*AC+ -0.0*BC + 0.02*A2 + 0.0129*B2 + 0.034*C2 + -0.04*ABC + 0.04*A2B + -0.027*A2C + -0.0199*AB2 + 0*AC2 + 0*B2 C + 0*BC2 + 0*A3 + 0*B3 + 0*C3 + -0.03*A2B2 + 0*A2BC + 0* A2C2 + 0* AB2C + 0 *ABC2 + 0* B2 C2 + 0*A3B + 0*A3C + 0*AB3 + 0*AC3 + 0*B3C + 0*BC3 + 0*A4 + 0*B4 + 0*C4 + 0* A2B2C + 0 *A2 B C2 + 0*AB2C2 + 0*A3B2 + 0*A3BC + 0*A3C2 + 0*A2B3 + 0*A2C3 + 0*AB3C + 0*ABC3 + 0*B3C2 + 0*B2C3 + 0*A4B + 0*A4C + 0*AB4 + 0*AC4 + 0*B4C + 0*BC4 + 0*A5 + 0*B5 + 0*C5
Y sugar consumed (%, v/v) = + 1.53254 + 0.029824 * A + 0.0998773 * B + -0.0382367 * C + 0.0994692 * AB + 0.0381083 * AC + 0.00524738 * BC + 0.0684663 * A2 + -0.0860421 * B2 + 0.0465525 * C2
Y PAC(g/l) = + 7.24 + 0.070 * A + 0.62*B + -0.64 *C+ -0.027*AB + -0.027*AC+ 0.03 *BC + -1.95 *A2 + -1.20 *B2 + -1.95 * C2
where y is the PDC activity and PAC produced (g/L), positive sign in front of the terms indicates synergetic effect, whereas negative sign indicates antagonistic effect. RSM models for sugar consumed and final pH were not significant.
3.5. Analysis of variance (ANOVA)
F-values of Model for PDC activity and PAC produced (3.71 and 3.69), P-Values were less than 0.05 (0.0265 and 0.0269), P-Value for lack of fit was greater than 0.05 (0.283 and 0.2808) so the models for these responces were significant and lack of fit nonsignificant.There are only 2.65 and 2.69% chances that the F-Value of Model could occur due to noise (Table 2). Whereas for sugar consumed and final pH lack of fit was significant and model was non-significant.
In case of Pyruvate decarboxylase activity (Table 2) and PAC produced, co-efficients of quadratic terms (A2, B2 and C2) were significant model terms. F and P- value of Lack of Fit (0.9529and 0.283) implied that the Lack of Fit was not significant relative to the pure error. There was a little chance (28.30 and 28.0% for PDC activity and PAC models) that a "Lack of Fit" could occur due to noise. Morover, “Adeq Precision" measures the signal to noise ratio. Adeq Precision for PDC activity (4.888) and PAC(4.882) indicated an adequate signal. So, These model can be used to navigate the design. Models for Sugar and final pH were not significant. Model for final pH considered in the Fit Summary was aliased in this case B, C², ABC were significant model terms. So, the model for final pH cannot be accurately fit with design and therefore was not considered for analysis.
All possible interactions and co-relations of Time, sugar concentration and temperature were very important and plotted as 3D surface graphs. Effects of these interactions over PDC activity and PAC produced (g/L) were studied as shown in Fig. 4A and 5A. Linear plots (Fig.3A.) between standardized effects and Normal % age probability for PDC activity, PAC and sugar Consumed during fermentation showed that the resultant values were uniformly distributed along a linear trend line. Box-Cox plots (Fig.3B.) for PDC activity, PAC and sugar Conc. showed that Lameda for PDC activity and PAC produced were close to ideal values that is 1 for PAC and 0.048 for PDC, whereas for sugar consumed it was beyond the limit for good models. Shown by blue lines in Fig 3B.
Moreover, as per analysis of Ramps (3C. I, II, III), temperature significantly affected activity of PDC and yield of PAC. Whereas smooth ramps for sugar conc. and time showed that these factors have no significant effect on production of PDC and its products. Positive impact of temperature have been reported by many researchers (Shukla and kulkarni, 2002; Andrews and McLeish, 2012). Optimume temperature for PDC production from Pichia cecembensis was 38oC (Fig. 3C.I) which goes with Shukla and kulkarni, 2002.
Interaction of Temperature & sugar and Temperature & Time positively affected (4A) the PDC activity which rose to 42.8 U/ml (indicated by concentric red data points). In simple words it could be suggested that temperature was a key factor to enhance the activity of PDC. Interaction of Temperature with sugar concentration and incubation Time produced significant Response Surface Models and 3D surface graphs. Co-relation of Time and Sugar was not effective for higher PDC activity (14 U/ml, indicated by red data spots). Standard error for PDC for all co-relations was almost similar as shown in 3D plots in Fig. 4B.
Interactions of Temperature with sugar conc. and Time enhanced PAC production but interactions of Time and Sugar conc. could not produce good yields of PAC (g/L). Standard error of interactions was uniform for all interactions (Fig.5A, 5B). Hence, Temperature was the factor which can produce higher activity of PDC and its Products. Arroyo-López et al (2009) reported temperature and sugar as significant parameters, affecting the microbial growth and product formation using CCD of RSM. They maximize the yield by process optimization for five factors (initial pH, initial molasses concentration %, v/v, incubation temperature °C, mixing rate rpm, and incubation period (Hrs). In present study eleven factors through PBM and three factors through CCD were optimized to enhance the yield of PDC and PAC. According to the RSM optimization process, the response for each fermentation parameter was defined within the studied levels range to get the highest performance.
In present model, maximum PDC activity was 56.27 U/ml producing 8.44 g/L PAC. Whereas Saccharomyces cerevisiae gave a yield of 2.4 g/l (Doostmohammadi, 2016) using molasses as a substrate. The yield was increased by 71% under optimized fermentation conditions, initial pH of 5.0, total sugar concentration at 18% (v /v), Incubation temperature of 33°C and 13 hrs of incubation. Hussain et al., (2012) reported different strains of Saccharomyces cerevisiae producing 2.58g/l maximum PAC. Retention times (Table 3A) for PAC, Benzoic acid and Benzyl alcohol were 5.5-6.0, 17.5 and 1.5-2.0 min as determined through HPLC purification.
Whole cells of Pichia cecembensis (Table 3B) have better half-life at 4oC incubated with and without 40mM benzaldehyde (240 hrs and 336 hrs, respectively) as compared to Candida utilis reported by Satianegara et al. (2006) who reported half-life of 228 hrs in the presence of benzaldehyde at same temperature. Whereas crude extract exhibits extended half-lives at 25oC with and without Benzaldehyed (24 and 32.5 Hrs) rather than crude extracts of Candida utilis 12.9 and 26.3 hrs, respectively which goes in agreement with Leksawasdi, (2004). Partially purified PDC in current research work have better half life time (72hrs) in the presence of benzaldehyde as compared to partially purified PDC of Candida utilis which lost its activity by one half in the presence of benzaldehyde at 6oC within 60.5hrs.