A refined medium to enhance the antimicrobial activity of postbiotic produced by Lactiplantibacillus plantarum RS5

Postbiotic RS5, produced by Lactiplantibacillus plantarum RS5, has been identified as a promising alternative feed supplement for various livestock. This study aimed to lower the production cost by enhancing the antimicrobial activity of the postbiotic RS5 by improving the culture density of L. plantarum RS5 and reducing the cost of growth medium. A combination of conventional and statistical-based approaches (Fractional Factorial Design and Central Composite Design of Response Surface Methodology) was employed to develop a refined medium for the enhancement of the antimicrobial activity of postbiotic RS5. A refined medium containing 20 g/L of glucose, 27.84 g/L of yeast extract, 5.75 g/L of sodium acetate, 1.12 g/L of Tween 80 and 0.05 g/L of manganese sulphate enhanced the antimicrobial activity of postbiotic RS5 by 108%. The cost of the production medium was reduced by 85% as compared to the commercially available de Man, Rogosa and Sharpe medium that is typically used for Lactobacillus cultivation. Hence, the refined medium has made the postbiotic RS5 more feasible and cost-effective to be adopted as a feed supplement for various livestock industries.


Results
Conventional refinement of antimicrobial activity of postbiotic RS5. Optimum concentration of yeast extract nitrogen source. The total nitrogen content (N) of KAT yeast extract was previously determined using Kjeltec 2400 (Foss, UK) equipment. Figure 1a Figure 1b shows the effect of different concentrations of yeast extract concentrations on the growth of L. plantarum RS5. As for the 27.84 g/L of yeast extract, the maximum biomass of L. plantarum RS5 (3.10 g/L) was achieved at 32 h of incubations. The biomass was increased to 3.26 g/L (32 h) when the yeast extract concentration was increased to 36.20 g/L. The significantly highest (p < 0.05) biomass of 3.41 g/L (24 h) was attained in the growth medium containing 44.55 g/L of yeast extract. However, significantly lower (p < 0.05) antimicrobial activity was detected in the growth medium containing 44.55 g/L of yeast extract, in comparison to the growth medium containing 27.84 g/L and 36.20 g/L of yeast extract, respectively. Table 1 shows the kinetic parameters of L. plantarum RS5 when cultivated in growth medium containing different concentrations of yeast extract. The highest productivity of antimicrobial activity of 37.33 MAU/mL h was achieved with 27.84 g/L of yeast extract as a comparison to the MRS medium that contributed to the lowest productivity of 26.67 MAU/mL h. Furthermore, the highest specific growth rate (µ) of L. plantarum RS5 (0.24 h −1 ) was achieved with 27.84 g/L of yeast extract, indicating that the 27.84 g/L (200.18 mM N) was the optimum concentration of yeast extract for the growth of L. plantarum RS5 to produce the highest antimicrobial activity in postbiotic RS5.
Effect of different carbon sources. Lactiplantibacillus plantarum RS5 produced different levels of antimicrobial activities when cultivated in growth medium containing different carbon sources. The production of antimicrobial activity was initiated after 10 h of incubation for all the tested carbon sources. The profile of antimicrobial activities (Fig. 2a) showed that significantly higher (p < 0.05) antimicrobial activities were achieved with the growth medium containing glucose or lactose as the carbon source in comparison to fructose and sucrose as carbon source throughout the incubation period. The highest antimicrobial activity of 1120 MAU/mL was observed www.nature.com/scientificreports/ from 24 to 30 h of incubation when lactose or glucose was used as the carbon source. The highest antimicrobial activity achieved using sucrose was only 800 MAU/mL, which was 29% lower than the highest antimicrobial activity that was achieved with glucose and lactose. However, the 800 MAU/mL of antimicrobial activity was 20% higher than the antimicrobial activity detected when MRS medium was used. The exponential growth of L. plantarum RS5 in growth medium containing sucrose was noted between 8 and 16 h, which was similar to the growth profile that observed in growth medium containing glucose as carbon source (Fig. 2b). Comparable biomass was detected when L. plantarum RS5 was grown in medium containing glucose or sucrose as carbon source until 24 h of incubation. The highest biomass (3.03 g/L) of L. plantarum RS5 was achieved in growth medium containing glucose at 30 h of incubation. Surprisingly, the lowest biomass was attained when L. plantarum RS5 was grown in medium containing lactose as a carbon source, followed by fructose and MRS medium.
The kinetic analysis (Table 2) showed that the highest specific growth rate of L. plantarum RS5 was achieved with glucose (0.32 h −1 ), followed by sucrose (0.24 h −1 ). Although L. plantarum RS5 produced the highest antimicrobial activity in lactose containing growth medium, the lowest specific growth rate (0.06 h −1 ) and biomass (1.73 g/L) were detected when lactose was supplemented in the growth medium as carbon source. The same level of the highest antimicrobial activity was produced at 26-30 h of incubation when L. plantarum RS5 was grown in medium containing either glucose or lactose as carbon source. In addition, the highest biomass of L. plantarum RS5 was also achieved in growth medium containing glucose. Thus, glucose was selected as the best carbon source for the subsequent experiments, since the cost of glucose was lower in comparison to lactose to produce L. plantarum RS5 biomass and the antimicrobial activity of postbiotic RS5.
Effect of glucose concentrations. Different levels of antimicrobial activities (Fig. 3a) were observed when L. plantarum RS5 was cultivated in growth medium containing different glucose concentrations. No antimicrobial activity was detected when glucose was not supplemented in the growth medium. An obvious increment of antimicrobial activity was observed when L. plantarum RS5 was cultivated in growth medium containing 20 g/L of glucose, whereby 1013.33 MAU/mL to 1120 MAU/mL of antimicrobial activity was achieved between 20 and 32 h of incubation. Surprisingly, a twofold reduction of antimicrobial activity (640-800 MAU/mL) was detected at 18-30 h of incubation, when the glucose concentration was increased to 40 g/L. Similarly, the antimicrobial activity of postbiotic RS5 was significantly lower (p < 0.05) when glucose was supplemented at a concentration that lower than 20 g/L, with 30 MAU/mL at 14 h of incubation, 76.67 MAU/mL at 22 h of incubation and 693.33 MAU/mL at 24 h was produced when 5 g/L, 10 g/L and 15 g/L of glucose was used, respectively.
The growth profile (Fig. 3b) showed that L. plantarum RS5 produced different levels of biomass when cultivated in a growth medium containing different glucose concentrations. Higher glucose concentrations (15 g/L, 20 g/L and 40 g/L) promoted higher biomass formation throughout the cultivation period. Comparable biomass was attained when L. plantarum RS5 was cultivated in the growth medium containing 20 or 40 g/L of glucose at 28-32 h of incubation. However, the highest biomass was achieved at 22 h of incubation when L. plantarum RS5 was grown in a growth medium containing 40 g/l glucose.
Similarly, different specific growth rates were detected when of L. plantarum RS5 was cultivated in growth medium containing different levels of glucose (Table 3). The highest specific growth rate of L. plantarum RS5 was achieved with growth medium containing 20 g/L of glucose.
Statistical refinement of antimicrobial activity of postbiotic RS5. Selection of significant ionic minerals and non-ionic surfactant by FFD. Table 4 shows the experimental design, antimicrobial activities, biomass production, lactate production and the corresponding pH of postbiotic RS5 produced by L. plantarum RS5. The growth media comprised different combinations of ionic minerals and a non-ionic surfactant in addition to 20 g/L of glucose and 27.84 g/L of yeast extract as carbon and nitrogen source. The antimicrobial activities and biomass varied markedly from 0 to 1280 MAU/mL and 0.94 g/L to 3.35 g/L, respectively. The results showed that Table 1. Effects of different concentrations of yeast extract on the growth and antimicrobial activity of postbiotic RS5 produced by L. plantarum RS5. *P max (MAU/mL)-maximum antimicrobial activity, t (h)cultivation time of maximum antimicrobial activity, Pr (MAU/mL h)-productivity of antimicrobial activity, X max (g/L)-maximum biomass, µ (h −1 )-specific growth rate, t x (h)-cultivation time of maximum biomass. **Values for P max (MAU/mL) and X max (g/L) are mean ± standard error of the mean (SEM), n = 3. Mean ± SEM within the same row that does not share a similar superscript are significantly different (p < 0.05). www.nature.com/scientificreports/ the highest antimicrobial activity (1280 MAU/mL) was achieved together with the highest biomass (3.35 g/L) and lactate (7.07 g/L) production (run 7). The interaction effects of media component on antimicrobial activity were evaluated by ANOVA (Table 5). From the ANOVA table, the linear effect of Tween 80 (X4) and manganese sulphate (X6), and the interaction between sodium acetate (X1) and Tween 80 (X4), and between Tween 80 (X4) and manganese sulphate (X6) affected the antimicrobial activity strongly at a p-value less than 0.05. However, the other mineral components did not affect the antimicrobial activity significantly (p > 0.05). The following equation 1 is the model of the desired response (Y MAU/mL ) and regression coefficients:

Kinetic parameters* Control MRS medium
whereby Y = antimicrobial activity; X 4 = tween 80; X 6 = manganese sulphate tetrahydrate; X 1 X 4 = interaction of sodium acetate and tween 80; X 4 X 6 = interaction of tween 80 and manganese sulphate tetrahydrate.  www.nature.com/scientificreports/ The R2 coefficient of the FFD model was 0.95, inferring that the model explained 95% of the data variability. The statistical significance of the model was confirmed by F-test. The F-value of 33.99 suggested that the model was significant and there was only a 0.01% chance that a model F-value this large could occur due to noise. Adequate precision measures the signal to noise ratio and generally a ration greater than 4 is desirable. The precision ratio of 17.36 indicated an adequate signal and this model could be used to navigate the design space.

Concentration refinement of significant ionic minerals and non-ionic surfactant by CCD. CCD
of RSM further refined the concentration of ionic minerals (sodium acetate and manganese sulphate) and nonionic surfactant (Tween 80) identified by FFD. The concentrations tested for sodium acetate, Tween 80 and manganese sulphate were 0-10 g/L, 0-2 g/L and 0-0.08 g/L, respectively. The responses, together with the coded and actual values of selected variables are presented in Table 6. The highest experimental value for antimicrobial activity was 1280 MAU/mL. The biomass of L. plantarum RS5 was in the range of 1.95 g/L to 3.51 g/L. Interestingly, the observation of CCD was also noted in FFD, whereby significantly higher lactate and lower pH were detected simultaneously with higher antimicrobial activity and biomass, indicating both antimicrobial activity and lactate production by L. plantarum RS5 were growth associated.
By applying multiple regression analyses to the experimental data, the following quadratic equation 2 explains well the antimicrobial activity produced by L. plantarum RS5 under the effects of sodium acetate, Tween 80 and manganese sulphate: whereby Y = antimicrobial activity; X 1 = sodium acetate; X 2 = tween 80; X 3 = manganese sulphate tetrahydrate.
The statistical significance of the quadratic equation was further confirmed by F-test. The ANOVA for the response surface quadratic model was shown in Table 7. The F-value, of 471.97 and the probability value at P > F was < 0.0001, implying the model was highly significant. There was just 0.01% chance that a "Model F-value" this large could occur because of noise. The coefficient that estimated the quadratic equation, along with the corresponding P values was indicated in the ANOVA analyses. The linear coefficients (X 1 , X 2 ), all quadratic coefficients, and two interaction coefficients, i.e. X 1 X 3 and X 2 X 3 , were highly significant (p < 0.01). The coefficient of R 2 value evaluated the goodness of the model. The R 2 value closed to 1.0 reflected the model's strength to predict the response more effectively. Meanwhile, the adequate precision of 54.77 showed that this polynomial quadratic model had an adequate signal and could be used to navigate the design space.
The three-dimensional response surface plots of Fig. 4 illustrate the interaction effects of sodium acetate, Tween 80 and manganese sulphate on the antimicrobial activity of postbiotic RS5. Higher antimicrobial activity was achieved when sodium acetate, Tween 80 and manganese sulphate were supplemented in growth medium ( Table 6). The optimum values of sodium acetate, Tween 80 and manganese sulphate were determined from the response surface plots. The quadratic model predicted the antimicrobial activity of 1355 MAU/mL could be achieved by using the CCD refined medium. Subsequently, the predicted antimicrobial activity was validated through experiment and 1333.33 MAU/mL was attained, which was comparable to the predicted antimicrobial activity of 1355 MAU/mL. A biomass of 3.48 g/L was achieved when L. plantarum RS5 was grown in CCD refined medium. The CCD refined medium consisted of glucose 20 g/L, yeast extract 27.84 g/L, sodium acetate 5.75 g/L, Tween 80 1.12 g/L, and manganese sulphate tetrahydrate 0.05 g/L. Tables 1 and 2 of supplementary results show the cost of the CCD refined growth medium reduced by 85% as compared to the control MRS medium.

Discussion
Determination of antimicrobial activity. Our previous studies have demonstrated that the postbiotic RS5 exhibited broad inhibitory activity against P. acidilactici, L. monocytogenes, S. enterica, E. coli and Vancomycin resistant enterococci 18,19 . LAB species that are tolerant to acidic condition have been commonly employed to determine the antimicrobial activity attributed to bacteriocin molecules that are present in crude postbiotic Table 2. Effects of carbon sources on the growth and antimicrobial activity of postbiotic RS5 produced by L. plantarum RS5. *P max (MAU/mL)-maximum antimicrobial activity, t (h)-cultivation time of maximum antimicrobial activity, Pr (MAU/mL h)-productivity of antimicrobial activity, X max (g/L)-maximum biomass, µ (h −1 )-specific growth rate, t x (h)-cultivation time of maximum biomass. **Values for P max (MAU/mL) and X max (g/L) are mean ± standard error of the mean (SEM), n = 3. Mean ± SEM within the same row that does not share a similar superscript are significantly different (p < 0.05). www.nature.com/scientificreports/ preparations. Therefore, P. acidilactici ATCC 4-46, a LAB that is tolerant to acidic condition, was employed as the indicator microorganism for the determination of antimicrobial activity in this study. The antimicrobial activity was expressed as modified arbitrary unit (MAU/mL) by considering the diameter of inhibition zone as described by Lim 5 to determine the antimicrobial activity quantitatively.

Kinetic parameters* Control MRS medium
Optimum concentration of yeast extract nitrogen source. Previous research findings have reported that medium composition can have a pronounced effect on the production of bacteriocin and the growth of LAB producer cells [28][29][30][31][32] . Therefore, we adopted a combination of conventional and statistical-based approaches to develop a refined medium to enhance the antimicrobial activity of postbiotic RS5. Yeast extract has been reported to have a strong effect on bacteriocin production by Lactiplantibacillus sp. 30,34,46 . Higher bacteriocin production could be due to inactivation of a bacteriocin synthesis inhibitor 39,40 . Moreover, yeast extract contains more growth factors and a relatively more free amino acids and short peptides than many other nutrient sources. These compounds could stimulate bacteriocin production by LAB as compared to other protein hydrolysates 30,34 . www.nature.com/scientificreports/ Table 3. Effect of glucose concentrations on the growth and antimicrobial activity of postbiotic RS5 produced by L. plantarum RS5. *P max (MAU/mL)-maximum antimicrobial activity, t (h)-cultivation time of maximum antimicrobial activity, Pr (MAU/mL h)-productivity of antimicrobial activity, X max (g/L) -maximum biomass, µ (h −1 )-specific growth rate, t x (h)-cultivation time of maximum biomass. **Values for P max (MAU/ mL) and X max (g/L) are mean ± standard error of the mean (SEM), n = 3. Mean ± SEM within the same row that does not share a similar superscript are significantly different (p < 0.05). www.nature.com/scientificreports/ www.nature.com/scientificreports/ The results obtained in this study also indicated that the antimicrobial activity of postbiotic RS5 was significantly affected by the concentration of yeast extract supplemented in the growth medium. The biphasic exponential production of antimicrobial activity (Fig. 1a) was observed in growth media containing 27.84 g/L (200.18 mM N) and 36.20 g/L (260.28 mM N) of yeast extract, respectively. Moghadam et al. 47 demonstrated that L. plantarum RS5 harboured both plnEF and plnW structural genes, which encode Plantaricin W lantibiotic (class 1) bacteriocin and Plantaricin EF non-lantibiotic (class 2) bacteriocin molecules 48 . The biphasic production of antimicrobial activity could be attributed to the production of Plantaricin W and Plantaricin EF by L. plantarum RS5. However, further study should be performed to confirm the type of bacteriocins that could account for the biphasic antimicrobial activities produced in postbiotic RS5 under these conditions.
The antimicrobial activity was enhanced 43% when the yeast extract concentration was increased from 11.89 to 36.20 g/L. Todorov et al. 49 reported a twofold increase in bacteriocin activity (3200 AU/mL) produced by L. sakei, when 20 g/L of yeast extract was supplemented. Similarly, the production of bacteriocin-like activity by Enterococcus durans E204 was increased from 320 to 640 AU/mL, when the yeast extract concentration was increased from 0.5 to 2.0% 50 . However, a twofold reduction of antimicrobial activity of postbiotic RS5 was noted when yeast extract was increased to 44.55 g/L, indicating that higher concentration of yeast extract could have induced substrate inhibition on bacteriocin synthesis when excessive nitrogen source was supplemented in the growth medium. The substrate inhibition observation was also reported for bacteriocin activity of L. rhamnosus 50 and L. brevis DF01 34 , respectively.
Lactiplantibacillus plantarum RS5 produced different levels of antimicrobial activities and biomass when different carbon sources were supplemented in the growth medium, demonstrating that it has different preference and ability to efficiently utilise different carbon sources. This could be related to a wide range of hydrolytic enzymes required for the metabolism of different carbohydrates. Carbohydrate catabolism provides energy to LAB via substrate-level phosphorylation for cell growth and metabolite production 52 . The extracellular hydrolytic enzyme activities have been reported recently for various LAB isolated from Malaysians foods, including L. plantarum RS5 43,44 . The diverse effects of different carbon sources on bacteriocin production by LAB have been reported previously 34,39,[49][50][51][52] . The positive effect of glucose was reported for bacteriocin produced by L. brevis DF01 34 , L. plantarum I-UL4 46 and Streptococcus bovis 52 . The results obtained in this study agreed with the observation reported by Todorov and Dicks 49 . Bacteriocin production was reduced by 50% and 87.5%, when L. pentosus was grown in medium containing sucrose and fructose, respectively. Xylose was reported to support bacteriocin production by Lactococcus lactis 39 . In contrast, fructose was a better carbon source for the bacteriocin production by L. mesenteroides E131 as compared to glucose 53 , suggesting that the carbon source requirement by LAB was varied and strain specific. Thus, it is important to test different carbon sources when aiming to enhance the antimicrobial activity produced by LAB.
The antimicrobial activities of postbiotic RS5 produced between 16 and 20 h of incubation using glucosecontaining growth medium was significantly higher (p < 0.05) compared to when lactose was used as carbon source (Fig. 2a). Between 26 and 30 h of incubation, antimicrobial activity of 1120 MAU/ml was achieved by both lactose and glucose-containing growth medium. The delay in the achievement of the highest antimicrobial activity of 1120 MAU/ml when lactose was used as the sole carbon source could be due to the disaccharide nature of lactose as compared to the simple sugar of glucose. Glucose could be catabolised directly via the glycolytic pathway to supply energy for bacteriocin production. The disaccharide lactose has to be taken up by a specific permease to be hydrolysed by β-galactosidase to produce glucose and d-galactose monomers, which will be subsequently converted to glucose-6-phosphate via the Leloir pathway prior to catabolism by the glycolytic pathway.
The highest biomass of L. plantarum RS5 (3.03 g/L) was achieved with glucose-containing growth medium. Polak-Berecka et al. 54 reported that the maximal biomass (5.50 g/L) of L. rhamnosus was achieved when it was Table 7. ANOVA analyses for CCD models of ionic minerals and non-ionic surfactant on the antimicrobial activity of postbiotic RS5 produced by L. plantarum RS5. *X 1 : sodium acetate; X 2 : Tween 80; X 3 : manganese sulphate tetrahydrate. **Value of "Prob > F" less than 0.05 indicates that the model terms are significant. www.nature.com/scientificreports/ cultivated in modified MRS medium supplemented with 13.40 g/L of glucose. Although L. plantarum RS5 produced comparable antimicrobial activity in lactose containing medium, the lowest specific growth rate (0.06 h −1 ) and biomass formation (1.73 g/L) were observed when lactose was supplemented as carbon source. A similar observation was also reported by Sanchez et al. 27 for nisin Z production by Lactococcus lactis subsp. lactis. A higher amount of nisin Z and a lower amount of biomass were produced by L. lactis subsp. lactis when it was grown in a lactose-containing medium. The lactose could have induced the transcription of the nisZ structural gene 27 . The effect of glucose concentration was also reported for bacteriocin production by different LAB 50 . Bacteriocin-like activity was detected at 160 AU/mL when E. duran E204 was cultured in a medium supplemented with 4.0% (w/v) glucose, which was 75% lower than the activity that was achieved with 3.0% (w/v) glucose. Similarly, bacteriocin production by L. plantarum LL441 was also significantly reduced when it was cultivated in the medium containing more than 3% (w/v) carbohydrates 55 . Homofermentative Lactobacilli have been reported www.nature.com/scientificreports/ to possess a phosphoenolpyruvate-sugar phosphotransferase system 56 . High glucose concentration has been shown to inhibit sugar uptake via phosphorylation of the heat-stable protein of the phosphoenolpyruvate-sugar phosphotransferase system 55 . The increased level of phosphorylated protein would eventually lead to a reduction in glucose intake and switching off the catabolite-sensitive operons 55 . Nonetheless, the synthesis of antimicrobial substances would be advantageous when the initial concentration of nutrient at an optimal level becomes limiting during the fermentation process. Under the limiting conditions, the bacteriocin would antagonise the potential competitors for exhausting nutrients such as glucose and yeast extract. This concept was proposed for antibiotic production 57 . Both the highest antimicrobial activity and biomass were achieved in growth medium containing 20 g/L of glucose, inferring that the production of antimicrobial compounds by L. plantarum RS5 directly correlated to the growth, which agreed with the study reported by Trinetta et al. 58 . Moreover, the production of pediocin AcH 59 and sakacin P 60 by LAB were stimulated when the medium and cultivation condition favored the growth of bacteriocin producer cells. However, there are also some cases in which higher bacteriocin activity were achieved under sub-optimal growth conditions 61,62 . The optimal concentration of glucose (20 g/L) and yeast extract (27.84 g/L) contributed to an optimal C:N ratio of 3.33, which enhanced the antimicrobial activity of postbiotic RS5 significantly (p < 0.05). The optimal C:N ratio improved bacteriocin production by Bifidobacterium thuringiensis 63 and L. curvatus L442 64 . A more than fourfold increase of bacteriocin production was reported when B. thuringiensis was cultured in medium with a C:N ratio of 9 63 . Similarly, nitrogen has been shown to be a limiting factor for bacteriocin synthesis by L. curvatus L442, and the optimal ratio of nitrogen and carbon sources has been demonstrated to substantially enhance bacteriocin activity 64 .

Statistical refinement of ionic minerals and non-ionic surfactant.
Tween 80 has been shown to be a crucial factor for bacteriocin production 30,45 . The inclusion of Tween 80 in the growth medium improved the production of bacteriocin ST194BZ by more than 50% 65 . Tween 80 plays a significant role as non-ionic surfactant that facilitates bacteriocin secretion by altering membrane fluidity 30 . The ionic minerals are essential components for biological production processes 66,67 . Different effects of manganese sulphate were reported for bacteriocin production by different LAB isolates. The supplementation of manganese sulphate in the growth medium improved bacteriocin production by L. pentosus 30 and L. plantarum 55 . However, manganese sulphate exerted no effect on bacteriocin production by L. brevis OG1 68 and an inhibitory effect on salivaricin production by L. salivarius CRL 1328 66 . The inhibitory effect of manganese sulphate on salivaricin production could be attributed to the growth stimulation of L. salivarius, whereby the available nutrients were used for biomass production. In the present study, manganese sulphate had a significant stimulatory effect on the antimicrobial activity of postbiotic RS5. In contrast, the other ionic minerals: Na + , K + and Mg 2+ did not show any significant linear effect on antimicrobial activity of postbiotic RS5, which was contradictory to the results reported for micrococcin production, whereby significant effects of Mg 2+ and K + were noted for micrococcin production by Micrococcus sp. Moreover, Micrococcus sp. produced a 32-fold increase of micrococcin activity, when 2.5% K 2 HP0 4 and 0.5% MgS0 4 .7H 2 0 were supplemented in the optimised medium, as compared to the control MRS medium 67 . However, MgSO 4 has been demonstrated to exert a positive effect and K 2 HPO 4 exerted an inhibitory effect on the bacteriocin production by L. salivarius 66 .
Citrate and phosphate ions did not enhance the antimicrobial activity of postbiotic RS5. This observation has further supported the notion that LAB, like L. plantarum, preferred an acidic environment to produce defensive molecules 57,69 . Phosphate control is a well-known regulatory mechanism to produce an inhibitory substance 70,71 . However, not all inhibitory substances were favourably produced under the presence of phosphate. The beneficial effect of phosphate was reported for nisin production by L. lactis strain 39,72 . But the negative effect of phosphate was demonstrated for both antimicrobial activities of postbiotic RS5 and salivaricin production 66 . The negative effect of phosphates could be due to a production inhibition mechanism, release or action of synthetases, inducers, precursors and other molecules, which were involved in the synthesis of inhibitory substance 70,71 .
The linear effect of Tween 80 (X4) and manganese sulphate (X6), the interaction effect between sodium acetate (X1) and Tween 80 (X4), and between Tween 80 (X4) and manganese sulphate (X6) strongly influenced the production of antimicrobial activity by L. plantarum RS5. The positive interaction between manganese sulphate and sodium acetate on antimicrobial activity of postbiotic RS5 agreed with a previous report, whereby salivaricin production reached the highest level when Mn 2+ and Na + were simultaneously added to the growth medium 64 . However, the presence of acetate, sulphate, citrate, and phosphate anions in growth medium exerted different effects on the antimicrobial activity of postbiotic RS5, whereby significant positive effects of acetate and sulphate were observed in this study. Similarly, the inclusion of acetate ions in the growth medium has been reported to increase the bacteriocin production by L. salivarius, attributed to the buffering capability of acetate ions to maintain the pH of the growth medium below 7 66 . In the current study, the concurrent addition of acetate and sulphate ions (0.05 g/L) in the growth medium increased the antimicrobial activity of postbiotic RS5, suggesting that the production of antimicrobial activity was favourable when mild acidic condition was maintained during the growth of L. plantarum RS5.
The lowest pH of 3.66 was noted in postbiotic RS5, demonstrating antimicrobial activity of 1120 MAU/mL (Run 19), did not correspond to the highest antimicrobial activity of 1280 MAU/mL and the highest concentration of lactate of 7.07 g/L of Run 3. Nonetheless, significantly higher (p < 0.05) biomass of L. plantarum RS5 (3.46-3.51 g/L and) and lactate concentration (6.90-7.07 g/L) were observed in postbiotic RS5 exhibiting the highest antimicrobial activity of 1280 MAU/mL, implying that the antimicrobial activity and lactate production by L. plantarum RS5 were growth associated. A similar observation was also reported by Chaline et al. 73 , whereby maximum lactic acid production was found in the growth medium that produced the maximum biomass of L. plantarum BL011. www.nature.com/scientificreports/

Conclusions
The combination of conventional and statistical-based optimisation techniques (FFD and CCD of RSM) was demonstrated to be a useful tool to develop a refined growth medium to significantly enhance the antimicrobial activity of postbiotic RS5. The defined growth medium that consisted of 20 g/L glucose, 27.84 g/L yeast extract, 5.75 g/L sodium acetate, 1.12 g/L Tween 80 and 0.05 g/L manganese sulphate exerted a significant impact on L. plantarum RS5 to produce an enhanced antimicrobial activity of 1333.33 MAU/mL. The antimicrobial activity was enhanced 108% compared to 640 MAU/mL antimicrobial activity that was attained using control MRS medium. The highest biomass concentration of 3.48 g/L was also obtained when L. plantarum RS5 was cultivated in the refined growth medium. The cost of the refined growth medium was substantially reduced, by 85%, compared to the control MRS medium. The results obtained in the current study have shown that significant enhancement of the antimicrobial activity of postbiotic RS5 was successfully achieved by refining the growth medium composition via a combination of conventional and statistical-based approaches. The improvements of growth medium have substantially lowered the production cost of the postbiotic RS5 and make it more feasible for the postbiotic RS5 to be used as a feed supplement in various animal production industries.

Materials and methods
Microorganisms. Lactiplantibacillus plantarum RS5 was used to produce postbiotic RS5. P. acidilactici ATCC 4-46 was employed as an indicator microorganism for the determination of antimicrobial activity 5  Preparation of postbiotic RS5. The postbiotic produced by L. plantarum RS5 was designated as postbiotic RS5 in this study. The preparation of postbiotic RS5 was performed according to the method of Ooi et al. 46 .
The postbiotic RS5 was collected by centrifugation (10,000×g for 15 min at 4 °C) and kept at 4 °C for the determination of antimicrobial activity 5,75 .
Determination of antimicrobial activity. The antimicrobial activity of postbiotic RS5 was determined using a modified Agar Well Diffusion Assay 5,75 , without neutralising the pH of the postbiotic RS5. A twofold serial dilution (2 0 to 2 -5 ) was conducted to dilute postbiotic RS5 using sterile 0.85% (w/v) NaCl solution. Diluted postbiotic RS5 (20 µL) was then allowed to diffuse around a well (5.0 mm in diameter) pre-punched in MRS agar medium, followed by overlaying with 3 mL of soft agar inoculated with 1% (v/v) P. acidilactici 4-46 (OD 600nm was adjusted to 1.0) before incubating at 30 °C for 24 h. A clear inhibition zone with a diameter of more than 1 cm (including 0.5 cm diameter of the well) was considered a positive antimicrobial activity. The antimicrobial activity was expressed as modified arbitrary unit (MAU/mL) using the following equation 3 3 : Determination of microbial biomass. Microbial biomass was determined according to the method described by Li and de Orduna 76 . A volume of 5 mL of the cultured broth was filtered through a pre-dried and pre-weighed 0.45 µm cellulose acetate filter paper (Sartorius Stedim Biotech, Germany). The filter paper was then dried overnight at 105 °C. The dried filter paper was then transferred to a desiccator for 4 h to allow the filter paper to cool down to room temperature. The difference between the final and initial weight of filter paper was taken as the weight of microbial biomass in the 5 mL of cultured broth. The microbial biomass was expressed as gram of dry weight per litre (g/L).

Determination of lactate concentration and pH.
Lactate concentration of postbiotic RS5 was determined using the Pico Trace Glucose Analyser (Trace Analytics GmbH, Germany) according to the manufacturer's instruction and the pH of the postbiotic RS5 was determined using a pH meter (Mettler Toledo, US).
Conventional refinement of antimicrobial activity of postbiotic RS5. The composition of MRS (Merck, Darmstadt, Germany) broth medium (Table 8) was used as a reference to refine the growth medium of L. plantarum RS5. The concentration of glucose and yeast extract were optimised by using a conventional approach 30,52 , whereas the ionic minerals and Tween-80 were optimised by using statistical approaches 34,35 . A volume of 1% (v/v) active inoculum (OD 600 nm = 1.0) was inoculated into 15 mL of growth medium and incubated at 30 °C under static condition. Each optimisation experiment was performed in triplicate.
Optimum concentration of yeast extract nitrogen source. Optimum concentration of carbon source. The effects of glucose, lactose, fructose and sucrose (Merck, Darmstadt, Germany) as carbon sources were determined at 20 g/L with the optimum concentration (g/L) of yeast extract (verified in the previous experiment), sodium acetate at 5.0, diammonium hydrogen citrate at 2.0, dipotassium hydrogen phosphate at 2.0, Tween 80 at 1.0, magnesium sulphate heptahydrate at 0.2 and manganese sulphate tetrahydrate at 0.04, which followed the composition of MRS broth (Table 8; Merck, Darmstadt, Germany). Subsequently, the optimum concentration of the best carbon source was then determined at 0, 5, 10, 15, 20 and 40 (g/L). The experiment was performed in triplicate with 2 h sampling intervals.
Statistical refinement of antimicrobial activity of postbiotic RS5. Statistical refinement of growth medium for the enhancement of the antimicrobial activity of postbiotic RS5 was performed using Design Expert statistical software version 6.0.10 (State-Ease Inc, Minneapolis, MN). The optimum concentration of carbon source and yeast extract (as described in previous sections) was used for both FFD and CCD experiments. Both FFD and CCD experiments were performed in triplicate, and sampling was performed at 36th hour of incubation.
Selection of significant ionic minerals and non-ionic surfactant by FFD. FFD was employed to determine the effect of ionic minerals (sodium acetate, diammonium hydrogen citrate, dipotassium hydrogen phosphate, magnesium sulphate heptahydrate, manganese sulphate tetrahydrate) and non-ionic surfactant, Tween 80, on the antimicrobial activity of postbiotic RS5. The effect of ionic minerals and non-ionic surfactant were examined at three different levels with a minimum (− 1), central (0) and maximum (+ 1) value of FFD, as shown in Table 4. FFD suggested 32 experimental runs. The response effect was analysed using analysis of variance (ANOVA), and the significance level of each component was determined at p-value < 0.05.
Concentration refinement of significant ionic minerals and non-ionic surfactant by CCD. The optimum concentration and the interaction effects of the three most significant ionic minerals and non-ionic surfactant (sodium acetate, Tween 80 and manganese sulphate) identified by FFD were then further refined using CCD. Each variable was evaluated at five levels (− α, − 1, 0, + 1, + α) as shown in Table 6. CCD suggested 20 experimental runs and made up to 2 3 factorial design with eight cube points, augmented with six replications of the centre point (all variables at level 0) and the six-star points, i.e., points having one variable at an axial distance to the centre of ± α. The α value was 1.68 as calculated by the following equation 4: The relationship among the variables was determined by fitting the second-order polynomial equation 5 for antimicrobial activity as the response obtained from the 20 experimental runs.
whereby y is the predicted response variable, b 0 is the intercept, b 1 , b 2, b 3 are the linear coefficients, b 11 , b 22 and b 33 are the squared coefficients, b 12 , b 13 and b 23 are the interaction coefficients, x 1 , is the sodium acetate, x 2 is the Tween 80, and x 3 is the manganese sulphate. ANOVA was used to obtain the coefficients of the equation. The analyses were performed for the responses obtained from the 20 experimental runs.