Introduction

Probiotics are viable microorganisms when administered in adequate amounts to confer health benefits to the host1. Probiotics represent a wide group of bacteria and yeast, mostly lactic acid bacteria (LAB) belonging to genera Lactobacillus, Lactococcus, Bifidobacterium, Pediococcus, Streptococcus, and other gram-positive bacteria like Bacillus2,3. In addition, yeasts, such as Saccharomyces cerevisiae, have been reported as probiotics for animal nutrition4. Probiotics are used as additive feed in animal nutrition5,6 due to their health benefits to the host, improving weight gain, immunomodulation, nutrient digestibility, and resistance against pathogens, etc.7.

Microorganisms require nutrients for their growth, including carbohydrates (glucose, dextrose, lactose, and maltose), amino acids (nitrogen sources), and minerals8. In addition, microorganisms require phosphorus for growth and metabolism9. These nutrients are usually provided through commercial culture media such as Brain–Heart Infusion (BHI), Man Rogosa & Sharpe (MRS), and Tryptic Soy Broth (TSB), among others10,11.

Normally, probiotics intended for fish nutrition are produced in a commercial media, and then probiotic biomass is separated by centrifugation, washed, and resuspended in saline or phosphate buffer solution before inclusion in animal feed12,13,14. Nevertheless, these culture media are expensive because of nitrogen sources such as beef meat, peptones, yeast extract, meat extract, etc.15 and could represent 30–40% of the production costs16.

An alternative to commercial media is the use of agro-industrial by-products17,18 that may have a double purpose: as a source of nutrients for biomass production and as a wall material in drying to protect probiotic bacteria and to increase solid content for a more efficient encapsulation process19. The use of these food-graded by-products avoids the separation process for biomass recovery20. Therefore, the use of by-products in the culture media could reduce the cost of production and favor the circular economy21. Furthermore, by-products such as whey, molasses, and palm kernel cake are used in animal nutrition22,23,24. Several authors have reported that the use of “low-cost” components has reduced the cost of culture media by 62–86%16,25 and biomass production was similar or higher than that of commercial media.11,26. Whey27,28 and sugarcane molasses29,30 have been used for the production of lactic acid bacteria and Bacillus species. Although palm kernel cake is not commonly used as a culture media component for probiotic production, it has been used to produce enzymes and biomass from Bacillus subtilis31 and L. plantarum32, respectively.

Some of the probiotics mentioned above are of great interest to the scientific community because of their ability to produce bioactive compounds and probiotic characteristics with a positive effect on animal nutrition, specifically in aquaculture33. Therefore, Melo et al.34 characterized the microbial community composition of a continuous-flow competitive exclusion culture (CFEC) from the gut microbiomes of Nile tilapia (Oreochromis. niloticus), from which some bacteria were isolated. From this CFEC, three isolates (L. lactis and two Priestia species) showed tolerance to acidic pH, bile salts, and antibacterial activity against pathogens such as Streptococcus agalactiae and Aeromonas hydrophila. Also, these probiotic bacteria did not present hemolytic activity35. Most recently, these bacteria were included in fish feed36 and administered to Nile tilapia fingerlings in an in vivo trial37. These authors found that monostrain and multistrain administration resulted in the improvement of growth parameters, gut histology, immune regulation, and resistance to the pathogen Streptococcus agalactiae. Therefore, it is important to evaluate the production of these bacteria using agro-industrial by-products as culture media components.

To the best of our knowledge, there are no reports in the literature on dealing with the design of a culture media composed of whey, sugarcane molasses, and palm kernel cake to produce potential probiotic bacteria under monoculture conditions. Therefore, this study aimed to design an agro-industrial by-product-based culture media using these substrates as components for biomass production by L. lactis A12, Priestia megaterium M4, and Priestia sp. M10.

Methods

Ethical statement

The project followed the Colombian national government’s regulations. The Permit for accessing genetic resources was issued by the Colombian Ministry of Environment Number 117 (Otrosí4) on the 8th of May 2018 over 5 years.

Microorganisms

L. lactis A12, P. megaterium M4, and Priestia sp. M10 were isolated from a competitive exclusion bacterial culture derived from Nile tilapia (O. niloticus) gut microbiota34. Potential probiotic bacteria were identified by molecular techniques and sequenced the whole genome35. Bacteria were deposited under codes A12 (L. lactis A12), M4-MR4 (Priestia megaterium), and M10-MR10 (Priestia sp. M10) in the Chilean Collection of Microbial Genetic Resources (CChRGM) at the Instituto de Investigaciones Agropecuarias (INIA, Chillan, Chile). This institute is registered in the World Data Centre for Microorganisms (WDCM) with registration number 1067. These bacteria were stored in 1.5 mL Eppendorf tubes with BHI (Oxoid, UK) and 40% v/v glycerol at − 20 °C in a bacterial suspension: BHI volume ratio of 1:1. Bacteria were activated on TSA (Tryptic Soy Agar, Sharlau, Spain) at 28 °C for 48 h. Then, a single colony was taken from the TSA, and inoculated in BHI broth (Brain Heart Infusion, Oxoid, UK), and incubated overnight at 28 °C.

Preparation of culture media and fermentation conditions

Whey powder (Saputo, Colombia), sugarcane molasses (VitaAgro, Colombia), and PKC (Hacienda La Cabaña, Colombia) were used to design the culture media. PKC was ground with a Mill MF 10 basic (IKA, Germany) and sieved through a 1.0 mm mesh. The proximate composition of the by-products was determined according to protocols described by the Association of Official Analytical Chemists (AOAC)38, and the results are shown in Table 1. The components were mixed at different proportions according to the experimental design (Table 2) and added to distilled water to a final volume of 45 mL in a 250 mL shake flask. The culture media was supplemented with yeast extract (Sharlau, Spain) at 1.5% w/v and pH was adjusted to 7.60 ± 0.10 NaOH 1.0N. The final mixture was sterilized at 121 °C for 30 min. The culture media was then inoculated with 5 mL (10% v/v) of the bacterial inoculum (5.54 Log10 CFU/mL) and placed in an orbital incubator shaker (Innova 42, New Brunswick Scientific, USA) at 28 °C for 24 h. The agitation was set according to the experimental design (0 or 75 RPM). The culture media for activation, temperature, and incubation time were selected according to the methodology described by Melo-Bolívar et al.35.

Table 1 Proximal composition of the culture media components.
Full size table
Table 2 Mixture design with experimental results.
Full size table

Initial bacterial counts of L. lactis A12, P. megaterium M4, and Priestia sp. M10 was 4.80 ± 0.08, 4.73 ± 0.14, and 4.57 ± 0.14 Log10 CFU/mL, respectively. Viable cell count was determined after inoculation and at the end of the fermentation process using the plate count method in TSA at 28 °C for 24 h. The bacterial count was expressed as Log10 CFU/mL, and the bacterial growth change (BGC) was determined using Eq. (1).

$$BGC(Log_{10} ) = Log_{10} CFU/mL(final) - Log_{10} CFU/mL(initial)$$
(1)

Experimental design

The culture media design was carried out for each bacterium using an L-optimal mixture design with the statistical software Design Expert (Stat-Ease Inc., Minneapolis, MN, U.S.A). The design consisted of 22 runs, with three replicates in two reference mixture runs9. The mixture components were considered as the numerical factors: whey powder (1.00–3.84% w/v), sugarcane molasses (5.47–9.19% w/v), and PKC (0.77–2.73% w/v). The categorical factor was agitation (0 and 75 RPM) (Table 2). The component mixtures shown in Table 2 were obtained using several restrictions (constraints) in the mixture design, including the component concentration and composition, as well as the culture media composition (data not shown). The ranges used in this research were chosen according to reports from different authors27,29,31,39,40. The response variables were the BGC of L. lactis A12, P. megaterium M4, and Priestia sp. M10.

The best component mixture and agitation conditions that maximized the BGC values individually were achieved using the desirability function41. Validation experiments were performed in triplicate under the optimal conditions for each bacteria, and the results obtained for each bacteria in the designed culture media were compared with those obtained with BHI broth using a two-tailed t-test with an alpha level of 0.05. In addition, homogeneity of variance for the t-test was confirmed using a F-test (alpha level of 0.05).

Results

Model fitting of mixture design

The response variables of the mixture design for each bacteria are shown in Table 2. Bacterial growth changes in the fitted models were obtained by Analysis of Variance (ANOVA) at a confidence level of 95% and are presented in Table 3. ANOVA assumptions such as homoscedasticity, normality, and data independence were verified using residuals vs. predicted, normal probability (%) vs. residuals, and residuals vs. run number plots, respectively. The experimental data from each response were fitted to a linear model. This selection was made considering the lowest p-value for the mixture and process factors, and the corrected Akaike information criterion (AICc) (data not shown).

Table 3 ANOVA and fitting statistical parameters of bacterial growth change (BGC).
Full size table

P. megaterium M4 and Priestia sp. M10 models showed determination coefficients (R2 and Adjusted R2) values higher than 0.80. The lack-of-fit p-values were not significant for the models, which indicates that the model appropriately fitted the experimental data. The L. lactis A12 model had an R2 value lower than 0.60. Adequate precision is another statistical parameter that measures the signal-to-noise ratio. Values higher than 4.0 are desirable. The Adeq precision values for the three bacteria were greater than 4.0. This indicated that the models could be used to navigate the experimental design space. The model terms for P. megaterium M4 and Priestia sp. M10, linear mixture (whey, sugarcane molasses, and PKC), and interaction with agitation (AD, BD, and CD) significantly influenced the response variable because their corresponding p-values were lower than 0.05. However, for the L. lactis A12 model, only the linear mixture and interaction CD were significant. Values of determination coefficient equal to or higher than 0.80 are desirable42. Similar R2 values were reported previously by other authors for B. subtilis31,43,44.

Equations (2)–(7) show the fitted linear models for each bacteria, with and without agitation, expressed in terms of the actual components and factor levels.

Without agitation (0 RPM)

$$Priestia \, sp. \, M10 \, \left( {BGC} \right) = 0.1578 \times \left[ A \right] + 0.0671 \times \left[ B \right] + 0.0412 \times \left[ C \right]$$
(2)
$$P. \, megaterium \, M4 \, \left( {BGC} \right) = 0.2366 \times \left[ A \right] + 0.0353 \times \left[ B \right] + 0.0152 \times \left[ C \right]$$
(3)
$$L. \, lactis A12 \, \left( {BGC} \right) = 0.3523 \times \left[ A \right] + 0.3191 \times \left[ B \right] + 0.3389 \times \left[ C \right]$$
(4)

With agitation (75 RPM)

$$P. \, megaterium \, M4 \, \left( {BGC} \right) = 0.4280 \times \left[ A \right] + 0.1310 \times \left[ B \right] + 0.0603 \times \left[ C \right]$$
(5)
$$Priestia sp. \, M10 \, \left( {BGC} \right) = 0.4320 \times \left[ A \right] + 0.0958 \times \left[ B \right] + 0.1483 \times \left[ C \right]$$
(6)
$$L. \, lactis A12 \, \left( {BGC} \right) = 0.4401 \times \left[ A \right] + 0.3339 \times \left[ B \right] + 0.1913 \times \left[ C \right]$$
(7)
$$A: \, Whey(\% w/v), \, B: \, Sugarcane \, molasses(\% w/v), \, C: \, PKC(\% w/v).$$

Effect of mixture composition on bacterial cycle change (BGC)

Table 2 shows the experimental results of the mixture design. Figures 1, 2, and 3 present contour, trace, and desirability plots of the BGC for P. megaterium M4, Priestia sp. M10, and L. lactis A12, respectively.

Figure 1
figure 1

Contour (a and c), trace (b and d), and desirability (e) plots of BGC for P. megaterium M4.

Full size image
Figure 2
figure 2

Contour (a and c), Trace (b and d), and desirability (e) plots of BGC for Priestia sp. M10.

Full size image
Figure 3
figure 3

Contour (a and c), trace (b and d), and desirability plots of BGC for L. lactis A12.

Full size image

Trace plots are useful for evaluating the importance of components in a mixture. The trace plots of P. megaterium M4 (Fig. 1b and d) and Priestia sp. M10 (Fig. 2b and d) shows that whey is the component that most affects changes in BGC values, followed by sugarcane molasses, and PKC with and without agitation. In the case of L. lactis A12, it can be observed that without agitation at different mixtures of the components, the BGC values were similar; however, with agitation (Fig. 3d), whey and PKC were the components that most affected the BGC values of L. lactis A12.

As can be observed in Fig. 1a and c, higher BGC values were obtained with agitation (75 RPM) than with static cultures (0 RPM). The BGC values of P. megaterium M4 ranged from 0.33 to 2.61 Log10 (5.06–7.41 Log10 CFU/mL). For P. megaterium M4, media component mixtures with high whey concentrations and low molasses and PKC concentrations led to higher BGC values (Fig. 1b and d). In the case of Priestia sp. M10 (Fig. 2a–d), BGC values presented the same trend as in P. megaterium M4 with BGC values ranging between 0.55 and2.57 Log10 (5.30–7−09 Log10 CFU/mL).

Figure 3a and c show contour plots of BGC values for L. lactis A12, which ranged from 3.69 to 4.37 Log10 (8.63–9.13 Log10 CFU/mL). Similar trends were observed for P. megaterium M4 (Fig. 1) and Priestia sp. M10 (Fig. 2) cultures. Higher BGC values were observed when agitation was used at high whey and low PKC concentrations. Meanwhile, varying sugarcane molasses concentrations in the mixture did not show a significant effect (p > 0.05) on the changes in BGC. However, as shown in Fig. 3a, the BGC values were similar in different mixtures when agitation was not used (3.90–4.00 Log10).

Optimization and validation of optimal mixture

The desirability function was used to determine the optimal mixture and agitation conditions that maximized the BGC value for each probiotic bacteria. Figures 1e, 2e, and 3e show that the desirability value increases as whey concentration increases, and sugarcane molasses and PKC concentrations decrease in the mixture for the three bacteria species. The desirability values for P. megaterium M4 and Priestia sp. M10, and L. lactis A12 were 1.000, 0.978, and 0.906, respectively. The optimal conditions for each bacteria; and predicted and the experimental value of BGC are shown in Table 4. For P. megaterium M4 (p = 0.005) and Priestia sp. M10 (p = 0.021), the BGC was significantly higher in BHI compared to optimal mixture. However, the BGC values for L. lactis A12 were similar (p = 0.120) between optimal mixture and BHI broth. The Final pH of the culture media under optimal mixtures for Priestia species and L. lactis A12 were 6.44 ± 0.00 and 5.35 ± 0.02, respectively.

Table 4 Validation and comparison of BGC values for probiotic bacteria under the optimal mixture condition with commercial media.
Full size table

The designed culture media contains whey (38.4 g/L), sugarcane molasses (73.90 g/L), PKC (7.70 g/L), and yeast extract (15.00 g/L). After estimation and considering component costs, the average cost of the proposed culture media is USD $ 3.01/L (see Table 5) which is 86.93% lower than the BHI broth (USD $ 23.04/L). BHI broth contains brain infusion solids (12.5 g/L), beef heart solids (5.0 g/L), and proteose peptone (10.0 g/L) as nitrogen source. Also contains glucose (2.0 g/L) as carbon source, NaCl (5.0 g/L), and disodium phosphate (2.5 g/L) as buffering agent.

Table 5 Estimation of total cost of the culture media.
Full size table

Discussion

An alternative probiotic culture media was designed based on the mixture of agro-industrial by-products including whey, sugarcane molasses, and palm kernel cake for biomass production of L. lactis A12, P. megaterium M4, and Priestia sp. M10 in monoculture conditions. The proposed culture media was obtained from a mixture combined design, which has numerical restrictions on component composition to achieve the desired mixture and, also evaluated the effect of agitation on bacterial growth change values.

Priestia megaterium M4 and Priestia sp. M10 has genes related to the metabolism of sugars such as ribose, fructose, glucose, and sucrose that are present in molasses, however, low BGC values were reported at high sugarcane molasses concentrations. A possible explanation for this behavior could be related to the presence of heavy metals such as magnesium, zinc, manganese, calcium, and iron that may interfere with the metabolic pathways of bacteria resulting in the inhibition of growth45. On the other hand, a high concentration of whey (3.84% w/v) improved BGC for the three bacteria, this indicated that bacteria could use lactose as a carbon source for their growth. A similar trend was reported by Papizadeh et al.45. They evaluated the effect of different mixtures of cheese whey, cane molasses, corn steep liquor, and wheat germ extract on the dry biomass of L. plantarum strain RPR42. They found that a low concentration of cane molasses and a high concentration of cheese whey increased the biomass production of L. plantarum (~ 9.97 Log10 CFU/mL).

Other studies reported the use of sugarcane molasses and whey for the production of L. paracasei ssp. paracasei F19 (6.51–9.58 Log10 CFU/mL)25 and L. lactis CECT 539 (~ 10.00 Log10 CFU/mL)46, respectively. These values were higher than those found in this research (8.63–9.13 Log10 CFU/mL) for L. lactis A12. A possible explanation in the case of L. paracasei ssp. paracasei F19 is that molasses was hydrolyzed and centrifugated to increase glucose content and reduce heavy metals. For L. lactis CECT 539, the whey-based media was supplemented with MRS broth nutrients (except glucose and tween 80) at different concentrations to increase biomass and nisin production.

In the case of PKC, L. lactis A12 and Priestia species presented high BGC values at low PKC concentrations. This could be related that PKC contains a high content of carbohydrates (36.81%) such as non-starch-polysaccharides and crude fiber (27.76%). Probiotic bacteria present metabolic pathways that prefer the use of monosaccharides and disaccharides as carbon sources like those present in whey and molasses instead of complex carbohydrates and fiber47. Norizan et al.31 reported values of viable cells for B. subtilis ATCC11774 of 8.02–9.47 Log10 CFU/mL of in a culture media composed of PKC, peptone, and yeast extract, among other components. These values are higher than those reported in our study for Priestia megaterium M4 (5.30–7.09 Log10 CFU/mL) and Priestia sp. M10 (5.06–7.41 Log10 CFU/mL). The difference between our results and those reported for B. subtilis ATCC11774 could be associated that in this study fermentation was carried out for 72 h and the culture media was supplemented with nitrogen sources such as peptone and yeast extract.

Additionally, agitation improved nutrient transfer from the media to bacterial cells, resulting in higher BGC values for L. lactis A12 and Priestia species9. Also, agitation provides homogeneous conditions in the media through continuous mixing48.

Desirability values obtained for the three bacteria were higher than 0.7, which indicates a better optimization of experimental data of each response variable. Experimental errors for L. lactis A12, P. megaterium M4, and Priestia sp. M10 was 2.08, 4.88, and 2.01%, respectively. These values were lower than 10%, this indicates that the desirability function was a useful statistical tool for the optimization of culture media mixtures.

Concerning the final pH of culture media optimal mixtures, L. lactis A12 showed lower pH than Priestia species. This could be related to the ability of L. lactis to produce lactic acid and then decrease media pH27,39. It has been reported that the optimal pH range of L. lactis for nutrient consumption is 5.80–6.5046. Even though the final pH was outside the pH range, BGC values were similar to those observed in BHI. This situation could be overcome by adding a buffering such as sodium or potassium phosphate to maintain pH within optimal values8,15.

Although using BHI showed better BGC values for P. megaterium M4 and Priestia sp. M10, similar BGC values were achieved with L. lactis A12, indicating that the optimal mixture could be suitable for the biomass production of these bacteria. The proposed culture media contains nitrogen sources such as whey, PKC, and yeast extract that could be an alternative to those present in BHI broth. Glucose (2 g/L) in BHI could be replaced by lactose (2.99 g/L), sucrose, fructose, and glucose (56.74 g total sugars/L) as carbon sources present in whey and sugarcane molasses, respectively. The agro-industrial by-products used in the culture media were the most cost-effective components representing 5.68% of the total cost.

Conclusion

It was possible to design an agro-industrial by-products-based media to produce lactic acid bacteria and Bacillus species in monoculture conditions. The results obtained showed that this agro-industrial by-product culture media can be used as an efficient low-cost alternative to produce these potential probiotic bacteria for use in animal feed supplementation. The use of agro-industrial by-products for bacteria production would avoid downstream processes like centrifugation and washing reducing production costs.

Future studies should focus on producing these probiotic bacteria in a bioreactor, its encapsulation, in vitro, and in vivo assays to evaluate probiotic characteristics and their feasibility for the scale-up process.