Optimization of fermentation medium for advancing postbiotic antibacterial activity

Abstract

Antimicrobial postbiotic compounds have attracted increasing interest due to their potential applications in the food industry, particularly as candidates for biopreservation strategies. From a technological perspective, process optimization represents a valuable strategy for improving the production of postbiotic compounds and supporting the development of efficient fermentation bioprocesses. This study aimed to optimize the culture medium to enhance the antibacterial activity of a postbiotic derived from Pediococcus acidilactici CECT 9879. Plackett–Burman design (PBD) was first used to screen carbon and nitrogen sources, followed by one-factor-at-a-time (OFAT) experiments to evaluate yeast-derived ingredients and MRS-derived basal medium components, and response surface methodology based on a central composite design (CCD–RSM) for final medium optimization. Significant linear, quadratic, and interaction effects of key medium components, particularly yeast-derived ingredients and sodium acetate, were identified, indicating well-defined optimal concentration ranges. Under optimized composition, the minimum inhibitory concentration (MIC) against Escherichia coli decreased from 48.33 ± 2.89% (v/v) in the non-optimized medium to 31.67 ± 2.89% (v/v). The results highlight the role of nutrient optimization in modulating postbiotic antibacterial activity during fermentation and provide insights that may contribute to improving process efficiency in the agri-food sector.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13568-026-02076-8.

Introduction

In recent years, postbiotics have gained attention within the food sector due to their reported health-related benefits while avoiding the concerns associated with the use of live probiotic microorganisms (Amobonye et al. 2025). Accordingly, their production involves the inactivation of microbial cells while preserving their biological activity (Rafique et al. 2023). Postbiotics have been proposed as promising tools for improving food safety, as they may represent natural alternatives to conventional preservatives and potentially contribute to shelf life extension and maintenance of food quality (Moghadam et al. 2026; Sharafi et al. 2024). In addition to food applications, postbiotic preparations have also been investigated in animal nutrition, where they have shown positive effects on gut microbiota, immune function, and pathogen control in livestock (Chwen Loh et al. 2021). The benefits of postbiotics are closely associated with the fermentation process carried out by lactic acid bacteria (LAB). These are widely recognized as one of the main sources of postbiotic production, due to their capacity to synthesize diverse bioactive molecules, including polysaccharides, peptides, enzymes, bacteriocins, and organic acids, which exhibit antioxidant, antimicrobial, and anti-inflammatory activities (Moradi et al. 2020).

Depending on the metabolic pathway involved, fermentation in LAB can be homofermentative, yielding primarily lactic acid, or heterofermentative, producing lactic acid together with other by-products such as acetic acid, ethanol, and carbon dioxide. In addition to these main products, this fermentative process gives rise to a variety of bioactive molecules and cell-derived components that contribute to host health (Bueno et al. 2025). The composition and functional properties of postbiotics depend on the bacterial strain, the culture medium, and the processing conditions (Tong et al. 2023). Specifically, the nutrient composition of the fermentation medium has been shown to influence metabolite profiles and, consequently, the biological activity of postbiotic preparations (Chang et al. 2021).

As members of LAB, Pediococcus species exhibit technological and functional attributes that have supported their use as probiotics, starter cultures, and biopreservative candidates, highlighting their relevance in food systems and their potential contribution to human and animal health (Todorov et al. 2022). In particular, the postbiotic derived from Pediococcus acidilactici CECT 9879 has shown beneficial effects on metabolic health by regulating pathways associated with glucose and lipid metabolism (Yavorov-Dayliev et al. 2025). Moreover, a recent study evaluated the influence of nutritional sources on postbiotic production by this strain, showing that variations in carbon and nitrogen sources affected bacterial growth and antibacterial activity (Garrote Achou et al. 2025).

Advancing the study and practical application of postbiotics in the health and food sectors requires efficient and well-controlled manufacturing strategies. Nevertheless, optimizing postbiotic production remains challenging due to the complexity of metabolic pathways, strain-dependent differences, and the need to carefully adjust fermentation conditions (Prajapati et al. 2023). In this context, Design of Experiments (DoE) has been increasingly adopted for bioprocess improvement. Because such systems often involve multiple interacting factors, a structured approach is required to evaluate their influence on system response. DoE is based on the systematic planning and execution of experimental trials, enabling efficient assessment of factor effects on process performance (Balakrishnan et al. 2022). Among the experimental strategies encompassed within the DoE framework, Plackett–Burman design (PBD) is frequently used as a screening tool to identify significant factors among multiple process parameters with a reduced number of trials. The factors identified can then be further optimized using response surface methodology (RSM), which applies structured designs and quadratic modeling to evaluate interactions and determine optimal conditions for maximum process efficiency (Shree and V 2025).

Recent studies have successfully applied optimization techniques to refine culture medium formulation for probiotic strains, including Enterococcus faecium NCIMB 11,181 (Soleymanzadeh et al. 2025), Lacticaseibacillus casei UT1 (Afshar et al. 2025), Bifidobacterium longum HSBL001 (Cheng et al. 2025), Bacillus subtilis Ö-4-68 (Demirhan-Yazıcı et al. 2025), Bifidobacterium animalis subsp. lactis INL1 (Zacarías et al. 2025), and Lactobacillus paracasei subsp. paracasei F19 (Acosta-Piantini et al. 2023). While previous studies have provided valuable insights into culture medium optimization, they have largely focused on biomass production or probiotic viability. In contrast, optimization strategies specifically aimed at improving the antibacterial performance of postbiotics, particularly using animal-free formulations, remain less explored.

Therefore, the aim of this study was to optimize an animal-free culture medium for Pediococcus acidilactici CECT 9879, with particular emphasis on enhancing the antibacterial activity of the resulting inactivated preparation. This work contributes to the application of statistical optimization strategies to improve postbiotic functionality, providing a basis for future studies in the context of food and feed-related applications.

Methodology

Experimental design

The overall experimental strategy is summarized in Fig. 1. The optimization of the culture medium considered carbon sources, nitrogen sources, and basal medium composition. The initial selection of carbon and nitrogen sources was based on previous preliminary OFAT experiments. In the present study, PBD was applied to identify the most influential carbon and nitrogen sources. Subsequently, targeted OFAT experiments were performed to explore alternative yeast-derived ingredients in terms of industrial feasibility and to evaluate the relevance of selected basal medium components, thereby refining factor selection and defining the experimental ranges for subsequent multivariate optimization. The selected factors were further optimized using CCD within the framework of RSM. Finally, the predicted optimum conditions were experimentally validated. The overall workflow comprised sequential stages of screening, refinement, modelling, optimization, and validation.

Fig. 1.

Schematic diagram of the experimental design for statistical optimization of the fermentation medium to enhance postbiotic antibacterial activity

General experimental conditions

The lactic acid bacterium Pediococcus acidilactici CECT 9879, obtained from the Spanish Type Culture Collection (CECT, Paterna, Spain), was used in this study. Escherichia coli strain 097040, obtained from the microorganism collection of PENTABIOL S.L., was used as the indicator strain for antibacterial activity assessment. Strain reactivation, subculturing, maintenance, fermentation at 37 °C for 24 h, and subsequent inactivation by thermal treatment at 80 °C for 1 h, corresponding to the postbiotic production process, were carried out based on a previously described method (Garrote Achou et al. 2025).

Cell growth, expressed as colony-forming units per millilitre (log CFU/mL), and antibacterial activity, expressed as minimum inhibitory concentration (MIC, % v/v), were determined as response variables throughout the study following the methodology previously described by Garrote Achou et al. (2025). Briefly, after fermentation, serial dilutions were prepared and viable cell counts were determined by the plate count method on MRS agar. For antibacterial activity assessment, cell-free supernatants were obtained by centrifugation of inactivated cultures and diluted in BHI medium. These supernatants contained the metabolites produced during fermentation. MIC values, expressed as % (v/v), corresponded to the proportion of supernatant present in the assay mixture prior to the addition of the indicator strain suspension. Equal volumes of diluted supernatant and indicator strain suspension were then combined. The following controls were included: the indicator strain in BHI medium, non-fermented medium inoculated with the indicator strain, and cell-free supernatant without the indicator strain.

Experimental culture media were formulated through a sequential statistical optimization strategy, including Plackett–Burman design (PBD), one-factor-at-a-time (OFAT) experiments, and response surface methodology (RSM). The carbon sources used throughout the study were glucose (Pintaluba S.A., Tarragona, Spain), xylose (Laboratoriumdiscounter, Ijmuiden, The Netherlands), and ribose (TCI, Tokyo Chemical Industry Co., Ltd., Tokyo, Japan). Selected nitrogen sources and yeast-derived ingredients were coded due to confidentiality agreements associated with the industrial collaboration.

Plackett–Burman design for carbon and nitrogen sources screening

A Plackett–Burman design (PBD) was applied to identify the nutritional factors significantly influencing responses related to postbiotic production, specifically antibacterial activity and cell growth, in P. acidilactici. Based on prior experiments, three carbon and three nitrogen sources were selected for evaluation due to their relevance for fermentation performance (Garrote Achou et al. 2025). The carbon sources tested were glucose, xylose, and ribose, while the nitrogen sources included soy peptone, cotton hydrolysate, and yeast extract.

All carbon and nitrogen sources were tested at two concentration levels, defined as high (+ 1) and low (− 1). Carbon sources were evaluated at 30 g/L (high level) and 10 g/L (low level), while nitrogen sources were tested at 10 g/L (high level) and 5 g/L (low level). The concentration levels were selected based on commonly reported values for lactic acid bacteria cultivation to define a biologically relevant range suitable for factor screening (Abbasiliasi et al. 2017).

The experimental design consisted of 12 runs. The Plackett–Burman design matrix defining the experimental combinations of carbon and nitrogen sources is shown in Supplementary File as Table 1.

Table 1.

Coded levels and actual values of the independent factors used in the CCD

Factor	Abbreviation	−α	−1	0	+ 1	+α
Yeast extract	YE	0.48	1.50	3.00	4.50	5.52
Yeast-derived ingredient 1	YDI1	3.18	10.00	20.00	30.00	36.82
Sodium acetate	SA	3.18	10.00	20.00	30.00	36.82

Culture media were prepared according to the combinations of carbon and nitrogen sources defined by the experimental design. In all cases, formulations were supplemented with a constant MRS-derived basal medium consisting of sodium acetate (5.00 g/L), dipotassium phosphate (2.00 g/L), polysorbate 80 (1.00 g/L), magnesium sulfate (0.20 g/L), and manganese sulfate (0.05 g/L).

OFAT evaluation of yeast-derived ingredients

Based on the relevance of yeast extract identified during the screening phase and its industrial applicability, four alternative yeast-derived ingredients (YDI) with distinct compositional profiles and degrees of processing, including hydrolyzed preparations, extract fractions, nucleotide-rich fractions, and cell wall-derived fractions with commercial application profiles, were selected for evaluation. The ingredients were assigned the codes YDI1–YDI4, which do not correspond to their compositional characteristics. OFAT experiments were conducted by substituting the nitrogen sources of the MRS formulation (bacteriological peptone, meat extract, and yeast extract) with each yeast-derived ingredient at a concentration of 22.00 g/L. In all cases, glucose (20.00 g/L) was used as the carbon source, while the remaining MRS-derived basal medium components, including ammonium citrate (2.00 g/L), were maintained at their standard concentrations.

OFAT evaluation of MRS-derived basal medium components

The influence of individual MRS-derived basal medium components on postbiotic production was assessed using a one-factor-at-a-time (OFAT) approach. In this strategy, one medium component was varied across its defined concentration range, while all remaining components were maintained at their standard MRS concentrations. Sodium acetate (0.00–50.00 g/L), ammonium citrate (0.00–25.00 g/L), dipotassium phosphate (0.00–25.00 g/L), polysorbate 80 (0.00–10.00 g/L), magnesium sulfate (0.00–25.60 g/L), and manganese sulfate (0.00–25.60 g/L) were evaluated individually. In all experiments, carbon and nitrogen sources were kept constant at 20.00 g/L glucose, 10.00 g/L bacteriological peptone, 8.00 g/L beef extract, and 4.00 g/L yeast extract.

Response surface methodology for medium optimization

Response surface methodology (RSM) was employed to optimize the fermentation medium composition for P. acidilactici. Based on the results obtained from the preceding Plackett–Burman design (PBD) and OFAT experiments, a central composite design (CCD) was selected to evaluate the linear, quadratic, and interaction effects of three independent factors on postbiotic antibacterial activity: yeast extract (YE), yeast-derived ingredient 1 (YDI1), and sodium acetate (SA). Each factor was studied at five coded levels (−α, − 1, 0, + 1, +α), with the corresponding actual values reported in Table 1.

−α, − 1, 0, + 1, and + α denote the coded factor levels; values are expressed in g/L.

The CCD consisted of a full two-level factorial design (2³), comprising eight factorial points, six axial points, and six replicates at the center point, resulting in a total of 20 experimental runs. The axial distance was set at α = 1.68 to ensure rotatability of the design. Culture media were formulated according to the factor combinations defined by the experimental design and supplemented with the MRS-derived basal medium described above, without ammonium citrate. The design matrix, expressed in coded factor levels, is presented in Supplementary File as Table 2.

Table 2.

Estimated effects of carbon and nitrogen sources on antibacterial activity (MIC) and bacterial growth (log CFU/mL) according to the Plackett–Burman design

Source	MIC			Log CFU/mL
	Effect	Coefficient	p-value	Effect	Coefficient	p-value
Glucose	2.500	1.250	0.219	0.012	0.006	0.782
Xylose	3.057	1.528	0.146	−0.145	−0.073	0.015
Ribose	0.833	0.417	0.659	−0.585	−0.292	0.000
Soy	−3.610	−1.805	0.098	0.118	0.059	0.031
Cotton	−0.277	−0.138	0.882	−0.002	0.001	0.968
Yeast	−8.610	−4.305	0.005	0.138	0.069	0.018

The experimental data obtained from the CCD were fitted to a second-order polynomial model to describe the relationship between the independent factors and the response and to predict the optimal medium composition, as expressed by the following equation:

In this equation, Y represents the predicted response; X_i and Xⱼ are the coded independent factors; β₀ is the intercept term; β_i, β_ii, and β_iⱼ represent the linear, quadratic, and interaction coefficients, respectively; and ε represents the experimental error.

Model validation was carried out by experimentally evaluating the optimal composition predicted by the CCD–RSM model and comparing the observed responses with the corresponding predicted values. The optimized fermentation medium was experimentally compared with a non-optimized reference medium corresponding to the standard MRS formulation, consisting of glucose (20.00 g/L), bacteriological peptone (10.00 g/L), beef extract (8.00 g/L), yeast extract (4.00 g/L), sodium acetate (5.00 g/L), ammonium citrate (2.00 g/L), dipotassium phosphate (2.00 g/L), polysorbate 80 (1.00 g/L), magnesium sulfate (0.20 g/L), and manganese sulfate (0.05 g/L). Antibacterial activity, expressed as minimum inhibitory concentration (MIC, % v/v), was used as the primary response variable, while cell growth, expressed as colony-forming units per milliliter (log CFU/mL), was determined as a complementary variable to assess postbiotic production under both conditions.

Statistical analysis

All experimental designs and statistical analyses were conducted using Minitab^® Statistical Software Version 22 (Minitab LLC, State College, PA, USA). Analysis of variance (ANOVA) was applied to identify significant factors in the Plackett–Burman design (PBD) and to evaluate model significance in the response surface methodology based on a central composite design (CCD–RSM). One-way ANOVA followed by Tukey’s post hoc test was used for the one-factor-at-a-time (OFAT) experiments. Comparisons between optimized and non-optimized fermentation media were performed using Student’s t-test. In all cases, statistical significance was established at p ≤ 0.05. Experiments were conducted in triplicate, and results are reported as mean values.

Results

Plackett–Burman design for carbon and nitrogen sources screening

The influence of selected carbon and nitrogen sources on antibacterial activity, expressed as minimum inhibitory concentration (MIC), and on bacterial growth (log CFU/mL) was evaluated using a Plackett–Burman design. Yeast extract was the only factor that significantly influenced antibacterial activity (Table 2), as increasing its concentration resulted in a significant reduction in MIC values, indicating enhanced antibacterial activity of the produced postbiotic. Regarding bacterial growth, increasing concentrations of xylose and ribose led to lower log CFU/mL values, whereas higher levels of yeast extract and soy peptone promoted cell growth.

The coefficients of determination (R²) of the fitted linear models were 0.87 and 0.98 for MIC and log CFU/mL, respectively, indicating that 87% and 98% of the variability in MIC and log CFU/mL, respectively, were explained by the model. The corresponding Pareto charts (Fig. 2) provide a visual representation of the absolute standardized effects and the associated significance threshold; factors exceeding this reference line were identified as significantly influencing postbiotic antibacterial activity and bacterial growth (p < 0.05) in P. acidilactici.

Fig. 2.

Pareto charts of standardized effects of carbon and nitrogen sources on a MIC and b log CFU/mL

Based on the Plackett–Burman screening results, yeast extract was selected for further optimization through response surface methodology (RSM), as it showed a significant effect on both postbiotic antibacterial activity and bacterial growth in the strain under study.

OFAT evaluation of yeast-derived ingredients

The replacement of the nitrogen sources in the MRS formulation with individual yeast-derived ingredients led to distinct differences in postbiotic antibacterial activity and cell growth (Table 3). Among the evaluated ingredients, YDI1 exhibited the lowest MIC value (63.33 ± 2.89%) and the highest cell density (9.43 ± 0.12 log CFU/mL). In contrast, YDI2 showed a higher MIC (96.67 ± 2.89%) together with reduced bacterial growth (8.79 ± 0.08 log CFU/mL). No inhibitory activity was detected for YDI3 and YDI4 under the tested conditions, although both supported cell densities comparable to those observed for YDI2.

Table 3.

Influence of yeast-derived ingredients on postbiotic antibacterial activity (MIC) and cell growth (log CFU/mL)

Yeast-Derived Ingredient (YDI)	MIC (% v/v)	Cell growth (log CFU/mL)
YDI1	63.33 ± 2.89	9.43 ± 0.12
YDI2	96.67 ± 2.89	8.79 ± 0.08
YDI3	NI*	8.59 ± 0.09
YDI4	NI*	8.78 ± 0.17

^* NI: No inhibitory activity detected against E. coli under the tested conditions

In the present study, different yeast-derived ingredients were therefore evaluated, and YDI1 was selected for subsequent medium optimization studies due to its more favorable balance between antibacterial activity and cell growth.

OFAT evaluation of MRS-derived basal medium components

Variations in MRS-derived basal medium composition resulted in observable changes in antibacterial activity (MIC) and cell growth (log CFU/mL). Figure 3 shows the corresponding response profiles obtained at increasing concentrations of individual medium components, allowing a visual assessment of stimulatory and inhibitory trends. Increasing sodium acetate concentrations were associated with a decrease in MIC values, whereas cell growth exhibited slight fluctuations across the tested range.

Fig. 3.

Influence of sodium acetate, dipotassium phosphate, and ammonium citrate on antibacterial activity (MIC) and cell growth (log CFU/mL)

In contrast, increasing dipotassium phosphate concentrations showed a trend toward higher MIC values, and no detectable antibacterial activity was observed at concentrations above 22.5 g/L. Regarding cell growth, log CFU/mL values decreased at elevated phosphate concentrations.

Ammonium citrate exhibited a response pattern similar to that observed for dipotassium phosphate, with MIC values showing an upward trend as its concentration increased. With respect to cell growth, values remained relatively stable across the tested concentration range. For polysorbate 80, magnesium sulfate, and manganese sulfate, both MIC and cell growth values remained largely unchanged across the tested concentration ranges (Fig. 4).

Fig. 4.

Influence of polysorbate 80, magnesium sulfate, and manganese sulfate on antibacterial activity (MIC) and cell growth (log CFU/mL)

Among the evaluated MRS-derived basal medium components, sodium acetate was selected for further optimization, as it was the only component that showed a reduction in MIC values without substantially affecting cell growth, making it a suitable candidate for improving antibacterial performance.

Response surface methodology for medium optimization

The experimental results obtained from the central composite design (CCD) were fitted to a second-order polynomial model to describe the effects of yeast extract (YE), yeast-derived ingredient 1 (YDI1), and sodium acetate (SA) on postbiotic antibacterial activity. A reduced quadratic model was obtained by removing non-significant terms (p > 0.05) while maintaining model hierarchy and overall adequacy to properly represent the response surface. The coded regression coefficients of the reduced quadratic model are presented in Table 4. The regression analysis showed that all linear terms (YE, YDI1, and SA) significantly affected MIC values (p < 0.01). The negative linear coefficients indicate that increasing the concentration of each factor individually reduced MIC within the experimental range. Significant quadratic effects were observed for YDI1 (p < 0.001) and SA (p < 0.05), both with positive coefficients, confirming the presence of curvature in the response surface. The quadratic term of YE was not statistically significant (p = 0.103). A significant interaction between YE and YDI1 was detected (p < 0.001), and the positive interaction coefficient indicates that simultaneous increases in both factors were associated with higher MIC values than expected from their individual effects. The interaction between YE and SA showed a marginal effect (p = 0.069).

Table 4.

Coded regression coefficients of the reduced quadratic model for postbiotic antibacterial activity

Term	Coefficients*	Standard Error	p-value
Intercept	37.520	1.430	< 0.001
YE	−6.241	0.948	< 0.001
YDI1	−11.516	0.948	< 0.001
SA	−3.927	0.948	0.002
YE²	1.640	0.923	0.103
YDI1²	6.059	0.923	< 0.001
SA²	2.524	0.923	0.019
YE x YDI1	6.250	1.240	< 0.001
YE x SA	−2.500	1.240	0.069

*Coefficients are expressed in coded units. Only terms retained in the reduced quadratic model are presented

The ANOVA (in Supplementary File as Table 3) revealed that the reduced quadratic model was statistically significant (p < 0.001), confirming that the model explains variation in the response. The lack-of-fit was not statistically significant (p = 0.207), suggesting that the model adequately describes the relationship between the response and the predictors. The coefficient of determination (R² = 0.963) indicates a good fit of the model to the experimental data, explaining 96.30% of the total variation in the response. The adjusted R² (0.936) further supports the adequacy of the model after accounting for the number of predictors. Moreover, the predicted R² (0.837) was reasonably close to these R² values, reflecting good predictive performance without evidence of model overfitting.

Three-dimensional response surfaces and the corresponding contour plots (Fig. 5) were generated to illustrate the relationship between medium components and antibacterial activity. The plots depict the simultaneous variation of two factors while the third was maintained at its central level. Since lower MIC values correspond to higher antibacterial activity, the optimal regions were identified at the lowest areas of the response surfaces and the darkest blue regions in the contour plots.

Fig. 5.

Three-dimensional response surface and corresponding contour plots illustrating the combined effects of yeast extract (YE), yeast-derived ingredient 1 (YDI1), and sodium acetate (SA) on antibacterial activity (MIC). Each plot represents the interaction between two factors while the third was maintained at its central level

The response surface analysis revealed distinct patterns among the factor combinations. For YE and YDI1, the surface exhibited a clear downward inclination along the YE axis, reflecting its significant negative linear effect on MIC, while a steeper slope along the YDI1 axis highlighted its stronger contribution to antibacterial activity. A twisting of the surface was observed, consistent with the significant interaction between both components and indicating that the effect of one factor depended on the level of the other. The YE–SA combination displayed curvature primarily along the SA axis, consistent with its significant quadratic effect. A surface asymmetry was observed, considerably less pronounced than in the YE–YDI1 combination, in agreement with the marginal interaction term identified in the fitted model. The YDI1–SA surface showed pronounced curvature along both axes, reflecting the positive quadratic terms of YDI1 and SA and indicating that the reduction in MIC became progressively less marked at higher concentrations. The curvature appeared more symmetrical, and no noticeable surface twisting was observed, supporting the absence of a significant interaction term.

The model predicted optimal concentrations of 5.52 g/L yeast extract (YE), 20.85 g/L yeast-derived ingredient 1 (YDI1), and 36.14 g/L sodium acetate (SA). To evaluate the predictive capability of the CCD–RSM model, experimental validation was performed under the predicted optimal conditions. The model predicted a MIC value of 25.07%, with a 95% confidence interval ranging from 14.17 to 35.97%. The experimentally obtained MIC was 31.67 ± 2.89%. The experimental mean fell within the model’s 95% confidence interval, confirming the adequacy of the CCD–RSM model to describe the system behavior within the studied experimental domain.

The antibacterial performance of the optimized medium was compared with that of the non-optimized formulation. The MIC decreased significantly (p = 0.002), from 48.33 ± 2.89% in the non-optimized medium to 31.67 ± 2.89% under optimized conditions. This represents an absolute reduction of 16.66% points and a relative reduction of 34.48%, calculated with respect to the MIC obtained in the non-optimized medium. Although the optimization model was developed using MIC as the sole response variable, bacterial growth was also evaluated to determine whether the improved antibacterial activity was associated with differences in biomass production. No significant differences in viable cell counts were observed between the non-optimized (9.64 ± 0.04 log CFU/mL) and optimized media (9.65 ± 0.06 log CFU/mL) (p = 0.886). The final pH of the inactivated fermentation culture differed significantly, with higher values observed under optimized conditions (4.68 ± 0.04) compared to the non-optimized formulation (3.89 ± 0.05) (p < 0.001).

Discussion

Plackett–Burman design for carbon and nitrogen sources screening

On the one hand, the antimicrobial properties of postbiotics have been associated with metabolites such as organic acids, bacteriocins, peptides, and volatile compounds (Prajapati et al. 2023). For example, phenyllactic acid (PLA) has been described as a broad-spectrum antimicrobial compound produced by several lactic acid bacteria through amino acid metabolism (Rajanikar et al. 2021). Previous studies have demonstrated that PLA and hydroxyphenyllactic acid (HPLA) production by P. acidilactici DSM 20,284 can be significantly enhanced by supplementing MRS medium with their corresponding amino acid and ketoacid precursors (Mu et al. 2012). Phenylalanine and tyrosine are present in yeast extract at concentrations ranging approximately from 2.64 to 5.30 g/100 g and from 0.40 to 5.30 g/100 g, respectively (Tao et al. 2022). Overall, the compositional profile of yeast extract, particularly its amino acid content, may represent a source of metabolic precursors potentially involved in the biosynthesis of antimicrobial compounds. In addition to its nutritional role, yeast extract may also contain peptides capable of influencing regulatory mechanisms. Increased expression of short coding sequences has been reported during the growth of Streptococcus thermophilus in yeast extract-containing media. These sequences are recognized as precursors of signaling peptides involved in quorum sensing-regulated processes, including bacteriocin production (Proust et al. 2020). Although these mechanisms and metabolites were not specifically identified in the resulting postbiotic preparation, previous literature suggests that yeast extract supplementation may favor the production of antimicrobial compounds such as organic acids, bacteriocins, PLA, HPLA, and other peptide-derived metabolites, which could plausibly contribute to the enhanced antibacterial activity observed under the evaluated conditions.

On the other hand, nitrogen utilization in lactic acid bacteria relies on a complex proteolytic system in which cell envelope proteinases (CEPs) degrade proteins into oligopeptides. These peptides are subsequently transferred into the cell through dedicated transport systems for oligopeptides, dipeptides, and tripeptides (Opp, DtpP, and DtpT, respectively), and are then hydrolyzed into free amino acids (Wang et al. 2021). These amino acids are essential for cellular functions, particularly protein synthesis and growth (Kieliszek et al. 2021). In this context, yeast extract represents a highly relevant nitrogen source for microbial fermentation, as it provides readily available peptides and amino acids that are key contributors to LAB development (Tao et al. 2022). Collectively, these properties may explain the growth-promoting effect of yeast extract on P. acidilactici observed in this study.

The significant effect of yeast extract observed in the present Plackett–Burman design is consistent with previous reports using similar statistical screening approaches. In Pediococcus pentosaceus TL-3, yeast extract exhibited the strongest stimulatory effect on growth compared to peptone and meat extract (Lim et al. 2019). Likewise, in Pediococcus acidilactici 72 N, yeast extract was identified as the only nitrogen source exerting a significant influence on viable cell production when compared with beef extract and sweet whey (Myo et al. 2025). In this context, the present study evaluated the influence of nutritional factors in relation to antibacterial activity associated with postbiotic production, highlighting the potential relevance of yeast extract not only as a growth-supporting ingredient but also as a possible contributor to the functional activity of the obtained postbiotic preparation.

OFAT evaluation of yeast-derived ingredients

The results suggest that antibacterial activity was influenced by the specific nutritional composition of the yeast-derived ingredients rather than being exclusively dependent on cell proliferation, as no inhibitory activity was detected for some ingredients (YDI3 and YDI4) despite supporting comparable cell growth. A similar observation has been reported in Lactococcus lactis Gh1, where growth in LB medium did not result in detectable bacteriocin-like inhibitory substance (BLIS) production, whereas other media supported both biomass formation and antimicrobial activity. In that study, BLIS synthesis was described as not strictly following the typical growth-associated bacteriocin production pattern (Jawan et al. 2020).

From an industrial perspective, raw materials used in fermentation media must be cost-effective and consistently available. However, complex nutrients such as yeast extracts are inherently subject to compositional variability due to differences in biological origin and manufacturing conditions. To overcome this challenge, the use of blends of the same ingredient type obtained from different manufacturers has been described as a common strategy to improve process robustness (Ummadi and Curic-Bawden 2010).

OFAT evaluation of MRS-derived basal medium components

A reported mechanism for acetate utilization in bacteria involves its conversion into acetyl-coenzyme A (Ac-CoA), which serves as a key metabolic intermediate supporting the formation of biosynthetic precursors (Hosmer et al. 2024). In Lactococcus lactis, the acetate pathway has been shown to contribute to growth yield under respiratory conditions (Cesselin et al. 2018), and the presence of acetate in culture media has been reported to improve cellular physiology and metabolic performance in Lactobacillus rhamnosus ATCC 7469 (Barboza et al. 2025). Acetate has also been implicated in the regulation of antimicrobial compound production in certain LAB species, suggesting a possible metabolic link between acetate availability and antimicrobial activity. Moreover, activation of bacteriocin synthesis through quorum-sensing control mechanisms has been reported in Lactobacillus species (Meng et al. 2021). Beyond the documented role of acetate in bacterial metabolism, sodium acetate is commonly incorporated into LAB media as a buffering agent to maintain optimal pH during acid accumulation associated with bacterial growth. Its omission has been linked to reduced biomass formation due to rapid pH decline (Hayek et al. 2019). In this context, these findings may help explain the reduced MIC values observed with increasing sodium acetate concentrations, suggesting a potential metabolic contribution to antibacterial activity.

Although extracellular phosphate plays an important regulatory role in central metabolism and efficient glucose utilization in lactic acid bacteria (Levering et al. 2012), its availability in culture media has also been associated with inhibitory effects on the production of antimicrobial compounds in certain strains. Specifically, phosphate has been reported to reduce bacteriocin production in Lactobacillus salivarius CRL 1328 (Juárez Tomás et al. 2010) and to negatively affect the antimicrobial activity of postbiotic derived from Lactiplantibacillus plantarum RS5 (Ooi et al. 2021). Taken together, these findings may account for the increased MIC values observed at higher phosphate concentrations in the present study.

While citrate utilization in LAB has been associated with enhanced growth performance and improved resistance to acid stress (Eicher et al. 2023), the ability to degrade citrate is not widespread among lactic acid bacteria and has been primarily reported in specific strains, such as Lactococcus lactis subsp. lactis biovar diacetylactis (van Mastrigt et al. 2018). This metabolic capability requires the presence of genes involved in the synthesis of citrate permease and citrate lyase (Eicher et al. 2023). Therefore, its potential contribution under the present experimental conditions remains to be elucidated, which may be associated with the increased MIC values observed at higher ammonium citrate concentrations in P. acidilactici, suggesting strain-dependent differences in citrate metabolism.

Response surface methodology for medium optimization

These surface patterns can be biologically interpreted considering the known role of nitrogen sources in antimicrobial production. Previous studies indicate that yeast extract is frequently a preferred nitrogen source for antimicrobial metabolite production in lactic acid bacteria, largely due to its high content of readily assimilable amino acids, low-molecular-weight peptides (< 1 kDa), vitamins, and growth factors (Peng et al. 2023). In the present study, the significant negative linear effects of yeast-derived ingredients (YE and YDI1) on MIC indicate that increasing their concentration enhanced overall antibacterial activity across most of the experimental range. However, the significant positive quadratic coefficient observed for YDI1 demonstrates that this relationship was not strictly linear. Although increasing YDI1 reduced MIC within the studied domain, the magnitude of improvement progressively diminished at higher concentrations. Nitrogen is required for biological growth, particularly for cellular protein synthesis and nucleic acid production, and adequate supplementation is essential for microbial metabolism during fermentation. Increasing nitrogen levels have been associated with greater availability of peptides and growth-promoting factors that may act as inducers of antimicrobial metabolite biosynthesis (Abbasiliasi et al. 2017). Nevertheless, excessive nitrogen availability has been reported to shift metabolism toward biomass formation rather than antimicrobial metabolite production in certain LAB species, which may result in reduced antibacterial activity despite continued growth (Phakathi et al. 2026). A comparable non-linear metabolic response has been reported in Lactobacillus plantarum Hui1, where nitrogen concentrations above 20 g/L reduced lactic acid production, indicating that excessive nitrogen levels can exert inhibitory or toxic effects (Saavedra et al. 2021). In addition, the significant YE–YDI1 interaction indicates that the combined influence of these yeast-derived ingredients was not simply additive. This finding agrees with previous reports suggesting that the performance of mixed nitrogen sources is influenced by their type, dosage levels, and interactions with other components of the medium (Phakathi et al. 2026).

The relevance of this behavior is further supported by studies employing RSM to optimize antimicrobial activity in lactic acid bacteria. For example, culture medium optimization increased the inhibition zone of cell-free supernatants (CFS) from Lactiplantibacillus plantarum K014 against Cutibacterium acnes from 17.33 mm to 21.67 mm following multivariate optimization of culture conditions, including yeast extract (Kong et al. 2025). Likewise, antimicrobial activity associated with bacteriocin production by Lactococcus lactis NCU036019 against Staphylococcus aureus increased from 46.19 AU/mL to 300.14 AU/mL, with yeast extract identified among the significant medium components (Wei et al. 2025). In Lactobacillus brevis DF01, bacteriocin activity increased approximately four-fold under optimized medium formulations where yeast extract concentration contributed significantly to the response (Lee et al. 2012). Overall, these findings highlight the suitability of RSM for culture medium optimization. These results are also consistent with the significant influence of yeast-derived ingredients observed in the present study and suggest the existence of an optimal nutrient concentration range for antibacterial activity.

Sodium acetate also exhibited significant linear and quadratic effects on MIC, indicating its role in modulating antibacterial activity. In Latilactobacillus sakei NRIC 1071(T), the presence of sodium acetate increased growth yield by 1.6 times, enhanced lactic acid production, and elevated the activity of enzymes involved in the glycolytic pathway compared to conditions in its absence (Iino et al. 2002). Consistently, response surface optimization studies have reported adjustments in sodium acetate concentration as part of medium refinement strategies aimed at enhancing antimicrobial metabolite synthesis. For example, in Lactococcus lactis NCU036019, increasing sodium acetate to 10 g/L contributed to a marked rise in the antibacterial activity of lactococcin036019 in the CFS (Wei et al. 2025). Similarly, in Enterococcus faecium LR/6, optimization of sodium acetate to 11 g/L resulted in enhanced enterocin production (Kumar and Srivastava 2010). Collectively, these observations identify sodium acetate as a relevant medium component influencing antibacterial performance, in agreement with its significant effects observed in the present study.

The response surface optimization identified the combination of medium components associated with the lowest predicted MIC. The predicted optimum concentrations for YDI1 and SA were located within the evaluated concentration range, whereas the predicted optimum for YE coincided with the upper axial point of the CCD, suggesting that exploration of higher YE concentrations may help refine the experimental domain. Overall, the obtained results suggest that the enhanced antibacterial activity may be associated with changes in bacterial metabolism rather than differences in cell growth, as viable cell counts remained similar between the optimized and non-optimized media despite differences in MIC values. In addition, the optimized condition exhibited stronger antibacterial activity despite showing a higher final pH, suggesting that acidity alone may not fully explain the observed inhibitory effect. Therefore, the improved postbiotic antibacterial activity observed with the optimized culture medium may be related to changes in metabolite production by P. acidilactici.

Conclusions

This study optimized the culture medium to enhance the antibacterial activity of the postbiotic derived from Pediococcus acidilactici CECT 9879 using response surface methodology (RSM), supporting the usefulness of this statistical approach for postbiotic production. The model identified optimal concentrations of yeast-derived ingredients (YE and YDI1) and sodium acetate (SA), with significant quadratic and interaction effects indicating that the antibacterial performance of the postbiotic depends on balanced nutrient modulation rather than simple linear effects. These findings support the feasibility of animal-free medium formulations in fermentation bioprocesses and may provide a useful basis for improving postbiotic production in related lactic acid bacteria, although validation in additional strains is still required. Additional studies including chemical characterization of the postbiotic and evaluation in food systems may further support the applicability and scalability of the optimized process.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

C.G.A: Conceptualization, Methodology, Investigation, Formal analysis, Writing—original draft. M.J.C.D.: Conceptualization, Methodology, Validation, Supervision, Writing—review and editing. A.F.C.: Investigation.

Funding

Open Access funding provided by Universidad Pública de Navarra. This research was financially supported by the Government of Navarre (Spain) under the “Industrial Ph.D. 2023” program. Chajira Garrote Achou was employed through a grant for the hiring of industrial Ph.D. candidates (File code: 0011-1408-2023-000025) awarded by the Government of Navarre to PENTABIOL S.L. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Government of Navarre.

Data availability

The datasets supporting the conclusions of this article are included within the article and its additional supplementary file. Further inquiries can be directed to the corresponding author.

Declarations

Competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.