The study, published in Frontiers in Antibiotics, applied response surface methodology using a three-factor, five-level central composite design to optimize supplementation of an African oil bean seed extract broth with glucose, sodium citrate and potassium chloride. Twenty experimental runs were performed for each organism, with bacteriocin activity measured as the zone of inhibition against S. aureus ATCC 19095 using agar well diffusion. The work matters because bacteriocins, ribosomally synthesized antimicrobial peptides produced by lactic acid bacteria, hold generally recognized as safe status and are increasingly sought as natural alternatives to chemical preservatives, particularly in regions where cold chains are discontinuous and synthetic additives are unaffordable or culturally unacceptable.

Bacteriocin biosynthesis in lactic acid bacteria follows primary-metabolite-like kinetics, meaning most production occurs in close association with cellular growth. Yield is therefore subject to the same constraints as biomass formation, including the availability and balance of carbon, nitrogen, phosphorus and trace elements, the pH and buffering capacity of the medium, and the ionic environment. Carbohydrate concentration influences bacteriocin output both as a substrate for biosynthesis and, at high levels, through carbon catabolite repression, a regulatory mechanism that redirects metabolic flux away from secondary peptide synthesis. Organic anions and salts such as citrate and potassium chloride contribute to pH buffering, osmotic balance and transport functions, and have been shown to modulate bacteriocin production in a strain-dependent manner.

Because bacteriocin yield depends on several medium variables simultaneously, and because their effects can be non-linear and interactive, one-factor-at-a-time approaches are at best inefficient and at worst misleading. They cannot capture quadratic optima or factor interactions, and the optimum identified by sequential one-factor experiments is conditional on the starting point. Response surface methodology addresses these limitations by modeling the joint linear, quadratic and interactive effects of several factors using a comparatively small number of designed experiments. The central composite design is widely used because it allows fitting of full second-order polynomial models with relatively few runs and provides good predictions across a continuous experimental region.

The experimental results revealed a striking pattern. For both isolates, glucose exerted a significant negative linear effect, and all three supplements exerted significant negative quadratic effects, while no two-factor interaction reached statistical significance, indicating an additive response. The response surfaces are dome-shaped, with interior optima rather than boundary maxima. The desirability-based optima were 6.35 grams per liter glucose, 0.55 grams per liter sodium citrate and 1.78 grams per liter potassium chloride for L. lactis, and 6.44, 0.55 and 1.59 grams per liter respectively for L. fermentum. The two strains differed quantitatively: L. fermentum showed stronger quadratic curvature in glucose and potassium chloride, while L. lactis showed stronger curvature in sodium citrate, implying that deviations from each strain’s optimum carry different penalties.

The biological interpretation of the glucose optimum is particularly interesting. Bacteriocin yield rises to a maximum at approximately 6.4 grams per liter glucose and falls at both lower and higher levels. This pattern is consistent with a carbon-limited rather than carbon-rich optimum. An excess of readily fermentable carbohydrate provokes carbon catabolite repression in lactic acid bacteria, redirecting flux toward growth and acidification at the expense of peptide synthesis. Intense acidification at high glucose can also carry the pH below the range optimal for bacteriocin secretion or stability, although the researchers note that final pH was not measured, making this a plausible hypothesis rather than a demonstrated cause. At very low glucose, carbon and energy become limiting for biomass and therefore for bacteriocin, generating the rising left-hand limb of the quadratic curve.

Sodium citrate was not significant as a linear factor but exerted a strongly significant negative quadratic effect in both models, indicating an intermediate optimum near 0.55 grams per liter. As a weak organic-acid buffer, citrate moderates the rapid acidification produced by lactic acid bacteria growth and extends the period over which the broth remains in the pH range conducive to bacteriocin synthesis. Too little citrate provides insufficient buffering, while too much can shift the pH above the optimum and impose ionic stress. Citrate also chelates divalent cations such as magnesium and manganese, which are cofactors for enzymes involved in central metabolism, potentially limiting enzyme activity at high concentrations. The fact that the optimum citrate concentration was identical for both organisms suggests the dominant mechanism is a relatively non-specific one, most plausibly buffering.

Potassium is the major intracellular cation in bacterial cells and is required for osmotic balance, membrane potential and the activity of several enzymes. At too low a concentration, intracellular potassium homeostasis cannot be maintained; at too high a concentration, ionic strength rises and osmotic stress depresses growth and metabolism. The negative quadratic coefficient of potassium chloride in both models is consistent with this dual-sided dependence. The optimum potassium chloride level was slightly higher for L. lactis than for L. fermentum, in keeping with the broader observation that lactococci are generally more tolerant of higher salt levels than heterofermentative lactobacilli.

However, the study is notable for its unusual statistical transparency. The fitted second-order models accounted for 71.85 and 73.15 percent of the variation, but the overall regression was only marginally significant, with p-values of 0.060 and 0.050. More strikingly, the predicted R-squared was 0.00 percent in both cases, and the lack-of-fit test was highly significant, meaning the polynomials have no demonstrated ability to predict the response at new, un-run factor combinations. The researchers therefore interpret their optima as an estimated optimal region rather than a quantitatively predictive optimum. A cross-validated re-analysis with machine-learning regressors, including Gaussian-process regression, support-vector regression and ensemble methods, confirmed that the polynomial has negative cross-validated Q-squared values, while flexible models recovered a modest predictive signal of approximately 0.5 to 0.6, though the optima located by these models were unstable across model families.

Validation fermentations carried out at the supplement optima combined with the previously established optimal culture conditions of 1.9 percent sodium chloride, 31 degrees Celsius and initial pH 6.0 produced zones of 15.22 and 14.96 millimeters, exceeding both the supplementation-only model estimates and the prior culture-conditions outcomes. The researchers caution that these percentage gains were not tested statistically and that part of the excess reflects the fixed sodium chloride, which was optimized separately and absent from the supplement design. The two optimization layers are therefore interpreted as combining additively rather than demonstrating established synergy. Within these limitations, the work demonstrates that moderate, balanced supplementation with inexpensive, food-grade additives can substantially enhance bacteriocin yield, supporting further development of these strains as starter or protective cultures for controlled fermentation of the African oil bean seed and potentially other protein-rich African fermented foods.

Subject of Research: Medium optimization for bacteriocin production by lactic acid bacteria isolated from African oil bean seed fermentation

Article Title: Statistical optimization of glucose, sodium citrate and potassium chloride supplementation for enhanced bacteriocin production by Lactococcus lactis MT186647 and Lactobacillus fermentum MT186598 isolated from fermenting Pentaclethra macrophylla benth

Article References: Onwuakor, C. E., & Ngene, A. C. (2026). Statistical optimization of glucose, sodium citrate and potassium chloride supplementation for enhanced bacteriocin production by Lactococcus lactis MT186647 and Lactobacillus fermentum MT186598 isolated from fermenting Pentaclethra macrophylla benth. Frontiers in Antibiotics, 5, Article 1907088. https://doi.org/10.3389/frabi.2026.1907088

Image Credits: AI Generated

DOI: 10.3389/frabi.2026.1907088

Keywords: bacteriocin, lactic acid bacteria, response surface methodology, food biopreservation, medium optimization, Pentaclethra macrophylla, starter cultures, Statistical, optimization, glucose, sodium, citrate