Four bacteriocinogenic bacterial strains, P. polymyxa OSY–EC, Bacillus velezensis GF610, B. velezensis CE2, and B. velezensis OSY–S3 (Table 1) were screened in the current work for ability to produce antimicrobials in acid whey. These strains were transferred from their stock cultures into tryptic soy broth (TSB; Becton, Dickinson and Company, Franklin Lakes, NJ, United States) and incubated for 24 h at 37°C before use in various experiments. A panel of bacterial and fungal strains were tested for sensitivity to the antimicrobials produced by a test strain. The targeted bacterial strains were Staphylococcus aureus ATCC 6538, Listeria innocua ATCC 33090, Pseudomonas aeruginosa ATCC 9027, Escherichia coli ATCC 8739, E. coli k12, Klebsiella pneumonia 556, Enterobacter gergoviae 565, Burkholderia cepacia ATCC 25416, and Serratia marcescens 562. The targeted fungal strains were Candia albicans ATCC 10231 and Aspergillus brasiliensis ATCC 16404. The minimum inhibitory concentration (MIC) of the antimicrobials against the targeted strains was determined using the standard CLSI broth microdilution method (Clinical and Laboratory Standards Institute, 2015), as described later. Before use in the MIC experiment, the targeted bacterial strains were grown in TSB for 24 h at 37°C whereas C. albicans and A. brasiliensis were grown in TSB for 2 and 7 days, respectively, at 30°C. L. innocua ATCC 33090, E. coli k12, and C. albicans P57072 were used as indicator strains for anti–Gram positive, anti–Gram negative, and anti–fungal activity (He et al., 2007; Gerst et al., 2015; Hirakawa et al., 2015), respectively. All strains were obtained from the food microbiology laboratory collection, Department of Food Science and Technology, The Ohio State University, Columbus, Ohio, United States.

Acid whey was procured from Superior Dairy Inc. (Canton, OH, United States) as a by–product from the production of cottage and ricotta cheeses. The composition of the acid whey was determined as follows. Lactose content was determined by an enzymatic “low-lactose determination” method (Mangan et al., 2019) using a commercial kit (Megazyme, Wicklow, Ireland). The protein content was determined by a dye-binding method (AOAC 967.12–1970, Protein in milk. Dye binding method I) using a protein analyzer (Sprint; CEM Corporation, Matthews, NC, United States). The ash content was determined (AOAC 945.46–1945, Ash of milk. Gravimetric method) using a microwave muffle furnace (CEM Corporation). Calcium concentration was determined by complexometric titration (Kamal, 1960). Fat content was determined as described by Cartwright et al., 2005, using a fat analyzer (SMART Trac II Analyzer; CEM Corporation). Measurement and adjustment of pH were done using pH meter (Mettler-Toledo Co., Columbus, OH, United States) equipped with a glass electrode, and calibrated at pH values of 4.0, 7.0 and 10.0, at 25°C.

Prior to use, acid whey was clarified by centrifugation to remove sediments and heated at 65°C for 30 min to decontaminate the product; no microorganisms were recoverable on microbiological media after heating. The acid whey was modified as follows: (a) pH adjustment to 6 or 7 with 5 N NaOH (Fisher Scientific, Fair Lawn, NJ, United States), (b) addition of ultra-high temperature (UHT) skim milk (1% w/v), and (c) supplementation with yeast extract (YE; Becton, Dickinson and Company) at 1% w/v level. Yeast extract was dissolved in water and autoclaved prior to addition to the whey.

The four selected antimicrobial producers were cultured at 1% (v/v) in 20 ml of the modified acid whey as shown in Table 2. The incubation was completed in 125–ml baffled Erlenmeyer flasks agitated at 150 revolutions/min (RPM) in a shaking incubator (New Brunswick Scientific, Edison, NJ, United States) at 30°C for 24 or 48 h. After incubation, cells of the producer bacteria were removed by centrifugation at 7,000 × g and 4°C for 20 min and the supernatant was collected and stored at 4°C until being used for antimicrobial testing, using the antimicrobial bioassay technique described later. Based on the antimicrobial activity assessment (Table 2), the highest antimicrobial producer in modified acid whey, under given bioreaction conditions, was selected for scaling up the antimicrobial production.

According to the screening results (Table 2), P. polymyxa OSY–EC was selected for scaled–up production. The P. polymyxa OSY–EC strain is a mutant of the parent P. polymyxa OSY–DF (Campbell et al., 2021) from which paenibacillin and polymyxin were isolated previously (He et al., 2007) and a fusaricidin gene cluster was also detected (Huang and Yousef, 2012). A process, for simultaneous microbial growth and concentration of the produced antimicrobials, was developed as illustrated in Figure 1. Briefly, acid whey medium was ultrafiltered using 800 kDa membrane (Membrane Specialists, Hamilton, OH, United States) to remove sediments. The acid whey permeate (AWP) was modified by neutralization to pH 7.0 with 5 N NaOH, and supplementation with yeast extract (1%; wt/v). The modified acid whey was then pasteurized at 65°C for 30 min. The pasteurized acid whey–based medium was poured into four sterile 2-L flasks with a final content of 800 ml per flask. This was followed by inoculating P. polymyxa OSY–EC culture, at 1% (v/v) level, and incubating at 30°C for 48 h with shaking at 120 rpm. After the first 24 h of incubation, beads of a polymeric adsorbent resin (Amberlite XAD7HP; Sigma Aldrich, St. Louis, MO, United States), pasteurized at 65°C for 30 min, were added at 10% (wt/v) to the cultured acid whey to allow binding of antimicrobials with the resin. After the conclusion of incubation (a total of 48 h), the beads of the polymeric resin in each flask were collected and washed with 1 L of sterile deionized water, twice; This was followed by washing once with 500 ml of 30% ethanol. To elute the bound antimicrobials, the washed resin was resuspended in 250 ml of acidified 70% ethanol (pH 2.5) and the resin suspension was incubated at 22–25°C for 4 h with shaking at 130 rpm. After incubation, the acidified solvent containing the antimicrobials was collected, and then the antimicrobials were concentrated by evaporating the solvent (ethanol) using a vacuum concentrator (Automatic Environmental SpeedVac^® System, Savant, Hyannis, MA, United States) to yield semi–purified antimicrobial extract. The latter was dried using a freezer dryer (Labconco, Kansas, MO, United States) and the resulting freeze–dried antimicrobials-rich powder (designated as OSY–EC AMRP) was used throughout the study. After suspension in water, the MIC of the AMRP was determined against the bacterial and fungal strains (Table 3) using the broth microdilution method as described in later section.

Antimicrobial bioassay was completed in following media: (i) culture supernatants produced during the initial screening for antimicrobial production in acid whey-based media, and (ii) the semi–purified antimicrobial extract produced during the scaled–up bioreactions (before drying). Aliquots (10 µl) of these media were spotted onto tryptic soy agar (TSA) plates overlaid with soft agar medium (TSA containing 0.75% agar) seeded with an indicator strain (L. innocua ATCC 33090, E. coli k12, or C. albicans P57072). The spotted agar was allowed to absorb the inoculum by leaving the plates undisturbed for 60 min before incubation for 24 h at 37°C for the bacterial indicators or for 48 h at 30°C for C. albicans P57072. After incubation, zones of inhibition were observed to determine the antimicrobial activity of spotted aliquots as described in previous studies (He et al., 2007; Campbell et al., 2021). The concentrations of paenibacillin (indicator L. innocua ATCC 33090), polymyxin (indicator E. coli k12), and fusaricidin (indicator C. albicans P57072), in the antimicrobial-containing solutions, were measured in arbitrary units (AU) as follows: A r b i t r a r y   u n i t s ( A U ) / m l = 1 D i l u t i o n   f a c t o r   f o r   l a s t   d i l u t i o n   w i t h   i n h i b i t i o n x 1000 v o l u m e   s p o t t e d ( μ l ) (1)

The AU per gram powder was then calculated from the AU/ml of the solution prepared from that powder.

For the MIC determination, the standard CLSI broth microdilution method (Clinical and Laboratory Standards Institute, 2015) was used with modifications. Two–fold serially–diluted AMRP in TSB was prepared and 50 µl aliquots of these preparations were dispensed into the wells of a sterile 96–well plate (Corning, Tewksbury, MA, United States). Overnight cultures of the test organisms (Table 3) were used and their 10⁻³ dilutions in TSB were prepared. Aliquots (50 µl) of diluted cultures were added to the wells of the sterile 96–well plate. The MIC was determined as the lowest concentration of the AMRP which inhibited the visible growth of bacterial or fungal culture after incubation at 37°C for 24 h (for bacteria) or at 22–25°C for 3–6 days (for fungi).

The antimicrobials in supernatants of cultured acid whey and the semi–purified antimicrobial extract were separated using fast protein liquid chromatographic system (Quest 10 plus system; Bio-Rad, Hercules, CA, United States) equipped with a cationic exchange 10 × 100 mm column (ENrich S; Bio-Rad). The elution solvents used were (A) 5 mM phosphate buffer, pH 7 and (B) 5 mM phosphate buffer, pH 7 and 1 M NaCl; these were applied as a gradient, from 0% to 100% solvent B. One ml of active supernatant of cultured acid whey or semi–purified antimicrobial extract was applied to the column after equilibration. Eluted peptides were detected using a multi-UV detector at a 215 nm wavelength (NGC™ Multi-Wavelength Detector; Bio-Rad). Eluted fractions were collected in 1–ml aliquots with a fraction collector (BioFrac™ Fraction; Bio-Rad). Retention times for peaks corresponding to paenibacillin and to a polymyxin–fusaricidin mixture were determined. Fractions with peaks corresponding to paenibacillin and polymyxin–fusaricidin were collected and antimicrobial activities of these agents were confirmed using the bioassay technique described previously (He et al., 2007; Campbell et al., 2021). The indicator strains used were L. innocua ATCC 33090, E. coli k12, and C. albicans P57072 for paenibacillin, polymyxin, and fusaricidin, respectively. The identities of these fractions were confirmed using matrix-assisted laser desorption/ionization time of flight mass spectrometer (MALDI-TOF MS) (UltrafleXtreme; Bruker Daltonics, Billerica, MA, United States) as described previously (Hussein et al., 2020).

For the mechanistic studies, L. innocua ATCC 33090 (paenibacillin–sensitive) and E. coli k12 (polymyxin–sensitive) were used for studying the antimicrobial mode of action of OSY–EC AMRP produced in the acid whey–based medium. The OSY–EC AMRP contains paenibacillin and polymyxin which are cytoplasmic–membrane active agents. Therefore, experiments to detect leakage of intracellular potassium ions, disruption of outer membrane, and perturbation of inner membrane, were conducted as follows.

Data were presented as mean ± standard deviation of two biological repeats, and each repeat was performed in triplicate. A statistical software (GraphPad Prism 9.0.0; GraphPad software, San Diego, CA, United States) was used to perform analysis of variance (ANOVA) with Tukey pairwise comparisons to determine significant differences between treatment groups or comparing pairs of treatment. Statistical significance was determined at p value <0.05.

Unmodified or modified acid whey was used as a growth medium for four bacterial strains known to produce antimicrobial metabolites (Table 2). The unmodified acid whey contained 3.97% lactose, 0.62% protein, 0.5% ash, 0.15% Ca ²⁺ and 0.07% fat; initial pH was 4.6. Using the unmodified acid whey product, growth of the tested antimicrobial-producing bacterial strains was undetectable; hence, no antimicrobial metabolites were detected. Therefore, acid whey neutralization to pH 6–7 was necessary to enable the growth of the antimicrobial producers. With yeast extract supplementation (1% wt/vol) in neutralized acid whey, B. velezensis GF610 and P. polymyxa OSY–EC strains exhibited considerable activity against the Gram-positive indicator bacterium, L. innocua ATCC 33090, after 24 h and 48 h of incubation. P. polymyxa OSY–EC was the most capable antimicrobials producer in acid whey, compared to other tested strains (Table 2). P. polymyxa OSY–EC produced the highest activity against both the Gram-positive and Gram-negative indicator bacteria. Based on this screening step, optimized condition for production of antimicrobial peptides in acid whey consisted of supplementation with yeast extract (1%), neutralization to pH 7, and incubation for 48 h at 30°C. P. polymyxa OSY–EC was selected as the antimicrobial producer strain in an acid whey–based medium for the remainder of the study. Antifungal activity (Table 3) was only identified in acid whey inoculated with OSY–EC strain, but this activity was detectable only after application of the concentration procedures shown in Figure 1.

An efficient method that allowed production and separation of antimicrobials simultaneously was developed (Figure 1). In this method, the transition from small scale (20 ml; Table 2) to an 800–ml reaction was achieved. Beads of amberlite-based resin, XAD–7HP, were added to the bioreaction after the first 24 h of incubation to effectively capture the antimicrobials being released by the producer during the remaining 24 h of incubation. Addition of the polymeric beads did not affect the growth of the producer strains nor inhibit the antimicrobial production. The final OSY–EC antimicrobials-rich powder contained 512,000, 32,000 and 8,000 arbitrary unit (AU)/g paenibacillin, polymyxin E, and fusaricidin, as assayed against L. innocua ATCC 33090, E. coli k12 and C. albicans P57072, respectively. Control experiment was completed in which no beads were added during incubation of P. polymyxa OSY–EC for 48 h in the acid whey-based medium. In this experiment, bioreaction product was centrifuged and the antimicrobials in the culture supernatant were extracted with the polymeric beads, eluted with acidified ethanol, concentrated under vacuum, and freeze-dried. After antimicrobial bioassay using L. innocua ATCC 33090, E. coli k12 and C. albicans P57072, the resulting powder yielded 10,200, 51,200 and 4,000 AU/g paenibacillin, polymyxin E, and fusaricidin, respectively. The recovered activities, except that for polymyxin E, were considerably smaller than those observed when the beads were added during the last 24 h of incubation.

The MIC of the OSY–EC AMRP produced during this large scale bioreaction was determined against selected strains of known problematic bacteria and fungi (Table 3). The OSY–EC AMRP showed potent antimicrobial effect against most of the targeted strains; however, high MIC values were observed in case of S. marcescens (6,200 μg ml⁻¹) and B. cepacia (25,000 μg ml⁻¹). It is worth noting that 10–µl aliquots of the resuspended OSY–EC AMRP (50 mg ml⁻¹) inhibited the mycelial growth of A. brasiliensis on an agar medium (Figure 2). It should be cautioned that the OSY–EC AMRP contain three antimicrobial peptides, which could synergistically inhibit the growth of the tested strains.

Liquid chromatographic analyses were performed to confirm the causes of antimicrobial activities exhibited by the extracts of P. polymyxa OSY–EC culture in the modified acid whey. This also helped exclude degradation products of whey components as a source for these antimicrobials. For example, hydrolysis of milk lactoferrin, could lead to the production of lactoferricin, which is an antimicrobial peptide (Gifford et al., 2005). Chromatographic analysis of the antimicrobials produced in modified acid whey by P. polymyxa OSY–EC (Figure 3) revealed peaks at known retention times for paenibacillin and for a polymyxin–fusaricidin mixture (Campbell, 2020). The presumed paenibacillin fraction exhibited anti–Gram positive activity (detected by L. innocua ATCC 33090) and the polymyxin–fusaricidin fraction produced anti–Gram negative activity (detected by E. coli k12) but no antifungal activity was detected. The antifungal activity due to fusaricidin was detected only after concentration of OSY–EC extracts using the polymeric beads. MALDI-TOF MS analysis confirmed the peaks of singly charged ions corresponding to paenibacillin, polymyxin, and fusaricidin, with m/z values of 3,007.029 [M + Na]⁺, 1,191.784 [M + Na]⁺, and 883.605 [M + H]⁺, respectively (Figure 4). In conclusion, three antimicrobials, produced in the modified acid whey media by P. polymyxa OSY–EC, were identified and found to contribute to the observed antimicrobial activity of the product.

Cell suspension of L. innocua ATCC 33090 or E. coli k12 in TSB was treated with 4x–MIC level of OSY–EC AMRP and the surviving populations during the 4–h incubation at 37°C were presented in Figure 5. Compared to the untreated control, the populations of the two indicators decreased by >2 log CFU ml⁻¹ after 1 h of the treatment whereas >4 log CFU ml⁻¹ population reduction was observed after 4 h of treatment of the two indicator strains. The successful inactivation of the two indicator bacteria by OSY–EC AMRP in TSB prompted us to explore the inactivation pattern in a food matrix. Milk was selected as a model food considering its susceptibility to contamination and ability to support growth of many pathogenic and spoilage microorganisms. The OSY–EC AMRP at 2x–MIC level was incorporated into sterile milk, inoculated with ∼ 10⁵ CFU ml⁻¹ of L. innocua or E. coli k12, and the inoculated milk was held at 4°C for 3 days. For L. innocua ATCC 33090, the population of the bacterium decreased by >2 log CFU ml⁻¹ (p < 0.05) compared to the untreated control after 1–day incubation (Figure 6A). After 3 days of incubation, L. innocua population in the OSY–EC AMRP–treated milk was lower by >3.5 log CFU ml⁻¹ (p < 0.05), compared to the untreated control. In contrast, E. coli k12 population in OSY–EC AMRP–treated milk was below the detection limit (<1 log CFU ml⁻¹) after 1 day of incubation and this inactivation level remained until the 3^rd day of incubation (Figure 6B). The susceptibility of E. coli k12 to OSY–EC AMRP is more profound than that of L. innocua. These results demonstrated the potent bactericidal activity of the OSY–EC AMRP against Gram-positive and Gram-negative bacteria in the model food.

Paenibacillin and polymyxin are the key antimicrobials in OSY–EC AMRP and both are known cytoplasmic membrane–active agents. Demonstrating the action of OSY–EC AMRP against bacterial cytoplasmic membranes further confirms the contribution of paenibacillin and polymyxin to the antimicrobial activity of the powder. Therefore, a panel of membrane perturbation assays was conducted to elucidate the antimicrobial mechanism of action of OSY–EC AMRP, which contained 512,000, or 32,000 AU/g of paenibacillin and polymyxin, respectively, against the indicator strains used in the current study. The MIC values for OSY–EC AMRP were 200 and 400 μg ml⁻¹ for E. coli k12 and L. innocua ATCC 33090, respectively (Table 3). The addition of OSY–EC AMRP at various levels, from sublethal to lethal concentrations, resulted in rapid leakage of intracellular potassium ions from cells of L. innocua ATCC 33090 and E. coli k12 compared to the untreated control (0 μg ml⁻¹ OSY–EC AMRP; p < 0.05) (Figure 7). The leakage of potassium ions was a concentration–dependent (p < 0.05) and similar trend was observed for the two indicator strains in response to the OSY–EC AMRP treatment.

For the outer membrane disruption, it was obvious that the NPN dye produced strong fluorescence signals which significantly (p < 0.05) depended on the concentration of OSY–EC AMRP (Figure 8). This increased fluorescence is attributed to the uptake of NPN by the disrupted outer membrane of E. coli k12 (Figure 8). Similarly, the cytoplasmic membrane permeability was compromised by the OSY–EC AMRP (Figure 9); this was evident by the fluorescence signals of Sytox™ Green dye which increased proportionally (p < 0.05) to the concentration of OSY–EC AMRP added to the cell suspension of the indicator strains. Thus, cell membranes of the two indicator strains were apparently disrupted by the OSY–EC AMRP. Collectively, these results proved that OSY–EC AMRP caused damages to membranes of targeted cells owing to the membrane–active peptides, paenibacillin (Anti–L. innocua) and polymyxin E (anti–E. coli).