Development of novel microbial synthetic consortia for the production of antimicrobial fermentates containing caproate

TL;DR

This study designed a microbial consortium to produce caproate, a natural antimicrobial, which showed strong activity against bacteria, yeast, and mold in food applications.

Contribution

A novel multi-kingdom microbial consortium was developed to produce caproate-containing fermentates with antimicrobial properties.

Findings

Caproate levels reached 63.9 mM in a 22-day bioprocess with ethanol addition.

Fermentates inhibited bacteria, yeast, and mold growth, especially at pH 4.5.

Lactate and caproate acted synergistically in minced meat to enhance antimicrobial activity.

Abstract

Short- and medium chain carboxylic acids (SCCA/MCCA) are natural antimicrobials produced by fermentation and chain elongation, but currently only a few SCCA/MCCA are used in food-related applications. With the aim to diversify the SCCA/MCCA profile of fermentates for biopreservation, we designed bioprocesses employing bacterial or multi-kingdom consortia to produce caproate-containing fermentates using a targeted cross-feeding strategy. We combined Limosilactobacillus reuteri, Clostridium kluyveri, and Saccharomyces cerevisiae, and quantified substrates utilization and metabolites. The antimicrobial activity of SCCA/MCCA and fermentates was analysed in vitro and in a meat model system. In a first bioprocess, the addition of ethanol (EtOH) initiated caproate formation by the bacterial consortium with levels of 63.9 mM (7.4 g L−1) after 22 days. Next, we run two shorter bioprocesses (12 days) and more caproate (28.9 mM, 3.4 g L−1) was produced if EtOH was regularly added than without EtOH supplementation (21.7 mM, 2.5 g L−1). When S. cerevisiae was included, 37.9 ± 11.4 mM (4.4 g L−1) caproate was formed without EtOH addition. Beyond the intended cross-feeding activity, consortia produced and re-metabolised mannitol and glycerol. Lactate and caproate were the major carboxylic acids in fermentates. Caproate inhibited bacteria, yeast and molds at pH 4.5 and 6.5 in vitro, while lactate and caproate acted synergistically in minced meat. Fermentates conferred antimicrobial activity against indicator microbes mostly at pH 4.5 and to a lesser extent at pH 6.5. In this study we successfully designed a consortium to produce caproate. We provide evidence that caproate-containing fermentates had strong antimicrobial activity, and our results indicate the complexity of metabolic interactions that occurred within consortium members.

• Design of a self-contained bioprocess with multi-kingdom consortium.

• Cross-feeding of bacteria and yeast to produce caproate fermentates.

• Fermentates containing caproate inhibited bacteria, yeast and mold growth.

The online version contains supplementary material available at 10.1007/s00253-026-13718-z.

Figures

Substrate utilization and metabolite formation of bacterial consortium bioprocess MF0. The bioprocess MF0 was conducted at anaerobic conditions with YC-GF medium at pH 6.7, 37 °C, 0.5 Hz for 22 days*. L. reuteri* and *C. kluyveri* were inoculated on day 0 and day 1, respectively. On day 6, EtOH was added, and from day 9, EtOH was regularly supplied to the bioreactor. Substrate utilization and metabolite formation was determined with HPLC-RI (sugars, alcohols and SCCA) and UPLC-DAD (caproate). (A) Glucose and fructose utilization and mannitol and EtOH formation. (B) Formation of SCCA and MCCA

Substrate utilization and metabolite formation in short bacterial or multi-kingdom consortia bioprocesses. Bacteria bioprocesses MF1-1 (A-C) and MF1-2 (D-F) were conducted at anaerobic conditions in YC-GF at pH 6.7, 37 °C, 0.5 Hz for 12 days. Fermentations MF1-1 and MF1-2 were started with *L. reuteri* at day 1, *C. kluyveri* was added on day 2. EtOH was regularly supplied to MF1-1 starting on day 2. Multi-kingdom consortia bioprocesses MF2-1, MF2-2 and MF2-3 (G-I) were initiated with *L. reuteri* and *S. cerevisiae* on day 0 and *C. kluyveri* was added on day 2. Bioreactors were run for 12 da

Glycerol formation and utilization by multi-kingdom consortia. Glycerol and 1,3-propanediol levels were determined in fermentates of multi-kingdom consortia of bioprocesses MF2 (A) and in static Hungate tube incubations of co-cultures of *S. cerevisiae* and *L. reuteri* (B) in YC medium supplied with glucose and fructose that were incubated at 37 C for 3–4 d. Substrate and metabolite levels including 1,3-propanediol and glycerol were determined with HPLC-RI and NMR

Mannitol utilization by *C. kluyveri* and *L. reuteri.* (A) *C. kluyveri* was grown in a bioreactor supplied with YC-mannitol (YC-M, solid line) YC-M containing acetate and ethanol (dashed line) (B) *L. reuteri* was cultivated in Hungate tubes with YC-M. Samples were collected in regular intervals to determine substrate utilization and metabolite formation

Schematic overview of metabolic interactions of synthetic microbial consortia. Proposed interactions of the bacterial consortium, and with the addition of *S. cerevisiae* (in blue) of the multi-kingdom consortium. ‘?’ Possible interactions

Full text

Introduction

Every year, approximately 1.05 billion tons of food are wasted globally, corresponding to about one-fifth of the food available for human consumption (World Health Organization 2006). During processing, production, transport, retail and domestic preparation, food is lost due to damage, microbial spoilage and mishandling (Girotto et al. 2015). To reduce food waste and deliver safe products, there is an urgent need to identify natural antimicrobial compounds that replace chemical food preservatives. Short-chain carboxylic acids (SCCA) produced by fermentative processes such as acetate, citrate, lactate, formate, and propionate and their salts are used as antimicrobials to reduce food contamination and for the disinfection of contact surfaces and food manufacturing equipment (Ng et al. 2023). For example, lactate is a major contributor to the pH lowering preservative effect of food fermented by Lactobacillaceae, e.g. fermented dairy, vegetables, and cereals. Most SCCA are only active at acidic pH (Ng et al. 2024), and a major challenge lies in finding SCCA that confer antimicrobial activity against foodborne pathogens and spoilage organisms also at neutral or alkaline conditions to broaden application possibilities beyond acidic fermentations.

In a recent study, Ng et al. (2024) compared the antimicrobial activity of 21 compositionally and structurally different SCCA against three strains of Salmonella enterica serovars Typhimurium and Enteritidis and included the medium-chain carboxylic acid (MCCA) valerate and caproate. Among the tested compounds, caproate conferred the strongest antimicrobial and antibiofilm activity at pH 5.5 and 6.5. Caproate can be produced by some species within the genus Clostridium (Yin et al. 2017). Very few studies reported the formation of caproate in food and beverages, for example in some Chinese liquor fermented with glucose and lactose (Wang et al. 2022a, b & 2022). The probably best-studied caproate producer, Clostridium kluyveri, produces caproate in the presence of ethanol (EtOH) and acetate through β-oxidation–dependent chain elongation (Yin et al. 2017). C. kluyveri converts EtOH to acetyl-CoA, a proportion of acetyl-CoA is oxidised to acetate for ATP-generation, while the remaining acetyl-CoA are combined with acetyl-CoA cycled in the reverse β-oxidation cycle to produce butyryl-CoA (Angenent et al. 2016). This butyryl-CoA can either be converted to butyrate or undergo another round of chain elongation with acetyl-CoA to form hexanoyl-CoA, which can be reduced to caproate (Angenent et al. 2016). Both key substrates, EtOH and acetate, can be delivered by fermentative systems through cross-feeding. In an open culture with complex communities of ruminal microbes, C. kluyveri produced 40–53 mM (4.6–6.2 g L^−1^) caproate (Kenealy et al. 1995, Weimer et al. 2015), while 29 mM (3.4 g L^−1^) caproate was formed in an anaerobic bioreactor that contained C. kluyveri, and was fed with corn-fermentation derived beer (Ge et al. 2015). These studies confirm the feasibility of producing caproate with mixed microbial communities*,* but such complex microbial systems are difficult to control due to possible fluctuation of substrates, complexity of the microbiota, and competition of metabolic pathways (Dong et al. 2023). Therefore, we aimed to design minimal synthetic consortia to produce caproate through cross-feeding strategies that involve fermentation. To deliver acetate and EtOH for caproate production of C. kluyveri, we selected the food-grade heterofermentative Limosilactobacillus reuteri. From 1 M of glucose, L. reuteri theoretically produces 1 M lactate, 1 M EtOH and 1 M CO_2_. In addition, 2 M acetate are formed, if 1 M fructose is reduced to 1 M mannitol (Gaenzle 2015). We performed co-fermentations of C. kluyveri DSM 556 and L. reuteri FMT 421 in the presence of glucose and fructose. As we observed that the availability of EtOH was a decisive factor of caproate formation, we also investigated the performance of a multi-kingdom consortium adding the yeast Saccharomyces cerevisiae FMT 3004 to L. reuteri and C. kluyveri to achieve a higher concentration of EtOH in the fermentative system. Finally, we tested the antimicrobial activity of lactate, caproate and caproate-containing fermentates against bacteria, yeast and mold indicators at pH 4.5 and 6.5 in vitro and in situ in a minced meat model system.

Materials and methods

Strains used and cultivation conditions

Limosilactobacillus reuteri FMT 421, Bacillus subtilis FMT 429 and FMT 431, Pantoea sp. FMT 32*, Pantoea ananatis* FMT 33*, Kosakonia cowanii* FMT 34, Pichia kluyveri FMT 3000, S. cerevisiae FMT 3003 and FMT 3004 were obtained from the strain collection of the Functional Microbe Technology group (Department of Biological and Chemical Engineering, Aarhus University). Salmonella enterica serovar Typhimurium DSM 17058, Escherichia coli DSM 19206, Klebsiella oxytoca DSM 5115, Listeria innocua DSM 20649*, Staphylococcus aureus* DSM 1104, C. kluyveri DSM 556, Candida albicans DSM 10697 and Penicillium purpurogenum DSM 21170 were purchased from the German Collection of Microorganisms and Cell Cultures GmbH (DSMZ). Salmonella enterica serovar Enteritidis CCM 4420 was bought from the Czech Collection of Microorganisms (Masaryk University). Aspergillus niger BCE M1 Penicillium roqueforti BCE M2 (originally purchased from Dansk Hjemmeproduktion (lot. no. 05110069) were obtained from the mold collection at the Department of Biological and Chemical Engineering at Aarhus University. Working cultures of bacteria, yeast and mold strains used in this study were prepared as outlined in Table 1. Table 1. Cultivation conditions of bacteria, yeast and molds strains used in this studyMicroorganismStock cultureFirst activationPropagationLast activation stepL. reuteri,**L. innocua,**S. aureusFrozen (−80 °C) in glycerol (25%, v/v)WCSP agar, 2 d at 37 °C anaerobicallyOne colony to WCSP broth, overnight at 37 °C anaerobically100 µL from previous step to fresh WCSP or mMRS broth, overnight at 37 °C anaerobicallyS. enterica,**E. coliFrozen (−80 °C) in glycerol (25%, v/v)LB agar, 2 d at 37 °C anaerobicallyOne colony to LB broth, overnight at 37 °C anaerobically100 µL from previous step to fresh LB broth, overnight at 37 °C anaerobicallyK. cowanii,**P. ananatis,**Pantoea sp.,**B. subtilis,**K. oxytocaFrozen (−80 °C) in glycerol (25%, v/v)LB agar, 2 d at 30 °C aerobicallyOne colony to LB broth, overnight at 37 °C aerobically100 µL from previous step to fresh LB broth, overnight at 37 °C aerobicallyC. kluyveriFrozen (−80 °C) in glycerol (25%, v/v)mWCSP broth, 3–4 d at 37 °C anaerobically100 µL from previous step to fresh mWCSP broth, 3–4 d at 37 °C anaerobically100 µL from previous step to fresh mWCSP broth, 3–4 d at 37 °C anaerobicallyS. cerevisiae,**C. albicans,**P. kluyveriFrozen (−80 °C) in glycerol (25%, v/v)YPD agar, 2–3 d at 30 °C aerobicallyOne colony to YPD broth, overnight at 30 °C aerobically100 µL from previous step to fresh YPD broth, overnight at 30 °C aerobicallyA. niger,P. roquefortiP. purpurogenumInoculated (stock) agar plates (4 °C)PDA, 7 d at 25 °C aerobicallyHarvesting spores by adding water to the surface of platesCounting spores and storage at 4 °CWCSP: Wilkins Chalgren anaerobic mediamWCSP: modified Wilkins Chalgren anaerobic broth supplied with potassium acetate (10.0 g/L) and EtOH 96% (20.0 mL/L)LB: Luria Bertani brothYPD: Yeast Extract–Peptone–Dextrose (10 g L^−1^ yeast extract, 20 g L^−1^ peptone, and 20 g L^−1^ glucose)PB: potassium phosphate buffer (50 mM, pH 6.5)PDA: potato dextrose agar

Bioreactor processes

The fermentations were carried out in MINIFOR (MF) bioreactors (LAMBDA Laboratory Instruments) with 400 mL of working volume, automatic control of temperature and pH by infrared radiation heater and peristaltic pumps that regulate the addition of NaOH (1 M) and H_2_SO_4_ (1 M) for pH control. Reactors were filled with 200 mL medium (headspace > 200 mL). Anaerobic conditions were generated by sparging with CO_2_ into the medium. Oxygen levels were monitored. The inflow of CO_2_ was 0.05–0.15 L min^−1^ and the vertical agitation frequency was set at 0.5 Hz. The LAMBDA system uses a non-rotational agitation system with up and down movement. The temperature was regulated at 37 °C and the pH was maintained at pH 6.8 to provide optimal conditions for caproate production (Ge et al. 2015; Kenealy et al 1995). The system was controlled by SIAMsoftware (LAMBDA). No antifoam was used.

We initiated our experiment MF0 using bacteria co-cultures of C. kluyveri DSM 556 and L. reuteri FMT 421. In the first run MF0, bioreactors were filled with 200 mL of yeast casitone broth (YC) (Buljubašić et al. 2024) supplemented with 115 mM glucose and 71 mM fructose (YC-GF). L. reuteri was inoculated at the beginning of fermentation (2% vol/vol), and after 24 h, C. kluyveri was added at 4–5% vol/vol. As we did not observe caproate formation, on day 6, EtOH (96%) was added in a single injection after collecting effluent for further analysis. From day 9, EtOH was regularly supplied to the bioreactor. The pump was calibrated to add EtOH for 1 min every 6 h. The run lasted for 22 days.

Next, we conducted two shorter (12 days) bioreactor runs, MF1-1 and MF1-2, with C. kluyveri DSM 556 and L. reuteri FMT 421. Bioreactors were filled with 200 mL of YC-GF (188–218 mM glucose, 53–64 mM fructose). L. reuteri was inoculated at the beginning of fermentation (2% vol/vol), and after 48 h, C. kluyveri was added at 4–5% vol/vol. A peristaltic pump was calibrated and configurated with the SIAM software to supply 96% EtOH for 1 min every 6 h right after the inoculation with C. kluyveri.

For cultivation of the multi-kingdom consortium (L. reuteri + S. cerevisiae + C. kluyveri), bioreactors were filled with 200 mL YC-GF medium (131–171 mM glucose and 38–52 mM fructose). L. reuteri (2% vol/vol) and S. cerevisiae (5% vol/vol) were inoculated at day 0. C. kluyveri was added (4–5% vol/vol) at day 2. The bioreactor was run for 12 days for MF2 bioprocesses. We conducted three runs with the multi-kingdom consortium (MF2-1, MF2-2 and MF2-3).

From all bioreactor runs, samples (3–5 mL, 1.5–2.5% of the volume) were taken in regular intervals, centrifuged (10.000 rpm, 5 min at room temperature) to collect supernatants for substrate utilization and metabolite formation, and biomass for DNA isolation. Samples were stored at −20 °C. When the runs (MF1-1, MF1-2, MF2-1, MF2-2, MF2-3) were stopped at day 12, the remaining content of the bioreactors was centrifuged and the cell-free fermentates were saved at −20 °C until further antimicrobial activity tests.

We tested the potential of C. kluyveri to use mannitol in bioreactors using YC medium (200 mL) supplied with around 40 mM mannitol (YC-M), or YC-M with acetate (77 mM) and ethanol (259 mM). The bioreactors were inoculated with 4% C. kluyveri. The duration of the fermentation was 8 days, and the samples were collected daily for analysis (details in Supplementary Methods).

Single and co-culture incubations in Hungate tubes

We investigated whether S. cerevisiae produces glycerol under osmotic stress caused by high substrate concentration, and possible glycerol cross-feeding within consortium members in Hungate tubes with YC-GF medium. S. cerevisiae was inoculated alone (2% vol/vol) or in combination with L. reuteri (2%). Tubes were incubated at 37 °C, turbidity was determined using a McFarland densiometer. To determine mannitol utilization of L. reuteri FMT 421, additional static incubations were conducted in Hungate tubes supplied with YC-M (50 mM mannitol). Hungate tubes were inoculated with L. reuteri overnight culture (10% vol/vol) that was grown in modified MRS containing maltose (29 mM), fructose and glucose (both 28 mM) to initiate mannitol metabolism (mMRS-F-G). Strains were incubated for up to 3–4 d at 37 °C. From all incubations, samples were collected in regular intervals, separated into cell pellet and supernatant by centrifugation (10.000 rpm, 5 min at room temperature) and stored at −20 °C until further analysis.

Analysis of substrate utilization and metabolite formation

Cell-free fermentates were diluted in 5 mM H_2_SO_4_, and sugar consumption and metabolite production including lactate, formate, acetate, propionate, butyrate and valerate was determined by high performance liquid chromatography with refractive index detector (HPLC-RI) using a 1220 Infinity II LC System (Agilent). The HPLC-RI was equipped with a Hi-Plex-H column (300 mm × 7.7 mm internal diameter × 8 μm particle size) (Agilent) and 5 mM H_2_SO_4_ was used as mobile phase at a flow rate of 0.6 mL min^−1^ at 40 °C. The compounds of interest were quantified using external standards. Since caproate could not be detected using the HPLC-RI method, we used ultra-performance liquid chromatography with diode array detector (UPLC-DAD) for analysis. The 1260 Infinity II Bio-Inert UPLC system with DAD was equipped with an InfinityLab Poroshell 120 EC-C18 column (3.0 × 50 mm, 2.7 Microns) (Agilent). Ten microliters of the supernatant were injected and samples were eluted with the following gradient of solvents A, 10 mM KH_2_PO_4_ in highly pure water at pH 2.4; and B, acetonitrile, at a flow rate of 0.5 mL min at 40 °C: 0 min 95% A, 5% B; 3 min 95% A, 5% B; 5 min 70% A, 30% B; 7 min 70% A, 30% B; 8 min 40% A, 60% B; 10 min 40% A, 60%B; 11 min 95%A, 5% B; 13 min 95% A, 5% B. The detector wavelength was set at 210 nm.

To validate caproate concentrations obtained by UPLC-DAD and to determine glycerol levels we used proton nuclear magnetic resonance (^1^H-NMR) as described (Huertas-Díaz et al. 2023). Briefly, 200 µL of sodium phosphate buffer (150 mM, pH 7.5) containing 5 mM trimethylsilylpropanoic acid (TSP) and 30% D_2_O were added to 400 µL sample. The measurements were performed using a Bruker Neo-IVDR 600 NMR spectrometer, operating at ^1^H frequency of 600.03 MHz and equipped with a 5 mm ^1^H BBI probe (Bruker BioSpin). All ^1^H spectra were referenced to the TSP signal at 0 ppm. Topspin 4.09 (Bruker Biospin) was used for manual phase and baseline correction. Chenomx NMR Suite 8.6 (Chenomx Inc) was used for signal assignment and metabolite quantification.

DNA isolation and quantitative polymerase chain reaction (qPCR)

The cell pellets taken from the reactors were used for DNA extraction using the GeneJET Genomic DNA Purification Kit with an additional step to include bead beating for cell lysis (40 s at 6 m s-^1^ using a FastPrep, MP Biochemicals). The targeted genes encoded the large subunit of the glycerol/diol dehydratase (pduC) of L. reuteri (Li et al. 2024), the 16S rRNA gene of C. kluyveri (CloKly1F 5′-GAGGAGCAAATCTCAAAAACTGC-3′ and CloKly1R 5′-CCTCCTTGGTTAGACTACGGACTT-3’ (Weimer and Stevenson 2012) and the D1/D2 region of the 26S rRNA gene of S. cerevisiae (26SrRNAF 5’-AGG AGT GCG GTT CTT TG-3', 26SrRNAR 5’-TAC TTA CCG AGG CAA GCT ACA-3’ (Chang et al. 2007). Standards (Table S1) were prepared from tenfold dilution series of linearized plasmids harbouring pduC of L. reuteri, or the purified (Monarch PCR & DNA Cleanup Kit, New England Biolabs) and quantified (Qubit 4 fluorometer, Thermo Fisher Scientific) PCR products for C. kluyveri and S. cerevisiae. qPCR reactions were run in 96-well plates with iTaq Universal SYBR Green Supermix 2 × on a CFX Connect Real-Time PCR Detection System (both Bio-Rad). The program started with denaturation at 95 °C for 3 min followed by 40 cycles of denaturation step (95 °C for 10 s) and annealing (60 °C for 30 s), and a final melting curve analysis. All samples were run as technical duplicates, and each run included a negative control. Due to multiple copies of 16S rRNA and 26S rRNA genes of C. kluyveri and S. cerevisiae respectively, we used correction factors to estimate cell counts based on rRNA gene quantification. According to the reported (mean) numbers of ribosome gene copies, we used n = 7 and n = 92 for C. kluyveri and S. cerevisiae (Sharma et al. 2022 and Stoddard et al. 2015).

Antimicrobial activity of caproate and fermentates

Antimicrobial activity of lactate, caproate and fermentates was determined by two-fold broth dilution method as described (Elvan Gezer et al. 2025; Ng et al. 2024). Briefly, the pH of lactate/caproate stock solutions (100 mM) and fermentates was adjusted to 4.5 or 6.5 with HCl (18.5%) and NaOH (10 M) and solutions were filter sterilized. Stock solutions and fermentates were diluted two-fold in 96-wells microplates with fresh broth using a PIPETMAX 268 liquid-handling robot (Gilson). Working cultures were diluted 10–100 fold in broth and added at 10% vol/vol for a final cell concentration of approximately 5–6 log CFU mL^−1^ for bacteria, 3 log cells mL^−1^ for yeast and 4 log spores mL ^−1^ for molds. Each plate included a positive control (inoculated broth) to confirm that strains grew at the set pH, and negative control (non-inoculated broth). The microplates were incubated for 24 h at 30 °C and 37 °C for yeasts and bacteria, respectively, and the optical density at 600 nm (OD_600_) was recorded using a Tecan Nanoquant Infinite M200 Pro Microplate Reader. For bacteria and yeasts, the minimal inhibitory concentration of lactate, caproate or the dilution factor of fermentates needed to decrease the initial OD_600_ of foodborne pathogens and spoilers to 50% (MIC_50_ or DF_50_) was calculated as described (Elvan Gezer et al. 2025; Ng et al. 2024). Briefly, MIC_50_ or DF_50_ were determined by a four-parameter logistic equation. Based on lactate/caproate levels of fermentates and the DF_50_ values, we calculated the actual levels of lactate and caproate in the diluted fermentates that conferred 50% inhibition. For molds, the microtiter plates were incubated at 25 °C for 2 days, and the MIC and DF were defined as the concentration/dilution factor of fermentates that visibly reduced mold growth compared to the positive control (Elvan Gezer et al. 2025).

Antimicrobial activity of lactate and caproate in minced meat

To establish the antimicrobial efficiency of caproate and lactate in a food system, we tested activity in minced meat. Minced pork (8–12%) or beef (14–18%) purchased from a local grocery store was mixed with 5 mL of SCCA/MCCA solutions (25 mmol kg^−1^, 0.3% caproate, 25 mmol kg^−1^, 0.18% lactate, or 30 mmol kg^−1^ caproate and 90 mmol kg^−1^ lactate) and blended in a stomacher (SmasherTM, AES Laboratories) at fast speed for 180 s. While we used a ratio of lactate and caproate that resembled fermentates in minced beef, we used lower concentration of lactate in minced pork because addition of 90 mM lactate resulted in protein coagulation due to acidification. Samples were vacuum-sealed before storage at 7 °C for 7 days. At day 0, 4 and 7, 90 mL of 0.1% peptone water was added to 10 g of meat samples before blending with stomacher at fast speed for 180 s. Meat suspensions were checked with pH meter (Mettler Toledo), and serially diluted tenfold with 0.1% peptone water, and dilutions (100 µL) were spread on Nutrient agar (Merck) for enumeration of total plate counts. The plates were incubated at 37 °C for 4 days. The experiment was conducted with three independent biological replicates unless otherwise stated.

Statistical analysis

The MIC_50_ and DF_50_ values were calculated by applying a four-parameter logistic equation in GraphPad Prism 10 (Elvan Gezer et al. 2025; Ng et al. 2024). Significant differences in antimicrobial activity of fermentates towards individual indicator strains (DF_50_ values) with pH values adjusted to pH 4.5 and 6.5, and of bacteria and multi-kingdom consortia at pH 4.5 were determined using Student’s t-test. For minced meat experiment, Tukey’s test was used to determine significant differences among different treatments within the same storage day, and among different storage days within the same treatment.

Results

A bacterial consortium consisting of L. reuteri and C. kluyveri produced caproate

To initiate our synthetic consortia, we tested the potential of L. reuteri and C. kluyveri to form caproate in a co-culture (MF0). Strains were cultivated in bioreactors using YC-GF medium. We added the strains sequentially to enable L. reuteri (added at day 0) to form acetate and ethanol (EtOH) before supplementation with C. kluyveri at day 1 (Fig. 1).Fig. 1. Substrate utilization and metabolite formation of bacterial consortium bioprocess MF0. The bioprocess MF0 was conducted at anaerobic conditions with YC-GF medium at pH 6.7, 37 °C, 0.5 Hz for 22 days*. L. reuteri* and C. kluyveri were inoculated on day 0 and day 1, respectively. On day 6, EtOH was added, and from day 9, EtOH was regularly supplied to the bioreactor. Substrate utilization and metabolite formation was determined with HPLC-RI (sugars, alcohols and SCCA) and UPLC-DAD (caproate). (A) Glucose and fructose utilization and mannitol and EtOH formation. (B) Formation of SCCA and MCCA

During the first 24 h of incubation, 33.7 mM of glucose was consumed and 35.8 mM lactate, and 3.9 mM EtOH were produced (Fig. 1). Fructose was converted to mannitol (61.1 mM) after 24 h, and 17.5 mM acetate was formed. Butyrate (6.3 mM) was first detected on day 2 after the supplementation of C. kluyveri. After 5 days of incubation, acetate and EtOH levels were 17.0 and 3.1 mM, respectively with a ratio of EtOH:acetate of 0.18 (Fig. S2). Caproate levels were low (2.4 mM) based on UPLC-DAD and NMR measurements (Fig. 1, Fig. S1).

To increase EtOH levels, we introduced a one-time injection of 96% EtOH to reach a concentration of 42.6 mM on day 6 (Fig. 1), which led to a productivity of 4.0 mM day^−1^ at day 7 (Fig. S2). After supplementation, levels of EtOH and acetate decreased by -Δ38.5 and -Δ22.9 mM, respectively within 24 h, while butyrate levels remained similar. As the addition of EtOH improved caproate formation, we started to regularly provide EtOH to the bioreactor on day 9 leading to the formation of caproate (36.4 mM, 4.2 gL−1) at day 12 and accumulation of 146.8 mM EtOH on day 14. From day 13, acetate was no longer detectable, levels of butyrate decreased to about 8 mM and caproate levels increased to 63.9 mM (day 22, 7.4 g L^−1^) (Fig. 1) with highest productivity at day 13 with 11.2 mM day^−1^ (1.3 g L^−1^ day^−1^) (Fig. S2), showcasing the potential of the L. reuteri/C. kluyveri consortium to produce caproate if ethanol was supplied. The final MF0 fermentate also contained 147.0 mM lactate, 7.4 mM acetate and 8.0 mM butyrate. Propionate and formate were not detected.

Caproate levels were higher if EtOH was provided to the bacterial consortium

As we observed that one-time and/or regular EtOH supplementation supported caproate formation by our bacterial consortium in MF0 (Fig. 1), we conducted two shorter fermentations (MF1-1 and MF1-2) to further investigate the impact of EtOH: MF1-1 with regular addition of EtOH after C. kluyveri supplementation on day 2 (Fig. 2A–C), and MF1-2 without EtOH supplementation (Fig. 2D–F). We determined cell counts of L. reuteri and C. kluyveri with quantitative PCR (Fig. 2C, F).Fig. 2. Substrate utilization and metabolite formation in short bacterial or multi-kingdom consortia bioprocesses. Bacteria bioprocesses MF1-1 (A-C) and MF1-2 (D-F) were conducted at anaerobic conditions in YC-GF at pH 6.7, 37 °C, 0.5 Hz for 12 days. Fermentations MF1-1 and MF1-2 were started with L. reuteri at day 1, C. kluyveri was added on day 2. EtOH was regularly supplied to MF1-1 starting on day 2. Multi-kingdom consortia bioprocesses MF2-1, MF2-2 and MF2-3 (G-I) were initiated with L. reuteri and S. cerevisiae on day 0 and C. kluyveri was added on day 2. Bioreactors were run for 12 days. Substrates utilization and metabolite formation was determined with HPLC-RI (sugars, alcohols and SCCA) and UPLC-DAD (caproate). Cell counts were determined with qPCR. A, D, G Fructose and glucose utilization and mannitol and EtOH formation. B, E, H SCCA and MCCA formation, C, F, I cell counts of members of the consortium. *sample was collected before C. kluyveri was added

The fermentation process was initiated by inoculation of L. reuteri at 2% vol/vol (4.4–5.2 log cells ml^−1^) (Fig. 2C, F). During the first 48 h of incubation, L. reuteri consumed 11.5 and 25.2% of the provided glucose and produced lactate in MF1-1 (8.0 mM) and MF1-2 (22.9 mM) (Fig. 2A, B, D, E). Fructose was fully reduced to mannitol on day 5 and 3 in MF1-1 and MF1-2, respectively (Fig. 2A, D). Here, 28.9 (MF1-1) and 21.7 mM (MF1-2) of caproate (2.5 and3.4 g L^−1^) were detected in fermentates on day 12 (Fig. 2B, E). Caproate productivity was highest on days 6 and 7 for MF1-1 and MF1-2, respectively (Fig. S2, 8.0 and 7.4 mM day^−1^). Other than caproate, the fermentates of MF1-1 and MF1-2 also contained lactate (95.7 and 137.5 mM), acetate (10.1 and 16.6 mM), butyrate (5.6 and 7.2 mM), mannitol (33.2 and 41.1 mM) and EtOH (15.5 and 0 mM). At the end of the bioreactor runs, cell counts of C. kluyveri and L. reuteri were 5.8–6.5 and 6.0–6.6 log cells ml^−1^, respectively (Fig. 2C, F).

L. reuteri, C. kluyveri, and S. cerevisiae produced caproate

As the addition of EtOH enhanced caproate formation of the bacterial consortium, we expanded the microbial community by adding the EtOH-producing S. cerevisiae FMT 3004 to the bacteria strains (Fig. 2G–I). On day 0, L. reuteri and S. cerevisiae were inoculated at 2 and 5% (vol/vol) resulting in initial cell counts of 5.8 ± 0.5 and 4.0 ± 0.8 log cells mL^−1^, respectively (Fig. 2I). On day 1, almost half of the initially added glucose was consumed, while fructose was fully used and converted to mannitol (30.7 ± 7.9 mM) (Fig. 2G). Cell counts of L. reuteri increased to 7.2 ± 0.5 log cells mL^−1^ while numbers of S. cerevisiae decreased by Δ0.9 log cells mL^−1^ during the first two days of fermentation (Fig. 2I). Lactate (65.4 ± 27.9 mM), acetate (22.0 ± 8.9 mM) and EtOH (32.1 ± 17.2 mM) were produced by the yeast-bacterium consortium (Fig. 2H). On day 2, C. kluyveri was inoculated at 4–5% vol/vol (5.8 ± 1.2 log cell mL^−1^) (Fig. 2I). On day 5, glucose was depleted, and butyrate (9.2 ± 0.7 mM), and caproate were detected (37.3 ± 18.1 mM). By day 12, the fermentates contained caproate (37.9 ± 11.4 mM), lactate (106.6 ± 42.6 mM), acetate (24.3 ± 12.6 mM), butyrate (8.4 ± 0.6 mM), EtOH (7.9 ± 3.5 mM), and mannitol (6.7 ± 5.4 mM) (Fig. 2G, H); cell counts of L. reuteri, S. cerevisiae and C. kluyveri were 6.8 ± 0.9, 3.0 ± 0.4 and 5.5 ± 0.4 log cells mL^−1^, respectively (Fig. 2I).

Glycerol production and transformation by the multi-kingdom consortium

In fermentations with S. cerevisiae, we observed the formation of up to 34.1 ± 11.8 mM 1,3-propanediol (1,3-PD) (Fig. 3A), which is an indicator of glycerol transformation by microbes that harbour the pdu operon (Engels et al. 2016a, b; Li et al. 2024). Glycerol levels based on NMR analysis were low (< 1 mM, Fig. 3A) in the fermentates, indicating immediate transformation. In agreement, we detected 1,3-PD after 4 h in MF2 (Fig. 3A).Fig. 3. Glycerol formation and utilization by multi-kingdom consortia. Glycerol and 1,3-propanediol levels were determined in fermentates of multi-kingdom consortia of bioprocesses MF2 (A) and in static Hungate tube incubations of co-cultures of S. cerevisiae and L. reuteri (B) in YC medium supplied with glucose and fructose that were incubated at 37 C for 3–4 d. Substrate and metabolite levels including 1,3-propanediol and glycerol were determined with HPLC-RI and NMR

To confirm glycerol formation of S. cerevisiae and PduCDE activity of L. reuteri, we conducted additional static batch fermentations in Hungate tubes with YC-GF (140 mM glucose and 60 mM fructose) (Fig. 3B). When S. cerevisiae was grown alone, 41.9 ± 4.2 mM glucose was used, and EtOH (43.0 ± 9.1 mM) and glycerol (2.8 ± 0.1 mM) were detected after 3–4 days of incubation (Fig. 3B). S. cerevisiae did not use any of the provided fructose possibly due to higher affinity for glucose (D’Amore et al. 1989). In co-culture with L. reuteri, more glucose was used (88.8 ± 19.3 mM) similar as in MF2 compared to MF0 and MF1 suggesting additive glucose utilization. Lactate (17.8 ± 4.5 mM) and EtOH (75.3 ± 35.5 mM) was produced. Fructose was reduced to 49.4 ± 6.6 mM mannitol, concurrently, 26.9 ± 4.6 mM acetate was formed (Fig. 3B). Glycerol was not detected, and 7.5 mM 1,3-PD was present at the end of incubation providing strong evidence for glycerol transformation by L. reuteri.

Mannitol was produced and utilized

In bioprocesses of bacterial and multi-kingdom consortia, we observed a decrease of mannitol (-∆6.1–13.9 and -∆24 ± 5.5 mM, respectively) during the 12-day fermentations of MF1 and MF2 (Fig. 1A and Fig. 2A, D, G). To test the mannitol utilization by the consortium members, additional experiments were conducted. In Hungate tubes and/or bioreactors, neither S. cerevisiae FMT 3004 (Fig. S3) nor C. kluyveri DSM 556 was not able to metabolize mannitol (Fig. 4A, Fig. S3). Yet, when ethanol and acetate were supplied to YC-M in the bioreactor, mannitol was reduced by C. kluyveri DSM 556 by Δ6.0 mM during the 8-day incubation (Fig. 4A). To test whether L. reuteri FMT 421 produced and utilized mannitol, we inoculated an overnight culture, which was grown in mMRS-G-F to initiate mannitol metabolism, at 10% (vol/vol) in YC-M. There was a concurrent transfer of about 5 mM lactate and 1.5 mM acetate and ethanol (Fig. 4B). Within 3 days, the concentration of mannitol was reduced by Δ2.3 mM while levels of acetate and ethanol increased (+ 0.4 mM) (Fig. 4B) suggesting some capacity of L. reuteri to re-use mannitol. Fig. 4. Mannitol utilization by C. kluyveri and L. reuteri. (A) C. kluyveri was grown in a bioreactor supplied with YC-mannitol (YC-M, solid line) YC-M containing acetate and ethanol (dashed line) (B) L. reuteri was cultivated in Hungate tubes with YC-M. Samples were collected in regular intervals to determine substrate utilization and metabolite formation

Antimicrobial activity against microbial indicators in media and in a food model

The major SCCA and MCCA in fermentates were lactate and caproate. To investigate the contribution of both carboxylic acids to the antimicrobial activity of our fermentates, we performed a two-fold dilution assay and used a selection of Gram-positive and negative bacteria, yeast, and molds as indicators at pH 4.5 and 6.5. All strains grew at pH 4.5 and 6.5.

In general, caproate was more inhibitory than lactate against the indicators (Table 2). Even if the pH was adjusted to pH 4.5, only a few strains were inhibited at concentrations up to 50 mM lactate (Table 2). The MIC_50_ of caproate ranged from 0.4–3.2 mM and 0.5–25.0 mM for bacteria and yeasts and molds, at pH 4.5 and 6.5, respectively. Fermentates conferred antimicrobial activity against all bacteria, yeast and molds at pH 4.5. The DF_50_ values of MF1 fermentates were consistently lower (8–145) than of fermentates of MF2 (14–193) likely due to higher caproate content (Table 2). When we calculated the concentrations of caproate and lactate presented in the fermentates at the DF_50_, levels of caproate were similar to the MIC_50_, while levels of lactate were too low to explain the antimicrobial activity indicating that caproate was the major antimicrobial contributor.Table 2. Antimicrobial activity of caproate and fermentates. Bacterial consortium produced caproate (25.5 ± 1.1 mM), lactate (116.6 ± 29.5 mM), acetate (0.2 ± 0.3 mM) and butyrate (6.4 ± 1.1 mM) (mean of MF1-1 and MF1-2); while multi-kingdom consortium produced caproate (37.9 ± 14.0 mM), lactate (106.6 ± 52.1 mM), acetate (10.2 ± 13.2 mM) and butyrate (8.4 ± 0.8) (mean of MF2-1, MF2-2 and MF2-3). Antimicrobial activity was tested using two-fold broth dilution assays in 96 well microtiter plates. The minimal inhibitory concentration of lactate and caproate, or the dilution factor of fermentates to reduce density to 50% (MIC_50_ or DF_50_) was determined by four-parameter logistic equation for bacteria and yeast, and was visually determined for molds. Based on lactate/caproate levels of fermentates and the DF_50_ values, we calculated the actual levels of lactate and caproate in the diluted fermentates that conferred 50% inhibition. Data is shown as average values of two to three independent biological replicates. Statistical difference between DF_50_ was compared using Student’s t-test. C, average caproate concentration in the fermentate (mM); L, average lactate concentration in the fermentate (mM)DomainStrainsMinimal inhibitory concentration(MIC_50_, mM)Dilution factor(DF_50_ and estimated concentrations of caproate (C) and lactate (L))CaproateLactateBacterial consortium(MF1-1, MF1-2)Multi-kingdom consortium(MF2-1, MF2-2, MF2-3)pH4.56.54.56.54.56.54.56.5BacteriaSalmonella enterica DSM 170582.0 ± 0.3^1^31.0 ± 0.8^1^ > 50^1^ > 50^1^NT^2^NT^2^NT^2^NT^2^Salmonella enterica CCM 44201.3 ± 0.3^1^NT^12^ > 50^1^NT^12^79.7 ± 8.2^Aa^C: 0.3L: 1.33.8 ± 0.1^B^C: 6.7L: 28.1111.0 ± 3.9^Ab^C: 0.3L: 1.03.0 ± 0.2^B^C: 12.6L: 38.8Escherichia coli DSM 192061.6 ± 0.224.0 ± 1.315.1 ± 6.0 > 50NTNTNTNTKosakonia cowanii FMT 341.7 ± 0.129.8 ± 4.1 > 50 > 5010.4 ± 3.8^a^C: 2.5L: 11.2NI^2^16.5 ± 0.9^Aa^C: 2.3L: 6.53.8 ± 0.4^B^C: 10.0L: 28.1Klebsiella oxytoca DSM 51151.4 ± 0.015.1 ± 3.3 > 50 > 5010.3 ± 2.9^a^C: 2.5L: 11.3NI^2^17.9 ± 0.7^Aa^C: 2.1L: 6.03.9 ± 0.0^B^C: 9.7L: 27.3Pantoea ananatis FMT 331.1 ± 0.414.4 ± 4.2 > 50 > 5024.4 ± 6.2^a^C: 1.0L: 4.8NI^2^40.1 ± 9.7^Aa^C: 0.9L: 2.74.8 ± 0.6^B^C: 7.9L: 22.2Pantoea sp. FMT 320.4 ± 0.25.0 ± 0.2 > 50 > 5054.3 ± 20.4^a^C: 0.5L: 2.2NI^2^107.7 ± 1.7^Aa^C: 0.4L: 1.017.7 ± 3.0^B^C: 2.1L: 6.0Bacillus subtilis FMT 4290.5 ± 0.28.4 ± 0.9 > 50 > 5034.3 ± 18.7^Aa^C: 0.7L: 3.42.6 ± 1.2^A^C: 9.8L: 44.852.0 ± 21.5^Aa^C: 0.7L: 2.16.2 ± 0.6^A^C: 6.1L: 17.2Bacillus subtilis FMT 4310.7 ± 0.26.9 ± 3.5 > 50 > 5025.9 ± 10.1^Aa^C: 1.0L: 4.54.3 ± 0.1^A^C: 5.9L: 27.1135.6 ± 10.0^Aa^C: 1.1L: 3.08.2 ± 0.4^A^C: 4.6L: 13Listeria innocua DSM 20649NG^2^20.5 ± 3.0NG^2^ > 50NT^2^NT^2^NT^2^NT^2^Staphylococcus aureus DSM 1104NG^2^25.8 ± 1.2NG^2^ > 50NT^2^NT^2^NT^2^NT^2^YeastCandida albicans DSM 106971.1 ± 0.22.9 ± 0.642.9 ± 1.7NT^2^111.8 ± 13.1^Aa^C: 0.2L: 1.018.7 ± 10.2^B^C: 1.4L: 6.2146.0 ± 18.0^Aa^C: 0.3L: 0,724.9 ± 2.0^B^C: 1.5L: 3.6Pichia kluyveri FMT 30000.8 ± 0.56.2 ± 4.7 > 50 > 5023.2 ± 3.8^Aa^C: 1.1L: 5.07.2 ± 3.9^A^C: 3.5L: 16.230.5 ± 19.0^Aa^C: 1.2L: 3.512.2 ± 6.8^A^C: 3.1L: 8.7Saccharomyces cerevisiae FMT 30030.4 ± 0.62.7 ± 2.1 > 50 > 5016.6 ± 2.9^Aa^C: 1.5L: 7.04.6 ± 0.9^B^C: 5.5L: 25.330.6 ± 5.5^Aa^C: 1.2L: 3.58.0 ± 2.0^B^C: 4.7L: 13.3Saccharomyces cerevisiae FMT 30040.3 ± 0.16.4 ± 0.4 > 50 > 5044.6 ± 9.3^Aa^C: 0.6L: 2.410.3 ± 2.5^B^C: 2.5L: 10.456.2 ± 7.6^Aa^C: 0.7L: 2.113.3 ± 4.5^B^C: 2.9L: 8.8MoldsAspergillus niger BCE M10.812.5 > 50 > 5020.0 ± 17.0^Aa^C: 1.3L: 5.84.0 ± 0.0^A^C: 6.4L: 29.224.0 ± 11.3^Aa^C: 1.6L: 4.44.7 ± 1.4^A^C: 8.1L: 22.7Penicillium roquefortii BCE M23.225 > 50 > 508.0 ± 0.0^Aa^C: 3.2L: 14.66.0 ± 2.8^A^C: 4.3L: 19.414.0 ± 2.8^Aa^C: 2.7L: 7.64.0 ± 0.0^B^C: 9.5L: 26.7Penicillium purpurogenum DSM 211700.46.25 > 50 > 5048.0 ± 22.6^Aa^C: 0.5L: 2.46.0 ± 2.8^A^C: 4.3L: 19.496.0 ± 45.3^Aa^C: 0.4L: 1.18.0 ± 0.0^A^C: 4.7L: 13.3^1^Data taken from Ng et al. 2024^2^NI no inhibition, NT not tested, NG no growth^A^^−^^B^For each strain, different uppercase letters indicate significant difference (P < 0.05) in DF_50_ values for the same fermentate (bacterial or multi-kingdom consortium) at pH 4.5 or 6.5 based on Student’s t-test^a^^−^^b^For each strain, different lowercase letters indicate significant difference (P < 0.05) in DF_50_ values of bacterial or multi-kingdom consortia fermentates at pH 4.5 based on Student’s t-test

To investigate the antimicrobial activity of lactate and caproate in a food system, we used minced pork and beef. Generally, the pH of minced meat was comparable to the controls after the addition of selected carboxylic acids (pH 5.5–6.1), with the exception of the combination of lactate and caproate in minced pork at day 0 (Table 3). The pH of pork meat did not change during storage, while the pH of caproate/lactate-treated beef decreased. Total plate counts increased from 4.2 to 7.0 log CFU g^−1^ (P < 0.05), and 4.4 to 7.7 log CFU g^−1^ in untreated pork and beef meat, respectively, during storage. On day 7, total plate counts of caproate treated minced pork were 1.5 log CFU g^−1^ lower compared to controls, while the combination of caproate and lactate reduced total cell counts by 2.1 log CFU g^−1^ (P < 0.05). In minced beef, the addition of lactate and caproate reduced counts by 1.4 log CFU g^−1^. These results indicate that a combination of caproate and lactate confers antimicrobial activity in meat with a limitation that experiments in minced beef were only conducted once. Table 3. Effects of lactate and/or caproate treatment on cell counts of minced meat samples*.* Minced meat samples were treated with caproate (CA) and/or lactate (LA) as indicated. Samples were stored at 7 °C for 7 d, and pH, total plate counts were determined on nutrient agar plates after 0, 4 and 7 d of storage. Minced meat experiments were conducted in independent triplicatesTotal plate counts (log CFU g^−1^)pHIncubation time (days)047047Minced porkControl4.2 ± 0.3 ^aX^5.9 ± 0.3 ^aY^7.0 ± 0.3 ^aZ^5.9 ± 0.3 ^aX^6.0 ± 0.1 ^aX^5.8 ± 0.1 ^aX^LA, 25 mmol kg^−1^3.9 ± 0.1 ^aX^5.2 ± 0.5 ^abXY^5.8 ± 0.9 ^bY^5.7 ± 0.2 ^abX^5.7 ± 0.2 ^aX^5.5 ± 0.2 ^aX^CA, 25 mmol kg^−1^3.8 ± 0.0 ^aX^4.8 ± 0.5 ^bY^5.5 ± 0.4 ^bY^5.9 ± 0.1 ^abX^6.0 ± 0.1 ^aX^5.8 ± 0.1 ^aX^LA + CA,25 and 25 mmol kg^−1^3.8 ± 0.0 ^aX^4.4 ± 0.7 ^bXY^4.9 ± 0.3 ^bY^5.3 ± 0.3 ^bX^5.7 ± 0.2 ^aX^5.5 ± 0.1 ^aX^Minced beefControl4.47.07.76.135.645.60LA + CA,90 and 30 mmol kg^−1^3.94.86.35.955.976.04^a^^−^^b^ indicate significant differences (P < 0.05) among different treatments within the same storage day using Tukey’s test^X^^–^^Y^ indicate significant differences (P < 0.05) among different storage days within the same treatment using Tukey’s test

Discussion

A multi-kingdom consortium of L. reuteri, C. kluyveri, and S. cerevisiae functioned as a self-sustained caproate producer

Our results present the successful design of a multi-kingdom consortium whose interaction was based on metabolite cross-feeding. The consortium produced caproate from glucose and fructose with a yield of 0.24 mM caproate per mM glucose, and a concentration of 3.9–8.5 g L^−1^. Co-cultures of Clostridium acetobutylicum ATCC 824 and Clostridium saccharolyticum ATCC 35040, which provide acetate and EtOH from glucose, and C. kluyveri DSM 555 (ATCC 8527) produced 30 mM (3.5 g L^−1^) and 161 mM (18.7 g L^−1^) caproate during a 5.6-day bioprocess with 140 mM initial levels of glucose and two additional glucose supplementations (Otten et al. 2022). Bioaugmentation of microalgae biomass with baker’s yeast enabled caproate production by anaerobic digestion sludge from 0 to 22 mM (Shi et al. 2024). In comparison, our multi-kingdom consortium composed of food-grade microbes in addition to C. kluyveri, was only supplied with carbohydrates once and maintained a self-sustained process leading to the formation of caproate. In our experiments, concentrations of primary (e.g. lactate, ethanol) and secondary glucose metabolites (1,3-PD) differed between experiments of the multi-kingdom consortium possibly because both S. cerevisiae and L. reuteri competed for glucose. These observations indicate the difficulties in working with complex communities (Dong et al. 2023). Nonetheless, caproate was consistently produced suggesting robustness of the system.

In previous reports, the chain elongation process at pH 5.5 and 30 °C was inhibited at pH 5.5 and 30 °C when the concentration of undissociated caproic acid was as low as 7.5 mM (Ge et al. 2015). The near-neutral pH of our process (pH 6.8) favored the presence of caproate in its dissociated form reducing the impact of product inhibition.

The importance of EtOH on caproate formation by the consortia

C. kluyveri cannot use glucose, or fermentation intermediates such as pyruvate (Bornstein and Barker 1948), and relies on the availability of EtOH and acetate to form caproate. In previous studies, CO_2_ contributed only a minor fraction of the total carbon incorporated into products (Kenealy et al. 1995; Kucek et al. 2016; Steinbusch et al. 2011), therefore we did not focus on CO_2_ in the current study. Based on the theoretical reaction scheme suggested by Seedorf et al. (Seedorf et al. 2008) (6 M EtOH + 3 M acetate → 3 M butyrate + 1 M caproate), at least twice as much EtOH than acetate needs to be present to enable caproate formation. If acetate is present in excess, butyrate is produced (Smith et al. 1985) in agreement with the observations from MF0. Caproate production was initiated once the EtOH:acetate ratio was > 1 suggesting that an insufficient amount of EtOH was the reason for low initial caproate formation.

When C. kluyveri was added on day 2 to fermentations MF1, the ratio of EtOH:acetate in the fermentates was 0.8:1.0 (MF1-1) and 0.7:1.0 (MF1-2); in MF1-1, the addition of EtOH changed the ratio to 10.9:1.0. In a previous study, C. kluyveri DSM 555 produced more caproate (111 mM, 12.9 ± 0.67 g L^−1^) when the EtOH:acetate ratio was 3.5:1.0 compared to 5.5:1.0 (78 mM, 9.1 g L^−1^) (Otten et al. 2022). The bacterial consortium of L. reuteri and C. kluyveri was able to synthesize caproate with a titer of 2.5 g L^−1^ and 3.4 g L^−1^ (21.7–29 mM) with the addition of EtOH within a 12-day bioprocess suggesting that caproate was produced even at not optimal EtOH:acetate ratios possibly because we used a dynamic system that provided both compounds throughout the fermentation.

The multi-kingdom consortium formed glycerol and produced reuterin

During anaerobic growth and in the presence of high levels of glucose, S. cerevisiae is known to produce glycerol (Nevoigt & Stahl 1997). L. reuteri harbours the pdu operon (Engels et al. 2016a, b; Li et al. 2024) and the cobalamin-dependent key enzyme glycerol/diol dehydratase PduCDE that transforms glycerol to 3-hydroxypropanal (3-HPA), which dehydrates to acrolein at physiological conditions (T = 4–45 °C; pH 6–9) (Engels et al. 2016a, b). 3-HPA and acrolein constitute the antimicrobial reuterin system (Engels et al. 2016a, b). 3-HPA can be further metabolised to 1,3-PD and 3-hydroxypropionate (3-HP); 3-HP also has antimicrobial activity (Engels et al. 2016a, b; Liang et al. 2021). The detection of 1,3-PD in bioreactors and in Hungate tubes suggests glycerol cross-feeding of S. cerevisiae and L. reuteri. As 1,3PD is the major final product from glycerol transformation of L. reuteri if glucose is present (Lüthi-Peng et al. 2002), likely only a small proportion of 3-HPA degraded to acrolein for overall low reuterin concentration. In addition, strains of C. kluyveri also harbour the pdu operon but lack the genetic potential to form the cofactor cobalamin (Seedorf et al. 2008). C. kluyveri could contribute to glycerol transformation (Fig. 5) if cobalamin was shared by L. reuteri, but experimental evidence remains to be obtained*.*Fig. 5. Schematic overview of metabolic interactions of synthetic microbial consortia. Proposed interactions of the bacterial consortium, and with the addition of S. cerevisiae (in blue) of the multi-kingdom consortium. ‘?’ Possible interactions

L. reuteri was capable of utilizing mannitol

In bioprocesses of bacterial and multi-kingdom consortia, we observed a decrease of mannitol. Mannitol utilization is not common in S. cerevisiae and may occur only after spontaneous mutation (Kazamias & Sperry 1995). In a previous investigation, sugar alcohols supported the growth of members of the Clostridia (Tiffany et al. 2021), yet the type strain C. kluyveri DSM 555 does not possess any genes encoding mannitol dehydrogenase (Seedorf et al. 2008). Other phosphorylation-dependent pathways exist, for example for Clostridium difficile (Kazamias & Sperry 1995). In our test assays, C. kluyveri DSM 556 was only able to metabolize mannitol in the presence of acetate and ethanol, which were available during consortium fermentations. Yet, mannitol consumption started between day 1 and 2 in our bioprocesses MF2 even before C. kluyveri was added pointing at L. reuteri as a concurrent mannitol producer and utilizer. The ability to use mannitol has been reported for a few Lactobacillaceae. Lactiplantibacillus plantarum metabolized mannitol in anaerobic conditions in the presence of electron acceptors; e.g. acetate that was transformed to EtOH (McFeeters & Chen. 1986). Together, our observations suggest that both L. reuteri and C. kluyveri had the ability to metabolize mannitol and possibly contributed to the mannitol reduction that was observed for the bacterial and multi-kingdom consortia (Fig. 5).

Caproate was more antimicrobial than lactate in vitro and in a food model

Lactate and caproate were the two major carboxylic acids in fermentates, and in general, caproate was more inhibitory than lactate against the bacteria, yeast and fungal indicators in vitro. Even if the pH was adjusted to pH 4.5, only a few strains were inhibited at concentrations up to 50 mM lactate. These observations are in agreement with previous studies (Elvan Gezer et al. 2025; Ng et al. 2024), suggesting that the antimicrobial activity of lactate is due to acidification and not direct microbial interaction if the initial pH is adjusted. In contrast, caproate conferred antimicrobial activity at pH 4.5 and 6.5, respectively, highlighting that caproate effectively inhibited a broad diversity of microorganisms even at near-neutral pH.

Previous studies demonstrated that the addition of 1–3% (134–402 mM) lactate reduced cell counts of E. coli, Listeria monocytogenes and S. enterica in meat samples (Van Ba et al. 2018), which were much higher than the lactate concentration (0.13%) used in current study indicating the synergistic effect of lactate and caproate. Myosin is the major protein in meat products with an isoelectric point (pI) of about 5.4 (Huff-Lonergan and Lonergan 2005). In our case, the addition of caproate and/or lactate was unlikely to induce water dripping in the minced meat, as the final pH were 5.5–5.8 and no visible water release was observed. Overall, our results showed that the main carboxylic acids of our fermentates acted synergistically in a food system.

Caproate containing fermentates reduced growth of indicator strains

In addition to lactate and caproate, we tested sensitivity of indicator microbes against the caproate-containing fermentates produced by the consortia. Compared to single chemicals, fermentates do not require extensive purification steps and can profit from complementary activity of compounds present in the fermentates. Sensitivity towards fermentates differed between bacteria, yeast and fungi, and between strains of a species, indicating individuality in susceptibility towards the antimicrobial activity of SCCA and MCCA. While inhibition activity was higher at pH 4.5, our data showed that caproate fermentates confer antimicrobial activity also at pH 6.5. According to the Henderson-Hasselbalch equation, it is likely that the much higher undissociated fraction of caproate (70%) at pH 4.5 compared to pH 6.5 (1.2%) enhances the antimicrobial activity of fermentates.

In previous work, we designed consortia to produce fermentates containing propionate through cross-feeding and tested similar indicators with the same experimental set-up (Elvan Gezer et al. 2025; Buljubašić et al. 2024). These propionate-fermentates also contained lactate in a ratio of approximately 1:1.4 (propionate:lactate). Compared to the propionate-fermentates, the caproate-fermentates were more antimicrobial against S. enterica CCM 4420 (DF_50_ of propionate fermentates 3.9–25.3) and K. oxytoca DSM 5115 (DF_50_ 2.5–7.1). Caproate-fermentates inhibited C. albicans DSM 10697, which was not affected by propionate-fermentates (Elvan Gezer et al. 2025) and were more effective against the mould indicators. These observations suggest that caproate fermentates are stronger antimicrobial than propionate fermentates, which have potential for further development as biopreservatives.

Taken together, we present here a novel approach using minimal and synthetic bacterial or multi-kingdom consortia to produce caproate-containing fermentates from fructose and glucose. Beyond the expected cross-feeding of acetate and ethanol, the consortium members were capable of glycerol transformation and mannitol formation and utilization indicating diverse metabolite dynamics of the consortium. The caproate-containing fermentates conferred antimicrobial activity against diverse bacteria, yeast and mold pathogens and spoilage microorganisms within a pH range of 4.5 and 6.5. While in a pH-adjusted system, most antimicrobial activity was due to caproate, caproate and lactate acted synergistically in a model food matrix and led to lower total plate counts, which might delay spoilage. Whether caproate-containing fermentates negatively affect the sensory aspects remains to be investigated. Other potential applications could include the use as sanitizers or disinfectants. Future work will include further bioprocess optimization, and bioprocesses at scale and downstream processes remain to be developed.

Supplementary Information

Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 1387 kb)