Development of Antifungal Lactic Acid Bacteria-Yeast Co-Cultures for Cocoa Bean Fermentation

Abstract

The fermentation of cocoa beans is a spontaneous process which is defined by a microbial succession of yeasts, lactic acid bacteria (LAB), and acetic acid bacteria (AAB). Filamentous fungi may also proliferate during cocoa bean fermentation, mainly in the well-aerated and cold outer layers of the fermentation mass, and during the drying process. Filamentous fungal growth may reduce cocoa bean quality, for example, by creating off-flavours or leading to internal mould, and a potential accumulation of mycotoxins, such as ochratoxin A, aflatoxins, or fumonisin, poses a health risk for consumers. Antifungal cultures of LAB and yeasts have been used for a long time to inhibit filamentous fungal growth and the concomitant mycotoxin production in diverse food products, but only little is known about antifungal cultures for cocoa bean fermentation. Therefore, the aim of this thesis was to develop antifungal LAB-yeast co-cultures for cocoa bean fermentation. A multiphasic approach was applied with the aim of selecting antifungal cultures that are well adapted to the cocoa bean fermentation process, reduce the growth of mycotoxigenic filamentous fungi, and do not negatively influence cocoa bean fermentation and the quality of the resulting cocoa beans. The selection of antifungal cultures encompassed three phases: (i) screening of LAB and yeast strains for antifungal activity, carbon metabolism, and stress tolerance in vitro, (ii) growth inhibition tests in vivo on cocoa beans against potentially mycotoxin-producing filamentous fungal strains, and (iii) the application of selected antifungal LAB-yeast co-cultures to the cocoa bean fermentation. In phase one (i), antifungal strains well adapted to the cocoa bean fermentation process were selected from 362 LAB and 384 yeast strains previously isolated from cocoa bean post-harvest processes. LAB and yeast strains were tested for antifungal activity using an agar plate assay based on buffered MRS and MEA, respectively, and modified cocoa pulp simulation medium (mCPSM) agar. Lactobacillus fermentum and Lactobacillus sp. strains showed considerable antifungal activity on modified cocoa pulp simulation medium (mCPSM), in contrast to strains of the species Lactobacillus plantarum that displayed inhibition zones on buffered MRS but not on mCPSM, emphasizing the importance of the growth medium in antifungal activity tests. Selected antifungal strains, i.e. 26 LAB and 63 yeasts, exhibited broad antifungal activity on mCPSM against up to seven filamentous fungal strains, i.e. Gibberella moniliformis S003, Penicillium citrinum S005, Aspergillus flavus S075, and four further strains of the genus Aspergillus. 26 antifungal LAB and 45 antifungal yeast strains were further tested for their metabolism and their metabolite and heat stress tolerance in mCPSM broth. The carbon metabolism of most antifungal strains was well adapted to the substrates of cocoa pulp. All LAB strains consumed citric acid and used glucose and fructose to produce lactic acid, acetic acid, and mannitol, while none of the yeasts metabolized citric acid, but 32 out of 45 strains used glucose and fructose to produce ethanol. Most antifungal strains tolerated ethanol at a concentration above the ones typically reported for cocoa bean fermentations. Furthermore, most LAB tolerated increased acetic acid concentrations and a temperature of 45 °C and yeasts grew well at elevated lactic acid levels. However, a temperature of 45 °C and elevated acetic acid concentrations seemed to be potentially limiting factors for the growth of yeasts, whereas lactic acid was the most critical factor for the growth of LAB. In phase two (ii), selected well-adapted antifungal strains, i.e. 14 LAB, 4 Hanseniaspora, and 12 Saccharomyces, were screened for in vivo antifungal activity in a growth inhibition test based on 20 g of cocoa pulp-bean mass. Six Lb. fermentum, two Hanseniaspora, and four Saccharomyces strains inhibited the aflatoxin-producer A. flavus S075 at 51-100% and of these six Lb. fermentum strains, five also completely inhibited the citrinin-producing strain P. citrinum S005 and the potentially fumonisin-producing strain G. moniliformis S003 after 11-14 days on the surface of cocoa beans. Finally, four selected LAB-yeast co-cultures completely inhibited A. flavus S075 after 14 days of incubation. In phase three (iii), application trials were performed at laboratory scale and industrial scale to evaluate the influence of antifungal co-cultures on the fermentation process and cocoa bean quality. The laboratory-scale fermentation, performed with 1 kg cocoa pulp-bean mass in plastic pots at a defined temperature profile, was compared to a traditional Honduran 300-kg on-farm box fermentation for validation purposes. In general, the spontaneous laboratory-scale process mimicked the spontaneous on-farm process regarding microbial counts, substrate consumption, and metabolite production and dry beans were of comparable quality. However, during laboratory-scale fermentation, up to 3 log CFU/g lower LAB and AAB counts were measured, which led to beans with up to 4 times higher residual sugars, 6 times lower lactic acid concentrations, and 3-12 times more polyphenols, and which were rated with 2 units higher off-flavours on a scale from 0 to 10. The laboratory-scale model system was used to evaluate six antifungal co-cultures, each composed of an LAB strain, i.e. Lactobacillus fermentum M017, Lb. fermentum M089, or Lb. fermentum 223, and a yeast strain, i.e. Hanseniaspora opuntiae H17 or Saccharomyces cerevisiae H290. The co-cultures with H. opuntiae H17 led to up to 3.6 log CFU/g lower maximal yeast counts compared to the spontaneous fermentations, which resulted in an incomplete metabolization of pulp sugars that instead were converted into gluconic acid. On the contrary, inoculation with S. cerevisiae H290 in co-culture with Lb. fermentum, led to 0.5-3.0 log CFU/g higher maximal yeast counts, accelerated the pulp sugar consumption, and resulted in up to 3 times lower residual sugars in dry beans and 1-22% more well-fermented beans. Furthermore, S. cerevisiae H290 positively influenced the flavour profiles of laboratory-scale fermented and dried beans in co-culture with Lb. fermentum M017 or Lb. fermentum 223, i.e. the resulting beans displaying 0.9-1.5 units lower astringency, bitterness, and off-flavours, and, in combination with Lb. fermentum 223, even 1.2-1.4 units more cocoa and fine flavours than spontaneously fermented beans. In contrast, no impact of the two selected LAB-yeast co-cultures, i.e. Lb. fermentum M017/S. cerevisiae H290 (A) and Lb. fermentum 223/S. cerevisiae H290 (B), was observed on microbial counts or substrate and metabolite concentrations when inoculated to 180-kg industrial scale fermentations, except for up to 10 times higher mannitol levels in bean cotyledons. However, at industrial scale, the co-cultures seemed to slow the fermentation process, which was observed in terms of up to 8-12 °C lower temperatures in the centre of the fermentation mass, where an increase from ambient temperature to 37-50 °C took place, and up to 22% fewer well-fermented beans after drying compared to the spontaneous fermentations. When comparing the influence of the two antifungal co-cultures at industrial scale, co-culture B, i.e. Lb. fermentum 223 / S. cerevisiae H290, led to superior cocoa bean quality, i.e. 0.5-1.9 mg/g lower acetic acid levels among levels of 2.5-9.1 mg/g in dry beans, 4-17% more well-fermented beans, less astringency and off-flavours, and more cocoa flavour compared to co-culture A. To conclude, co-culture B, Lb. fermentum 223/S. cerevisiae H290, was selected from 362 LAB and 384 yeast strains due to its antifungal activity on cocoa beans, metabolite and heat stress tolerance, well-adapted carbon metabolism of both strains in mCPSM, and a neutral to positive influence on cocoa bean quality although it slowed industrial-scale fermentations. In future studies, the selected antifungal LAB-yeast co-culture needs to be optimised with respect to an accelerated fermentation process and its antifungal activity needs to be investigated in on-farm fermentations, i.e. its ability to inhibit filamentous fungal growth and to reduce mycotoxin levels in dry beans. Future research should focus on antifungal mechanisms of the selected LAB and yeast strains against mycotoxigenic filamentous fungi, including the interaction of the co-culture strains and the influence of the cocoa pulp-bean matrix.