Fermented dry cocoa beans are the basic raw material for chocolate

Transcription

1 Oxidation of Metabolites Highlights the Microbial Interactions and Role of Acetobacter pasteurianus during Cocoa Bean Fermentation Frédéric Moens, Timothy Lefeber, Luc De Vuyst Research Group of Industrial Microbiology and Food Biotechnology, Faculty of Sciences and Bioengineering Sciences, Vrije Universiteit Brussel, Brussels, Belgium Four cocoa-specific acetic acid bacterium (AAB) strains, namely, Acetobacter pasteurianus 386B, Acetobacter ghanensis LMG T, Acetobacter fabarum LMG T, and Acetobacter senegalensis 108B, were analyzed kinetically and metabolically during monoculture laboratory fermentations. A cocoa pulp simulation medium (CPSM) for AAB, containing ethanol, lactic acid, and mannitol, was used. All AAB strains differed in their ethanol and lactic acid oxidation kinetics, whereby only A. pasteurianus 386B performed a fast oxidation of ethanol and lactic acid into acetic acid and acetoin, respectively. Only A. pasteurianus 386B and A. ghanensis LMG T oxidized mannitol into fructose. Coculture fermentations with A. pasteurianus 386B or A. ghanensis LMG T and Lactobacillus fermentum 222 in CPSM for lactic acid bacteria (LAB) containing glucose, fructose, and citric acid revealed oxidation of lactic acid produced by the LAB strain into acetic acid and acetoin that was faster in the case of A. pasteurianus 386B. A triculture fermentation with Saccharomyces cerevisiae H5S5K23, L. fermentum 222, and A. pasteurianus 386B, using CPSM for LAB, showed oxidation of ethanol and lactic acid produced by the yeast and LAB strain, respectively, into acetic acid and acetoin. Hence, acetic acid and acetoin are the major end metabolites of cocoa bean fermentation. All data highlight that A. pasteurianus 386B displayed beneficial functional roles to be used as a starter culture, namely, a fast oxidation of ethanol and lactic acid, and that these metabolites play a key role as substrates for A. pasteurianus in its indispensable crossfeeding interactions with yeast and LAB during cocoa bean fermentation. Fermented dry cocoa beans are the basic raw material for chocolate production. Cocoa beans are the seeds of the cocoa tree, Theobroma cacao L(1 4). Cocoa bean fermentation is still a spontaneous on-farm process. The key microorganisms for successful cocoa bean fermentations are yeasts, lactic acid bacteria (LAB), and acetic acid bacteria (AAB). Their succession during cocoa bean fermentation is important for the required metabolic activities in the fermenting cocoa pulp-bean mass, although the occurrence of these microbial communities may overlap (2, 3, 5 7). Yeast species involved in cocoa bean fermentation mainly encompass Saccharomyces cerevisiae and Hanseniaspora guilliermondii, H. opuntiae, and H. uvarum, which are responsible for pectinolysis of the cocoa pulp and the production of ethanol from carbohydrates (mainly glucose) (8 15). LAB species involved in cocoa bean fermentation, in particular Lactobacillus fermentum, consume carbohydrates (glucose and fructose) and citric acid and produce lactic acid, acetic acid, and mannitol (6, 7, 12, 16 19). Concerning AAB, mainly Acetobacter species are found, which oxidize ethanol (produced by yeast) and lactic acid (produced by LAB) into acetic acid (6, 7, 12, 15, 17, 18, 20, 21). Yeasts and LAB display a remarkable tolerance toward the low ph value (ph 3.0 to 3.5 because of the presence of citric acid) and anaerobic conditions of the initial cocoa pulp-bean mass. Also, AAB are able to survive under these anaerobic conditions (despite their obligatory aerobic metabolism) and to oxidize their substrates at low ph and high temperature (6, 13, 20, 21). In general, this occurs after 24 to 48 h of cocoa bean fermentation, when the ph slightly increases due to citric acid metabolism by the LAB and when aeration increases due to air ingress because of pulp breakdown by the yeasts (6, 16, 21, 22). Besides Acetobacter pasteurianus, which prevails throughout cocoa bean fermentation, different Acetobacter species may be present at the onset of a spontaneous cocoa bean fermentation process. These environmental contaminants encompass Acetobacter fabarum, Acetobacter ghanensis, and Acetobacter senegalensis (6, 13, 18, 20 25). However, not much is known about how cocoaspecific Acetobacter species, in particular A. pasteurianus, adapt physiologically to the cocoa pulp-bean mass. Alternatively, Acetobacter species may play an important role from the start of the fermentation, in addition to yeasts and LAB, but this has not been examined before (5, 6, 20, 21). Moreover, while some studies show a sharp decrease of the Acetobacter communities after the first 24 h of fermentation (5), others report their persistence throughout fermentation even when mixing of the cocoa pulp-bean mass is not performed (6, 7, 18, 20 22, 25). These divergent microbial community dynamics often observed during cocoa bean fermentation processes necessitate further investigation of the functional properties and ecophysiology of the microbial species involved, in particular Acetobacter species, for instance, by making use of cocoa pulp simulation media during laboratory fermentations. Among various LAB strains tested in a cocoa pulp simulation medium for LAB, cocoa-specific strains of L. fermentum are best adapted to the cocoa pulp ecosystem (17, 19). Also, a cocoa pulp simulation medium for the growth of AAB has been formulated to mimic the growth and metabolite production of strains of Acetobacter species under cocoa bean fermentation conditions (16). However, only the oxidation of ethanol into acetic acid by A. pasteurianus strains (including 386B) has been studied, in particular under aerobic conditions Received 8 October 2013 Accepted 1 January 2014 Published ahead of print 10 January 2014 Editor: D. W. Schaffner Address correspondence to Luc De Vuyst, ldvuyst@vub.ac.be. Copyright 2014, American Society for Microbiology. All Rights Reserved. doi: /aem aem.asm.org Applied and Environmental Microbiology p March 2014 Volume 80 Number 6

2 Simulated Cocoa Pulp Fermentations and at relatively high ph (4.5), mimicking the end phase of a spontaneous cocoa bean fermentation (16). The latter study did not allow the examination of the role and metabolic activity of this and other Acetobacter species at the onset of natural cocoa bean fermentations (anaerobic phase, lower ph than at the end). Furthermore, no compensation for ethanol losses through evaporation during oxidation of ethanol into acetic acid (low levels of acetic acid were produced out of ethanol by A. pasteurianus 386B in the study of Lefeber et al. [16]), due to intensive aeration and a high fermentation temperature, was taken into account. Hence, the former study did not allow the calculation of the stoichiometric oxidation of ethanol and lactic acid into acetic acid during cocoa bean fermentation and its further oxidation into carbon dioxide and water during the aerobic phase at the end of the cocoa bean fermentation process (6, 20, 22). The present study aimed at a detailed kinetic analysis of the oxidation of energy sources relevant for Acetobacter species occurring in fermenting cocoa pulp-bean mass during monoculture laboratory fermentations in cocoa pulp simulation media. Based on these data, the study also aimed to investigate microbial interactions of some of these species with L. fermentum and S. cerevisiae through biculture and triculture fermentations. MATERIALS AND METHODS Microorganisms and media. Four cocoa-specific AAB strains (Acetobacter pasteurianus 386B, Acetobacter ghanensis LMG T, Acetobacter fabarum LMG T, and Acetobacter senegalensis 108B), one LAB strain (Lactobacillus fermentum 222), and one yeast strain (Saccharomyces cerevisiae H5S5K23) were used throughout this study. All strains were isolated from Ghanaian cocoa bean heap fermentations (6, 10). The A. pasteurianus 386B, L. fermentum 222, and S. cerevisiae H5S5K23 strains have been used for experimental starter culture-added on-farm cocoa bean fermentations before (26). The strains were stored at 80 C in mannitol-yeast extract-peptone (MYP) medium (2.5% [wt/vol] D-mannitol, 0.5% [wt/ vol] yeast extract, and 0.3% [wt/vol] bacteriological peptone [Oxoid, Basingstoke, United Kingdom]) (for AAB), in de Man-Rogosa-Sharpe (MRS) medium (Oxoid) (for LAB), or in glucose-yeast extract medium (GY; 2.0% [wt/vol] glucose and 0.5% [wt/vol] yeast extract) (for yeasts), in all cases supplemented with 25% (vol/vol) glycerol as a cryoprotectant. A cocoa pulp simulation medium (CPSM) for AAB was used for the AAB monoculture fermentation experiments. It consisted of the following ingredients: ethanol, 10 g liter 1 ; lactic acid, 10 g liter 1 ; granulated yeast extract, 10 g liter 1 ; neutralized soya peptone (Oxoid), 5 g liter 1 ; and Tween 80, 1 ml liter 1 ; ph 4.5 (16). In all of these fermentation experiments, 10 g liter 1 D-mannitol was added to CPSM. CPSM for LAB was used for some monoculture fermentations and for the coculture fermentation experiments. It consisted of the following ingredients: glucose, 25 g liter 1 ; fructose, 25 g liter 1 ; citric acid, 10 g liter 1 ; granulated yeast extract, 5 g liter 1 ; neutralized soya peptone (Oxoid), 5 g liter 1 ; MgSO 4 7H 2 O, 0.5 g liter 1 ; MnSO 4 H 2 O, 0.2 g liter 1 ; and Tween 80, 1 ml liter 1 ;ph3.5(16). In some fermentation experiments, 10 g liter 1 lactic acid was added to CPSM-LAB. All chemicals were obtained from VWR International (Darmstadt, Germany) unless stated otherwise. Solid media were prepared by adding 1.5% (wt/vol) agar (Oxoid). Selective plating was performed on malt extract agar (MEA; Oxoid) plus chloramphenicol (100 mg liter 1 ; Sigma-Aldrich, St. Louis, MO) for yeast (incubation in a standard incubator at 30 C for 24 h), CPSM agar plus cycloheximide (100 mg liter 1 ; Sigma-Aldrich) for LAB (ph 5.5; incubation in a modular atmosphere-controlled system [MG anaerobic work station; Don Withley Scientific, West Yorkshire, United Kingdom] that was continuously sparged with a mixture of 80% nitrogen, 10% carbon dioxide, and 10% hydrogen [Air Liquide, Paris, France] at 37 C for 24 h), and MYP agar plus cycloheximide (100 mg liter 1 ; Sigma-Aldrich) and ampicillin (100 mg liter 1 ; Sigma-Aldrich) for AAB (ph 5.5; incubation in a standard incubator at 30 C for 48 h). Fermentation experiments. All fermentations were carried out in the appropriate cocoa pulp simulation media in 15-liter Biostat C fermentors (Sartorius AG, Melsungen, Germany) for 48 h. All fermentations were performed at 30 C, which is a temperature allowing good growth of AAB. Control of the ph and aeration was fermentation dependent (see below). Agitation was performed at 300 rpm. The inocula were prepared as follows. All strains were transferred from 80 C stock cultures. AAB strains were inoculated into MYP medium (100 ml) and incubated aerobically at 30 C for 24 h in a rotary shaker. Subsequently, the strains were propagated twice in CPSM (400 ml) for AAB (ph 4.5) to obtain final precultures. During the inoculum buildup, the transferred volume was always 5% (vol/vol). Inocula of L. fermentum 222 were prepared in MRS medium (10 ml) and incubated statically at 30 C for 12 h. Subsequently, the strain was propagated twice in CPSM (80 ml) for LAB (ph 5.5) to obtain final precultures. During this inoculum buildup, the transferred volume was always 1% (vol/vol). The S. cerevisiae H5S5K23 strain was inoculated into YG medium (50 ml) and incubated statically at 30 C for 24 h. Subsequently, the strain was propagated twice in CPSM (400 ml) for LAB (ph 5.5) to obtain final precultures. During this inoculum buildup, the transferred volume was always 5% (vol/vol). The final precultures were added to the fermentation vessels aseptically. Fermentation strategy to compensate for the evaporation of ethanol. A control fermentation experiment was carried out in CPSM for AAB without inoculation of an AAB strain to estimate the impact of ethanol evaporation on the fermentation dynamics. Therefore, the evaporation of ethanol was monitored through offline analysis of the ethanol concentrations as a function of time. This allowed us to calculate the amount of ethanol to be added through continuous supply during the fermentations to compensate for evaporation losses (Fig. 1). Monoculture fermentations. (i) Monoculture fermentations with AAB strains in CPSM for AAB with mannitol. Monoculture fermentations with all four cocoa-specific AAB strains (Acetobacter pasteurianus 386B, Acetobacter ghanensis LMG T, Acetobacter fabarum LMG T, and Acetobacter senegalensis 108B) were carried out in 8 liters of CPSM for AAB, which was supplemented with mannitol. Aerobic conditions were ensured during the fermentation experiments by continuously sparging the medium with 5 liters min 1 of air. Ethanol was added at a constant flow rate of 2.0 mm h 1 throughout the fermentation to compensate for evaporation losses. A constant ph of 4.5 was imposed and controlled automatically using 5.0 M solutions of NaOH and HCl. (ii) Monoculture fermentations with specific LAB, AAB, and yeast strains in CPSM for LAB. Monoculture fermentations with L. fermentum 222, A. pasteurianus 386B, A. ghanensis LMG T, and S. cerevisiae H5S5K23 were carried out in either 8 liters of CPSM for LAB (in the case of LAB and yeast) or CPSM for LAB supplemented with 10 g liter 1 of lactic acid to simulate the activity of LAB (in the case of AAB). During the first6hofthefermentations, no air sparging of the medium was applied (LAB, AAB, and yeast), and ethanol was added at a constant flow rate of 36.0 mm h 1 (to compensate for ethanol production by yeast and evaporation losses) to reach a final concentration of 10 g liter 1 (LAB and AAB). This was done to simulate anaerobic conditions and the activity of yeast that takes place during the initial phase of spontaneous cocoa bean fermentations. After 6 h, the medium was sparged with 5 liters min 1 of air, and ethanol was added at a constant flow rate of 2.0 mm h 1 to compensate for evaporation losses. A linear ph profile, mimicking the ph evolution from 3.5 to 4.5 in 48 h during spontaneous cocoa bean fermentation processes, was imposed and controlled automatically using 5.0 M solutions of NaOH and HCl. Biculture fermentations with specific LAB and AAB strains. Coculture fermentations of L. fermentum 222 (representing an optimal strain that is strictly heterofermentative, citric acid converting, and mannitol producing) and A. pasteurianus 386B (representative of an AAB strain that March 2014 Volume 80 Number 6 aem.asm.org 1849

3 Moens et al. FIG 1 Evaporation of ethanol (A) and its compensation (B) during fermentations in CPSM for AAB. Symbols: {, lactic acid;, ethanol. oxidizes ethanol [fast], lactic acid [fast], and mannitol [slow]) or A. ghanensis LMG T (representative of an AAB strain that oxidizes ethanol [fast], lactic acid [slow], and mannitol [slow]) were carried out in 8 liters of CPSM for LAB. During the first 6 h of the fermentations, no sparging of the medium with air was applied and ethanol was added continuously at a flow rate of 36.0 mm h 1 (to compensate for ethanol production by yeast and evaporation losses) to reach a final concentration of 10 g liter 1. After this 6-h period, the medium was sparged with 5 liters min 1 of air and ethanol was added at a constant flow rate of 2.0 mm h 1 to compensate for evaporation losses. A linear ph profile from 3.5 to 4.5 in 48 h was imposed and controlled automatically using 5.0 M solutions of NaOH and HCl. Triculture fermentations with specific LAB, AAB, and yeast strains. Coculture fermentations of L. fermentum 222, A. pasteurianus 386B, and S. cerevisiae H5S5K23, each representing an optimal strain, displaying interesting functional roles, and tested as a starter culture mix during on-farm fermentations before (26), were carried out in 8 liters of CPSM for LAB. During the first 6 h of the fermentations, no air sparging of the medium was applied to simulate the anaerobic conditions during the initial phase of spontaneous cocoa bean fermentations. Ethanol was added at a constant flow rate of 2.0 mm h 1 throughout the fermentation to compensate for evaporation losses. A linear ph profile from 3.5 to 4.5 in 48 h was imposed and controlled automatically using 5.0 M solutions of NaOH and HCl. During all fermentation experiments, temperature, ph, agitation speed, and flow rate of ethanol were controlled online (MFCS/win 2.1 software; Sartorius AG). Samples were withdrawn at regular time intervals for offline analysis. All fermentations were performed in duplicate. The results and figures presented are representative for both fermentations. Analysis of bacterial growth, carbohydrate and citric acid consumption, and metabolite production. (i) Determination of growth. During fermentation, growth (expressed in CFU) was quantified through plating of 10-fold serial dilutions of the samples in saline (0.85%, wt/vol, NaCl solution) on the appropriate selective agar media mentioned above. (ii) Determination of ethanol, acetic acid, and acetoin concentrations and qualitative determination of 2,3-butanediol. Concentrations of ethanol, acetic acid, and acetoin were determined with gas chromatography using a Focus gas chromatograph (Interscience, Breda, The Netherlands) equipped with a Stabilwax-DA column (Restek, Bellefonte, PA), a flame ionization detector, and an AS 3000 autosampler. Hydrogen (Air Liquide) was used as a carrier gas at a constant flow rate of 1 ml min 1 ; nitrogen (Air Liquide) was used as a make-up gas. The injector and detector temperatures were set at 240 C and 250 C, respectively. The column temperature program was 0 min, 40 C; 10 min, 140 C; 12 min, 230 C; and 22 min, 230 C. Cell-free fermentation samples (obtained through centrifugation at 4,168 g for 20 min at 4 C) were diluted (1:4) with a mixture composed of acetonitrile (Sigma-Aldrich), 1% (vol/vol) formic acid (Merck), and 0.2% (vol/vol) 1-butanol (internal standard; Merck); centrifuged (18,894 g, 15 min, 4 C); filtered (0.2- m filters; Minisart RC 4; Sartorius AG) prior to injection; and run with the appropriate external standards of ethanol, acetic acid, and acetoin before and after the samples. Injection was performed in split mode with a split ratio of 40:1; the injected volume was 1.0 l. Data processing was performed using the ChromCard for Windows (version 1.18) software (Thermo Electro Corp., Milan, Italy). All analyses were performed in triplicate. Determination of 2,3-butanediol was performed qualitatively by gas chromatography, as described above for ethanol, acetic acid, and acetoin. (iii) Determination of carbohydrate, mannitol, citric acid, and gluconic acid concentrations. Residual concentrations of glucose, fructose, and mannitol were determined through high-performance anion-exchange chromatography (HPAEC) with pulsed amperometric detection (PAD), using an ICS3000 chromatograph (Dionex, Sunnyvale, CA) equipped with a Carbopac PA10 column, as described previously (18). Quantification was performed by standard addition; concentrations with standard deviations of the carbohydrates and mannitol were calculated as described previously (27). Therefore, four standard solutions with the following compositions were made: ultrapure water (solution A); 0.1 g liter 1 glucose, fructose, and mannitol (solution B); 0.2 g liter 1 glucose, fructose, and mannitol (solution C); and 0.3 g liter 1 glucose, fructose, and mannitol (solution D). Residual concentrations of citric acid and gluconic acid were measured by HPAEC with conductivity under ion suppression (CIS), using an ICS3000 chromatograph (Dionex) equipped with an AS-19 column (Dionex), as described previously (18). Quantification was performed by standard addition; concentrations with standard deviations of citric acid and gluconic acid were calculated as described previously (27). Therefore, four standard solutions with the following compositions were made: ultrapure water (solution A); 0.1 g liter 1 citric acid and 5.0 g liter 1 gluconic acid (solution B); 0.2 g liter 1 citric acid and 10.0 g liter 1 gluconic acid (solution C); and 0.3 g liter 1 citric acid and 15.0 g liter 1 gluconic acid (solution D). (iv) Determination of lactic acid concentrations. Concentrations of lactic acid were determined through high-performance liquid chromatography (HPLC), as described previously, except for the use of 5 mm H 2 SO 4 as the mobile phase (17). External standards were used for quantification. All samples were analyzed in triplicate. Determination of substrate consumption rates. To be able to compare the substrate consumption rates (mm h 1 ) of different AAB strains, those of ethanol and lactic acid were determined for the monoculture fermentations with the AAB strains in CPSM for AAB with mannitol and for the monoculture fermentations with A. pasteurianus 386B and A. ghanensis LMG T in CPSM for LAB. These substrate consumption rates were calculated, based on the time of substrate depletion, by following modeling of the bacterial growth and substrate consumption as de aem.asm.org Applied and Environmental Microbiology

4 Simulated Cocoa Pulp Fermentations scribed previously (27). To check the reproducibility of the fermentations, the monoculture fermentation with A. pasteurianus 386B in CPSM for AAB with mannitol was performed in triplicate. The relative standard deviations of the ethanol and lactic acid consumption rates of all three replicates were equal to 6.2% and 4.1%, respectively. CR. Carbon recovery (CR; expressed in percentages) was calculated by dividing the total amount of carbon recovered in the metabolites by the total amount of carbon present in the energy sources at the time the acetic acid concentrations were maximal. Also, the maximal concentration of acetoin produced throughout fermentation was taken into account for the calculation of the carbon recovery. Since carbon dioxide and 2,3-butanediol, an end metabolite of pyruvate catabolism of L. fermentum 222, were not measured quantitatively during the fermentations, the following assumptions were made for the calculations of the carbon recoveries. When oxidizing lactic acid, AAB will first convert lactic acid into pyruvate, which is then decarboxylated into acetaldehyde, with the latter being further oxidized into acetic acid. Thus, the oxidation of 1 mol lactic acid results in the production of 1 mol carbon dioxide and 1 mol acetic acid (28). Therefore, for each mole of lactic acid oxidized, the production of 1 mol of carbon dioxide was taken into account for the carbon balance of the fermentation experiments. However, AAB can also oxidize lactic acid into acetoin. In this pathway, 2 mol lactic acid is oxidized into 2 mol carbon dioxide and 1 mol acetoin (29). Therefore, for each mole of acetoin produced, the production of 2 mol carbon dioxide was taken into account for the carbon balance of the fermentation experiments. Further, L. fermentum 222, which is a strictly heterofermentative LAB strain, produces 1 mol carbon dioxide out of each mole of glucose that is converted into lactic acid or lactic acid and acetic acid (17). Moreover, when citric acid is used, equimolar amounts of carbon dioxide and pyruvate, which is further metabolized, are produced. Therefore, for each mole of glucose and citric acid converted by L. fermentum 222, 1 mol carbon dioxide and 1 mol pyruvate, respectively, was taken into account for the carbon balance. The carbon dioxide produced by the glucose and fructose metabolism of S. cerevisiae H5S5K23 was also taken into account for the calculations (30). Concerning ethanol oxidation by AAB, 1 mol ethanol is oxidized into 1 mol acetic acid. RESULTS Fermentation strategy to compensate for the evaporation of ethanol. Control fermentation experiments in CPSM for AAB revealed the occurrence of evaporation losses of 2.0 mm h 1 (Fig. 1A). If ethanol was continuously added to the fermentation vessels at a rate of 2.0 mm h 1, the ethanol concentrations stayed constant (Fig. 1B). During these control fermentation experiments, the lactic acid concentrations also were constant (Fig. 1A and B). Monoculture fermentations. (i) Monoculture fermentations with AAB strains in CPSM for AAB with mannitol. The cocoaspecific A. pasteurianus 386B, A. ghanensis LMG T, A. senegalensis 108B, and A. fabarum LMG T strains grew on ethanol (all four AAB strains), lactic acid (all four AAB strains), and mannitol (A. pasteurianus 386B and A. ghanensis LMG T )in CPSM for AAB with mannitol. During these fermentations, A. pasteurianus 386B oxidized ethanol and lactic acid from the start of the fermentation. Both ethanol and lactic acid were oxidized at a high rate (12.15 mm h 1 and 8.43 mm h 1, respectively). They were depleted after 18 h and 24 h of fermentation, respectively. Ethanol was oxidized into acetic acid, whereas the oxidation of lactic acid resulted in the production of acetoin. Acetobacter ghanensis LMG T oxidized ethanol completely into acetic acid after 24 h of fermentation. The oxidation of lactic acid (2.35 mm h 1 ) by this strain was slower than the oxidation of ethanol (11.47 mm h 1 ), and lactic acid was not depleted after 48 h of fermentation. The oxidation of lactic acid resulted primarily in the production of acetoin; however, small amounts of acetic acid also were produced from this substrate. The oxidation of mannitol into fructose by A. pasteurianus 386B and A. ghanensis LMG T was not as profound as the oxidation of ethanol and lactic acid, and it only occurred at the end of the fermentation, when ethanol was depleted. This depletion of ethanol from the fermentation medium also resulted in a substantial overoxidation of acetic acid during the fermentation experiments with A. pasteurianus 386B and A. ghanensis LMG T (Fig. 2A and B and Table 1). Compared to ethanol oxidation for A. pasteurianus 386B and A. ghanensis LMG T, ethanol was oxidized slower during the fermentations with A. fabarum LMG T (8.09 mm h 1 ) and A. senegalensis 108B (4.25 mm h 1 ). Ethanol was oxidized completely into acetic acid after 30 h in both fermentations. Lactic acid was rapidly oxidized by A. senegalensis 108B (7.54 mm h 1 ) and completely converted into acetoin after 24 h of fermentation. Acetobacter fabarum LMG T oxidized lactic acid (3.03 mm h 1 ) mainly into acetoin; however, small amounts of acetic acid also were produced from this substrate. Lactic acid was depleted after 48 h of fermentation. Acetobacter senegalensis 108B and A. fabarum LMG T did not oxidize mannitol into fructose. Both strains overoxidized acetic acid once ethanol was depleted (Fig. 2C and D and Table 1). Acetobacter pasteurianus 386B and A. ghanensis LMG T were the only two cocoa-specific AAB strains that performed a fast oxidation of ethanol combined with the capability of oxidizing both lactic acid and mannitol. Therefore, these strains were used for further mono- and coculture fermentation experiments in CPSM for LAB. (ii) Monoculture fermentations with A. pasteurianus 386B or A. ghanensis LMG T in CPSM for LAB. The cocoa-specific strains A. pasteurianus 386B (representative of an AAB strain with beneficial ethanol and lactic acid oxidation kinetics) and A. ghanensis LMG T (representative of an AAB strain with beneficial ethanol oxidation kinetics) grew on ethanol and lactic acid in CPSM for LAB. Acetobacter pasteurianus 386B first oxidized lactic acid (8.51 mm h 1 ) into acetoin from the start of the fermentation. After 6 h of fermentation, A. pasteurianus 386B oxidized ethanol (13.04 mm h 1 ) into acetic acid completely. Lactic acid oxidation (4.06 mm h 1 )bya. ghanensis LMG T did not occur during this initial period but started when ethanol was oxidized (6.50 mm h 1 ) into acetic acid. Acetobacter ghanensis LMG T did not grow on glucose, fructose, or citric acid, whereas A. pasteurianus 386B oxidized glucose into gluconic acid at the end of the fermentation when ethanol was depleted (Fig. 3C and D and Table 2). (iii) Monoculture fermentations with L. fermentum 222 in CPSM for LAB. The cocoa-specific and strictly heterofermentative L. fermentum 222 grew on glucose in CPSM for LAB. Fructose was used as an alternative external electron acceptor and was reduced to mannitol. Glucose was fermented into lactic acid and acetic acid. Citric acid was used to provide an additional source of pyruvate, which was converted into acetic acid and acetoin; qualitative gas chromatography analysis revealed that 2,3-butanediol was not produced. Citric acid was depleted after 12 h of fermentation. This coincided with a sharp decrease of the cell counts (Fig. 3A and Table 3). (iv) Monoculture fermentations with S. cerevisiae H5S5K23 in CPSM for LAB. The cocoa-specific S. cerevisiae H5S5K23 strain March 2014 Volume 80 Number 6 aem.asm.org 1851

5 Moens et al. FIG 2 Growth of and substrate consumption and metabolite production by Acetobacter pasteurianus 386B (A), Acetobacter ghanensis LMG T (B), Acetobacter fabarum LMG T (C), and Acetobacter senegalensis 108B (D) in CPSM for AAB with added mannitol. Symbols:, ethanol; {, lactic acid; Œ, fructose;, mannitol;, acetic acid;, acetoin; Œ, growth. Downloaded from grew on glucose and fructose in CPSM for LAB. Glucose was initially fermented into ethanol, after which fructose also was fermented. Saccharomyces cerevisiae H5S5K23 did not grow on citric acid (Fig. 3B and Table 3). Only trace amounts of acetoin were produced during this fermentation. Biculture fermentations with L. fermentum 222 and A. pasteurianus 386B or A. ghanensis LMG T in CPSM for LAB. During the coculture fermentations, the cocoa-specific and strictly heterofermentative L. fermentum 222 strain grew in CPSM for LAB as described above. Both Acetobacter strains tested survived the first 6 h of fermentation, when no sparging of the fermentation medium with air was applied. Afterwards, A. pasteurianus 386B and A. ghanensis LMG T oxidized ethanol into acetic acid. Lactic acid, produced throughout the fermentation by L. fermentum 222, was simultaneously oxidized into acetoin and acetic acid. This was fastest in the case of A. pasteurianus 386B. Lactobacillus fermentum 222 did not produce acetoin from citric acid; qualitative gas chromatography revealed the production of 2,3-butanediol by L. fermentum 222 during these fermentations. Mannitol, produced by L. fermentum 222, was not oxidized by the Acetobacter strains during these fermentations. Acetobacter pasteurianus 386B did not oxidize glucose into gluconic acid, whereas A. ghanensis LMG T did toward the end of the fermentation, when ethanol was nearly depleted (Fig. 4A and B and Tables 2 and 3). Triculture fermentations with S. cerevisiae H5S5K23, L. fermentum 222, and A. pasteurianus 386B in CPSM for LAB. During the triculture fermentation, S. cerevisiae H5S5K23 grew on fructose and glucose, while L. fermentum 222 only grew on glu- on October 30, 2018 by guest TABLE 1 Metabolite production by cocoa-specific AAB strains during monoculture fermentation experiments c Strain Mean consumption SD (mm) of indicated energy source (at maximal acetic acid concn) Mean production SD (mm) of indicated metabolite (at maximal acetic acid concn) Ethanol Lactic acid Mannitol Acetoin b Acetic acid CO 2 a Fructose Carbon recovery (%) A. pasteurianus 386B A. ghanensis LMG T A. fabarum LMG T A. senegalensis 108B a Theoretical estimated carbon dioxide production. b Maximal concentration of acetoin produced throughout the fermentation. c Experiments were performed in a cocoa pulp simulation medium for acetic acid bacteria (CPSM-AAB) containing ethanol, lactic acid, and mannitol as the added energy sources aem.asm.org Applied and Environmental Microbiology

6 Simulated Cocoa Pulp Fermentations FIG 3 Growth of and substrate consumption and metabolite production by Lactobacillus fermentum 222 (A), Saccharomyces cerevisiae H5S5K23 (B), Acetobacter pasteurianus 386B (C), and Acetobacter ghanensis LMG T (D) in CPSM for LAB. Symbols:, ethanol; {, lactic acid;, acetic acid;, acetoin;, mannitol; o, glucose; Œ, fructose;, citric acid;, gluconic acid; Œ, AAB growth;, LAB growth;, yeast growth. cose. The cocoa-specific yeast strain fermented both carbohydrates into ethanol. The cocoa-specific LAB strain fermented the residual glucose into acetic acid and lactic acid. Fructose was used by the LAB strain as an alternative external electron acceptor, resulting in the production of mannitol. Citric acid was depleted by the LAB strain after 12 h of fermentation and converted into acetic acid and 2,3-butanediol, as revealed by gas chromatography analysis. After the initial yeast and LAB activities, the cocoa-specific A. pasteurianus strain oxidized ethanol produced by the yeast strain into acetic acid and lactic acid produced by the LAB strain into acetoin. No oxidation of mannitol, produced by L. fermentum 222, occurred. Acetobacter pasteurianus 386B did not oxidize glucose into gluconic acid during these fermentations (Fig. 4C and Tables 2 and 3). DISCUSSION This is the first study that investigated the kinetics of ethanol, lactic acid, and mannitol oxidation by various Acetobacter spe- TABLE 2 Metabolite production by cocoa-specific AAB strains during mono- and coculture fermentation experiments h Mean consumption SD (mm) of indicated energy source (at maximal acetic acid concn) Mean production SD (mm) of indicated metabolite (at maximal acetic acid concn) Strain Ethanol Lactic acid Mannitol Glucose Fructose Acetoin f Acetic acid CO 2 e Gluconic acid Carbon recovery (%) A. pasteurianus 386B a NR g A. pasteurianus 386B b d d NR NR NR d A. pasteurianus 386B c d d NR NR NR d A. ghanensis LMG Ta NR A. ghanensis LMG Tb d d NR NR d a Monoculture fermentation experiments in CPSM-LAB containing glucose, fructose, citric acid, ethanol, and lactic acid. b Coculture fermentation experiments of the respective AAB strains and L. fermentum 222 in CPSM-LAB containing glucose, fructose, citric acid, and ethanol. c Coculture fermentation experiments of A. pasteurianus 386B, L. fermentum 222, and S. cerevisiae H5S5K23 in CPSM-LAB containing glucose, fructose, and citric acid. d Calculated values based on the analytical measurements and metabolisms of the respective strains. e Theoretical estimated carbon dioxide production. f Maximal concentration of acetoin produced throughout the fermentation. g NR, not relevant. h Mono- and coculture fermentation experiments, with Lactobacillus fermentum 222 and/or Saccharomyces cerevisiae H5S5K23, were performed in a cocoa pulp simulation medium for lactic acid bacteria (CPSM-LAB) containing glucose, fructose, and citric acid as the added energy sources. March 2014 Volume 80 Number 6 aem.asm.org 1853

7 Moens et al. TABLE 3 Metabolite production by cocoa-specific Saccharomyces cerevisiae and Lactobacillus fermentum strains during mono- and coculture fermentation experiments i Strain Mean consumption SD (mm) of indicated energy source (at the maximal concn of acetic acid) cies occurring in fermenting cocoa pulp-bean mass through monoculture laboratory fermentations in cocoa pulp simulation media. Moreover, this study incorporated evaporation losses of ethanol due to the high fermentation temperature and aeration, which was not done before. Further, bi- and triculture fermentations unraveled the functional role of these cocoa-specific species (representing different oxidation patterns of ethanol and lactic Mean production SD (mm) of indicated metabolite (at the maximal concn of acetic acid) Glucose Fructose Citric acid Acetoin g Lactic acid Acetic acid Ethanol Mannitol CO 2 f Carbon recovery (%) L. fermentum 222 a NR h L. fermentum 222 b NR e e NR L. fermentum 222 c NR e e NR L. fermentum 222 d e e NR e e NR S. cerevisiae H5S5K23 a NR NR NR S. cerevisiae H5S5K23 d e e NR NR NR NR e NR a Monoculture fermentation experiments in CPSM-LAB containing glucose, fructose, and citric acid. b Coculture fermentation experiments with L. fermentum 222 and A. pasteurianus 386B in CPSM-LAB containing glucose, fructose, and citric acid. c Coculture fermentation experiments with L. fermentum 222 and A. ghanensis LMG T in CPSM-LAB containing glucose, fructose, and citric acid. d Coculture fermentation experiments with S. cerevisiae H5S5K23, L. fermentum 222, and A. pasteurianus 386B in CPSM-LAB containing glucose, fructose, and citric acid. e Calculated values based on the analytical measurements and metabolisms of the respective strains. f Theoretical estimated carbon dioxide production. g Maximal concentration of acetoin produced throughout the fermentation. h NR, not relevant. i Mono- and coculture fermentation experiments, with Acetobacter pasteurianus 386B or Acetobacter ghanensis LMG T, were performed in a cocoa pulp simulation medium for lactic acid bacteria (CPSM-LAB) containing glucose, fructose, and citric acid as the added energy sources. acid) in shaping the course of the community dynamics of successful cocoa bean fermentations and highlighted the indispensable interactions occurring during spontaneous cocoa bean fermentations, particularly with respect to the dominance of A. pasteurianus 386B together with selected strains of L. fermentum and S. cerevisiae. As obligate aerobic bacteria, AAB consume oxygen during ox- Downloaded from on October 30, 2018 by guest FIG 4 Growth of and substrate consumption and metabolite production by a coculture of Lactobacillus fermentum 222 and Acetobacter pasteurianus 386B (A); Lactobacillus fermentum 222 and Acetobacter ghanensis LMG T (B); and Saccharomyces cerevisiae H5S5K23, Lactobacillus fermentum 222, and Acetobacter pasteurianus 386B (C) in CPSM for LAB. Symbols:, ethanol; {, lactic acid;, acetic acid;, acetoin;, mannitol; o, glucose; Œ, fructose;, citric acid;, gluconic acid; Œ, AAB growth;, LAB growth;, yeast growth aem.asm.org Applied and Environmental Microbiology

8 Simulated Cocoa Pulp Fermentations idation of ethanol and lactic acid into acetic acid and/or acetoin (29, 31, 32). This is why aerobic conditions have to be maintained during fermentations with AAB, which is generally accomplished by continuous dispersion of oxygen into the fermentation medium. However, only part of the oxygen is consumed by the AAB; the major part escapes from the fermentor through the gas exit, which is accompanied with a significant loss of volatile components (33). During the present study, an open fermentation system was used, whereby a constant flow of ethanol was fed to the fermentation medium to compensate for evaporation losses. The temperature thereby was kept constant at 30 C, allowing a constant rate of ethanol evaporation; this temperature represents optimal growth of AAB. Concerning oxidation of ethanol into acetic acid and the oxidation of lactic acid into acetoin or acetoin plus acetic acid, two groups of cocoa-specific AAB strains could be distinguished based on monoculture fermentations in CPSM for AAB with mannitol. One group, represented by A. pasteurianus 386B and A. ghanensis LMG T, carried out a fast oxidation of ethanol into acetic acid coupled with a fast oxidation of lactic acid into acetoin (A. pasteurianus 386B) or a slow oxidation of lactic acid into acetoin plus acetic acid (A. ghanensis LMG T ). A second group, represented by A. senegalensis 108B and A. fabarum LMG T, carried out a slow oxidation of ethanol into acetic acid coupled to a fast oxidation of lactic acid into acetoin (A. senegalensis 108B) or a slow oxidation of lactic acid into acetoin plus acetic acid (A. fabarum LMG T ). The oxidation of ethanol into acetic acid is catalyzed by two sequential catalytic reactions of membranebound, pyrroloquinoline quinone (PQQ)-dependent alcohol dehydrogenase, which oxidizes ethanol into acetaldehyde, and aldehyde dehydrogenase, which oxidizes acetaldehyde into acetic acid (31, 32, 34 39). Lactic acid oxidation starts with the conversion of lactic acid into pyruvate by a lactate dehydrogenase. In the case of oxidation into acetic acid, pyruvate is further decarboxylated to acetaldehyde by a pyruvate decarboxylase, and acetaldehyde is finally oxidized into acetic acid by an aldehyde dehydrogenase (28, 40). In the case of oxidation of lactic acid into acetoin, two pathways are possible. One of these pathways is identical to the formation of acetoin in yeast (41 43). In this pathway, pyruvate is decarboxylated by a pyruvate decarboxylase, whereby bound active acetaldehyde undergoes carboligation with another molecule of acetaldehyde to produce acetoin (29, 42, 44). An alternative pathway involves decarboxylation of -acetolactate, which is formed out of two molecules of pyruvate (29). In all cases, the activity of the dehydrogenases during ethanol and lactic acid oxidation generates electrons that enter the respiratory chain localized in the cytoplasmic membrane. They are transferred to ubiquinone as intermediate electron acceptors, reducing it to ubiquinol. Ubiquinol is subsequently reoxidized into ubiquinone by the action of a quinol oxidase that reduces oxygen to water (32, 37, 38). Hence, acetic acid and acetoin are the major end metabolites of cocoa bean fermentation, reflecting the final activity of AAB (energy production) during this process. During monoculture fermentations in CPSM for AAB with A. pasteurianus 386B in the study of Lefeber et al. (16), both ethanol and lactic acid were oxidized into acetic acid. However, the oxidation of the initial large amounts of ethanol present in the medium did not result in the production of significant amounts of acetic acid. This can be explained by the fact that ethanol was not added to the fermentation medium to compensate for evaporation losses. Hence, the evaporation of ethanol resulted in a sharp decrease of the ethanol concentration from the start of the fermentations without a concomitant increase in the concentration of acetic acid. The present study indicates that an AAB metabolism that is forced to a fast initial oxidation of ethanol and lactic acid could be an adaptive advantage. Indeed, during spontaneous cocoa bean fermentations, yeast and LAB grow well under the initial anaerobic to microaerophilic conditions of the cocoa pulp-bean mass and convert the available carbohydrates (glucose and fructose) into a range of metabolic end products, among which are ethanol and lactic acid, which serve as energy sources for AAB (2, 6, 10). However, although AAB will prevail in spontaneous cocoa bean fermentations under aerobic conditions, yeasts such as S. cerevisiae can also oxidize lactic acid into pyruvate (45 47). This suggests that lactic acid consumption by yeast, when glucose is depleted in the cocoa pulp and aerobic conditions are prevailing, can be a competitive threat for lactic acid oxidation by AAB. Acetobacter pasteurianus 386B and A. ghanensis LMG T were the only two strains tested that oxidized mannitol into fructose. This oxidation only took place when ethanol was no longer present in the medium and when its oxidation into acetic acid halted. Oxidation of mannitol into fructose is catalyzed by a membrane-bound PQQ-dependent polyol dehydrogenase, as is the case for Gluconobacter oxydans strains (32, 48). Mannitol oxidation by A. pasteurianus 386B and A. ghanensis LMG T will give these strains an energetic advantage, since this oxidation reaction also is coupled to their respiratory chain (34). However, it can be questioned whether this metabolic activity itself is beneficial for the cocoa bean fermentation process as a whole, since this conversion generates fermentable carbohydrates at the end of the fermentation process which subsequently can be used by undesirable fungi and spoilage bacteria that proliferate when the cocoa bean fermentation actually comes to a finish. Both A. pasteurianus 386B and A. ghanensis LMG T, which are representative of the group of AAB strains that performs a fast oxidation of ethanol into acetic acid, survived and grew at the initial low ph of 3.5, as shown through mono- and biculture fermentations in CPSM for LAB. Despite AAB growing optimally at ph 5.0 to 6.5, previous studies have proven that AAB can grow at low ph too (49 51). The results of the mono- and biculture fermentations in CPSM for LAB support the hypothesis that AAB, prevailing throughout the main and end courses of a natural cocoa bean fermentation process, survive the initial environmental conditions of low ph and anaerobiosis and originate from the initial contamination of the cocoa pulp after opening of the pods. Acetobacter pasteurianus 386B oxidized glucose into gluconic acid at the end of a monoculture fermentation in CPSM for LAB. Similarly, gluconic acid was produced in a coculture fermentation with A. ghanensis LMG T and L. fermentum 222 in CPSM for LAB. In both cases, glucose oxidation occurred when ethanol was depleted. The oxidation of glucose did not occur during the monoculture fermentation with A. ghanensis LMG T during either the biculture fermentation with A. pasteurianus 386B and L. fermentum 222 or the triculture fermentation with A. pasteurianus 386B, L. fermentum 222, and S. cerevisiae H5S5K23 in CPSM for LAB. In all cases, ethanol oxidation occurred throughout the fermentation. Although other studies have proven that A. fabarum LMG T and A. senegalensis 108B are also able to oxidize glucose into gluconic acid (23, 24, 52), it is unlikely that AAB will do this in the competitive ecosystem of successful spontaneous March 2014 Volume 80 Number 6 aem.asm.org 1855

9 Moens et al. cocoa bean fermentations, during which glucose and fructose are depleted by yeast and LAB and lactic acid and ethanol are available as energy sources for AAB (25). However, when too much glucose is left in the fermenting cocoa pulp-bean mass, for instance, because of a nonoptimal succession of microbial activities, it can lead to the production of gluconic acid late in the fermentation process (22). The production of gluconic acid in an early phase of spontaneous cocoa bean fermentation results from the metabolic activity of Enterobacteriaceae (25). During the triple fermentation in CPSM for LAB, A. pasteurianus 386B oxidized lactic acid and ethanol, initially produced by L. fermentum 222 and S. cerevisiae H5S5K23, respectively, into acetic acid and acetoin. Also, mannitol was produced by L. fermentum 222. However, further oxidation of mannitol into fructose did not occur, which is a general observation during spontaneous cocoa bean fermentations (6, 20, 25). However, A. pasteurianus may be responsible for a decrease of the mannitol concentration and a concomitant increase of the fructose concentration in cocoa pulp during spontaneous cocoa bean fermentation (22). To conclude, the detailed kinetics of acetic acid production by cocoa-specific AAB strains revealed a deeper insight into their ethanol and lactic acid oxidation potential and acetic acid and acetoin production patterns. These results indicate the existence of two ethanol and lactic acid oxidation patterns among the cocoa-specific Acetobacter strains tested, characterized by either a fast or slow ethanol oxidation into acetic acid and always coupled to a fast or slow oxidation of lactic acid. Coculture fermentations of A. pasteurianus 386B and A. ghanensis LMG T with specific strains of L. fermentum and S. cerevisiae highlighted indispensable interactions occurring during spontaneous cocoa bean fermentations. These results revealed the importance of the metabolites lactic acid and ethanol as substrates for AAB during interactions with yeast and LAB. The results also indicated the minor and undesirable role of glucose and mannitol as substrates for AAB in the competitive ecosystem of cocoa bean fermentations. Moreover, the results indicated that the oxidation of these substrates by AAB is mediated by the presence of ethanol in the fermenting cocoa pulp-bean mass. In particular, this study showed that A. pasteurianus 386B displayed a specific and beneficial lactic acid and ethanol oxidation pattern, oxidized mannitol into fructose, oxidized glucose into gluconic acid, and was able to survive and grow under suboptimal ph and aeration conditions. Further, it tolerated high concentrations of ethanol and acetic acid and interacted beneficially with S. cerevisiae and L. fermentum strains, explaining its dominance during spontaneous cocoa bean fermentations. All of these data will contribute to an optimal choice and use of starter cultures for improved cocoa bean fermentation processes. ACKNOWLEDGMENTS Parts of this research were funded by the Research Council of the Vrije Universiteit Brussel (OZR, GOA, SRP, and IOF projects), the Research Foundation Flanders (FWO-Vlaanderen), the Hercules Foundation, the Institute for the Promotion of Innovation through Science and Technology in Flanders (in particular the IWT project), and Barry Callebaut N.V. F.M. is the recipient of a predoctoral fellowship of the FWO- Vlaanderen. REFERENCES 1. Thompson SS, Miller KB, Lopez AS Cocoa and coffee, p In Doyle MP, Beuchat LR, Montville TJ (ed), Food microbiology: fundamentals and frontiers, 2nd ed. ASM Press, Washington, DC. 2. Schwan RF, Wheals AE The microbiology of cocoa bean fermentation and its role in chocolate quality. Crit. Rev. Food Sci. Nutr. 44: De Vuyst L, Lefeber T, Papalexandratou Z, Camu N The functional role of lactic acid bacteria in cocoa bean fermentation, p In Mozzi F, Raya RR, Vignolo GM (ed), Biotechnology of lactic acid bacteria: novel applications. Wiley-Blackwell, Ames, IA. 4. Lima LJR, Almeida MH, Nout MJR, Zwietering MH Theobroma cacao L., the food of the gods : quality determinants of commercial cocoa beans, with particular reference to the impact of fermentation. Crit. Rev. Food Sci. Nutr. 51: / Ardhana MM, Fleet G The microbial ecology of cocoa bean fermentations in Indonesia. Int. J. Food Microbiol. 86: Camu N, De Winter T, Verbrugghe K, Cleenwerck I, Vandamme P, Takrama JS, Vancanneyt M, De Vuyst L Dynamics and biodiversity of populations of lactic acid bacteria and acetic acid bacteria involved in spontaneous heap fermentation of cocoa beans in Ghana. Appl. Environ. Microbiol. 73: Nielsen DS, Teniola OD, Ban-Koffi L, Owusu M, Andersson TS, Holzaphel WH The microbiology of Ghanaian cocoa fermentations analysed using culture-dependent and culture-independent methods. Int. J. Food Microbiol. 114: Jespersen L, Nielsen DS, Hønholt S, Jakobsen M Occurrence and diversity of yeasts involved in fermentation of West African cocoa beans. FEMS Yeast Res. 5: Nielsen DS, Hønholt S, Tano-Debrah K, Jespersen J Yeast populations associated with Ghanaian cocoa fermentations analyzed using denaturing gradient gel electrophoresis (DGGE). Yeast 22: http: //dx.doi.org/ /yea Daniel H-M, Vrancken G, Takrama JF, Camu N, De Vos P, De Vuyst L Yeast diversity of Ghanaian cocoa bean heap fermentations. FEMS Yeast Res. 9: Papalexandratou Z, De Vuyst L Assessment of the yeast species composition of cocoa bean fermentations in different cocoa-producing regions using denaturing gradient gel electrophoresis. FEMS Yeast Res. 11: Illeghems K, De Vuyst L, Papalexandratou Z, Weckx S Phylogenetic analysis of a spontaneous cocoa bean fermentation metagenome reveals new insights into its bacterial and fungal community diversity. PLoS One 7:e Pereira GVDM, Magalhães KT, de Almeida EG, Coelho IDS, Schwan RF Spontaneous cocoa bean fermentation carried out in a noveldesign stainless steel tank: influence on the dynamics of microbial populations and physical-chemical properties. Int. J. Food Microbiol. 16: Crafack M, Mikkelsen MB, Saerens S, Knudsen M, Blennow A, Lowor S, Takrama J, Swiegers JH, Petersen GB, Heimdal H, Nielsen DS Influencing cocoa flavour using Pichia kluyveri and Kluyveromyces marxianus in a defined mixed starter culture for cocoa fermentation. Int. J. Food Microbiol. 167: Papalexandratou Z, Lefeber T, Bahrim B, Lee OS, Daniel HM, De Vuyst L Hanseniaspora opuntiae, Saccharomyces cerevisiae, Lactobacillus fermentum, and Acetobacter pasteurianus predominate during wellperformed Malaysian cocoa bean box fermentations, underlining the importance of these microbial species for a successful cocoa bean fermentation process. Food Microbiol. 35: Lefeber T, Janssens M, Camu N, De Vuyst L Kinetic analysis of strains of lactic acid bacteria and acetic acid bacteria in cocoa pulp simulation media toward development of a starter culture for cocoa bean fermentation. Appl. Environ. Microbiol. 76: aem.asm.org Applied and Environmental Microbiology