Cadmium (Cd) levels in chocolate are a significant concern due to the toxic nature of this trace metal. This study presents a bioremediation treatment using Bacillus xiamenensis Luk70´s biomass in the post-harvesting process. The effect of the percentage of fermented beans, treatment time, nib pH, and concentration of bacteria applied to the reduction of Cd concentration was also studied, in addition to a cost analysis of pilot-scale bacteria production. Luk70 biomass with a 1:1 vegetative cell spore ratio applied to fermented cocoa beans at 90% can reduce nib Cd concentrations by a factor of 1.45 or 1.77, which to the best of our knowledge are the highest reductions in nib Cd concentration reductions reported to date without affecting the chocolate quality. The cost of producing Luk70 is estimated to be USD 0.56 per kg of wet cocoa beans, representing 14.4% of the current average price of Colombia´s wet cocoa beans. Although there are potential biosolutions for removing cadmium from cocoa, the economic costs and impact on the cocoa chain and producers must be considered.
Article
Cadmium bioremediation from cocoa with Bacillus xiamenensis Luk70 in post-harvesting, a technological and economic challenge to achieve
https://doi.org/10.21203/rs.3.rs-4829441/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
fine-flavor cocoa
Bacillus
chocolate
toxic
trace metal
microbial remediation
Colombia
Cadmium (Cd) is a toxic trace metal accumulated in plants, which could enter and accumulate in the human body through different food chains. Cadmium does not fulfill a biological function in plants or animals. In humans, it can cause renal tubular dysfunction and disrupt calcium metabolism leading to osteomalacia 1,2. Some regulatory agencies in various countries have set permissible limits for Cd in food to mitigate Cd exposure. For instance, the European Union (EU) has regulated the maximum allowed Cd levels for cocoa-derived products since 2019. These limits include 0.10 mg kg-1 for milk chocolate containing < 30% total dry cocoa solids; 0.30 mg kg-1 for chocolate with < 50% total dry cocoa solids and milk chocolate with ≥ 30% total dry cocoa solids; 0.8 mg kg-1 for chocolate with > 50% total dry cocoa solids and 0.6 mg kg-1 for cocoa powder sold to final consumers or used as an ingredient in sweetened cocoa powder 3. More regulations have been or are expected to be implemented in countries such as Australia, New Zealand, Russia, and member countries of the Southern Common Market (Mercosur, by its Spanish acronym). Finally, Proposition 65 in California mandates that chocolates containing elevated Cd concentrations include a warning on their packaging 4.
Cd in cocoa comes from the soil, from which the cocoa tree roots absorb it and is then translocated to the beans. Mucilage plays an essential role in the transport and storage of Cd within cocoa trees 5. Several studies have reported the Cd concentrations in cocoa-producing regions worldwide. Latin America and the Caribbean (LAC) show higher levels of Cd in their soils compared to those in Africa. This difference can be attributed to the relative age of LAC's soils, resulting in less reduction in Cd concentrations due to weather compared to African soils. 4. Furthermore, cocoa production areas situated near industrial or mining regions may be affected by Cd-contaminated irrigation water or flooding of riparian areas. Finally, the application of fertilizers to enhance soil productivity could act as another source of Cd contamination of soils. 4. In Colombia, the average soil Cd content is 1.43 mg kg-1 according to a study that sampled different cocoa production areas 6; a more recent study of Cd in soils of Arauca (the cocoa used in this study came from this region) reported values from 0.98 to 1.4 mg kg-17. The average soil Cd content in Colombia exceeds the threshold value of 1 mg kg-1 established for agricultural soils by countries such as Finland, Sweden, Belgium, and Austria 6. In this sense, mitigation strategies should be developed to decrease Cd concentration in both soils and cocoa beans, as it is crucial to mitigate the impact of regulations on the global commercialization of Colombian cocoa.
In recent years, several microbial remediation strategies for Cd have been proposed, mainly focused on soil decontamination in cocoa crops, that take advantage of the ability of specific bacterial genera (especially the genus Bacillus sp.) to detoxify metals through mechanisms such as resistance, valence transformation, biosorption, extracellular chemical precipitation, and efflux 2,8. To achieve this, the use of strains such as Bacillus cereus, Bacillus vietnamensis, and Azospirillum brasilense in different crops for reducing Cd levels has been reported 9–12. Furthermore, some studies have focused on removing cadmium from cocoa beans during postharvest, suggesting that conventional fermentation practices may decrease the Cd concentration in cocoa nibs by up to a factor of 1.3, which could be increased to 1.6 by using acetic acid and adjusting the temperature to 65°C 13. This reduction in Cd concentration in nibs has been attributed to a combination of high temperature (> 45°C) and acidic pH (< 5)13–17. Recently, a controlled cocoa fermentation with a mean temperature of 44.25°C and pH of 4.66 reached a mean nib Cd reduction throughout fermentation of 1.37 factor in TCS 01 cacao genotypes 18.
Bacillus xiamenensis Luk70 is a spore-forming Gram-positive bacteria isolated from the fermentation of Colombian cocoa beans 19. Previously, B. xiamenensis Luk70 has been reported as a plant-growth-promoting rhizobacteria (PGPR) and biocontrol agent of red dot disease in sugarcane at greenhouse experiments 20. It has also shown the capacity of zinc solubilization and promotion of drought tolerance in crops such as tomato, rice, sugarcane, and beans 21–25. Moreover, its ability to withstand high concentrations of lead (II), cadmium, chromium, arsenic, copper, nickel, and zinc in water-contaminated experiments has been reported 26. Luk70 had tolerance to high concentrations of Cd (12 mg kg− 1) and showed great potential for bioremediation of cadmium from cocoa beans in laboratory experiments. In this context, this study proposes a novel methodology for the in situ mitigation of Cd from cocoa beans during the post-harvest stage at a laboratory scale using biomass from B. xiamenensis Luk70. In addition, the ability of the different structures of Luk70 for Cd removal was studied. Subsequently, a biotechnological process was developed to produce a specific structure of B. xiamenensis Luk70´s with high activity for Cd bioremediation. Finally, a cost analysis was conducted from a pilot scale production for the active component of a potential bioproduct.
2.1. Microorganism
The strain used in these experiments has been identified as Bacillus xiamenensis Luk70 and is part of the collection of microorganisms associated with industrial cocoa fermentations 19. It was selected from a previous series of experiments, not reported here, for its ability to remove cadmium.
2.2. Bacillus xiamenensis Luk70´s submerged fermentation (SmF)
2.2.1. Inoculum production and culture medium
From a vial stored at -20°C, a suspension was made using 100 mL of sterile culture medium JM2 in a 500 mL Erlenmeyer flask sealed with a cotton stopper to allow gas exchange. This procedure was carried out five (5) times to obtain an inoculum volume of 450 mL, equivalent to 5% of the total volume of the subsequent stage. The flasks were incubated at 30°C for 24 hours, and their purity was assessed via Olympus BX53 microscope inspection of the fresh culture using phase contrast objectives. The culture medium used in this work was coded as JM2, a modification of a culture medium JM, standardized in previous works 27.
2.2.2. Fermentation in a laboratory-scale bioreactor
A sterile Sartorius Biostat B® model bioreactor containing 8500 mL of the culture medium JM2 was inoculated. Fermentations were conducted at 30°C, aeration of 1vvm and 150 rpm for 72 hours. Foam control was performed manually by adding silicone antifoam (Merck® Ref. 107743).
2.3. Effect of the type of biomass on the cadmium removal efficiency
Four types of Luk70 biomasses were selected for the assay: Biomass with a similar number of vegetative cells and spores, with more vegetative cells than spores, with only spores, and non-viable biomass. Control treatments included water and sterile JM2 broth supplemented with 6 ppm of cadmium. Biomasses were obtained as follows: i) Biomass with a similar number of vegetative cells and spores (long fermentation) were produced by fermentation following the protocol described on section 2.2. The fermented broth was centrifuged for 10 minutes at 7000 RPM, and the pellet was stored until further use. ii) Biomass with more vegetative cells than spores (short fermentation) were produced as before (section 2.2), yet the fermentation was terminated after 24 hours. iii) Biomass with only spores (spores) was produced by cellular lysis of Luk70 cells 28. iv) Non-viable biomass was obtained by sterilization in an autoclave at 121°C and 1.5 atm for 40 minutes (dead).
The biomass was inoculated individually in Erlenmeyer flasks with 200 mL of JM2 broth pH 4 supplemented with 6 ppm of cadmium (Cd). Each treatment was inoculated from stock until reaching an optical density (OD) of 1.0, measured by a spectrophotometer at 600 nm wavelength. Samples were collected at 0, 24, and 48 hours to measure the Cd content in the supernatant and the concentration of vegetative cells and spores (expressed as UFC mL− 1). The controls were sampled at 0 and 48 hours to measure the amount of Cd in the supernatant. All treatments were conducted in triplicates.
Samples were centrifuged for 10 minutes at 7000 rpm, followed by filtration using 0.22µm hydrophilic polyether sulfone (pes) membrane filters. Then, nitric acid (HNO3) 55% (v/v) was added and stored at 5°C until analysis. Cadmium levels were assessed according to the EPA 6010D/AOAC:2006 protocol. Cadmium removal rates (%) were calculated as shown in Eq. 1.
Where, \(\:{Cd}_{t0}\) is the Cd concentration (mg Kg− 1) in the supernatant at the initial time, and \(\:{Cd}_{tf}\) is the Cd concentration (mg Kg− 1) in the supernatant at the end of the experiment.
2.4. Kinetic analysis of Luk70´s submerged fermentation (SmF)
The kinetic analysis was focused on operational variables such as temperature, aeration, and agitation. Likewise, monitoring of indirect response variables including pH, dissolved oxygen, and weight loss (due to foam formation) was carried out using the Sartorius BioScada® software. Simultaneously, periodic samples from the fermentation broth were taken to determine dry biomass (g L− 1), vegetative cells and spore concentration (UFC mL− 1), and sugar concentration (sucrose, glucose and fructose) (mg mL− 1).
2.4.1. Reproducibility analysis of the fermentation kinetics of Luk70
Initially, an analysis of tolerance intervals was applied to verify that the range of values (upper and lower) of the indirect response variables evaluated in the fermentation process (pH and dissolved oxygen) throughout the two batches were adjusted to a normal distribution.
If the normality requirement was met, a graphical analysis of the response variables was applied, similarly to that from process control charts, and the graphs of individual observations and moving range (I-MR) were analyzed per batch since subgroups did not carry out the sampling. This allowed us not only to see differences in the central line and the limits of the surrogate response variables due to the implementation of the two fermentation batches over time, but also to calculate the short-term standard deviation (potential process capacity) and the long-term standard deviation (actual process capability) which are quantitative indicators of process acceptability. For this case, a short-term and long-term coefficient of variation was calculated for each fermentation batch using Equations 2 and 3.
| \(\:{CV}_{cp}=\frac{StDev}{Mean}\) | Equation 2 |
|---|---|
| \(\:{CV}_{lp}=\frac{StDev}{Mean}\) | Equation 3 |
Where: \(\:{CV}_{cp}\) is the short-term variation coefficient, and \(\:{CV}_{lp}\) is the long-term variation coefficient.
The CV between batches of the response variables spore concentration and vegetative cells concentration was also calculated. If the CV obtained was below 10% for the two batches, it was considered acceptable in the context of phenomena involving biological processes.
2.5 Effect of biomass concentration, fermentation time, and contact time over the in situ cadmium removal at lab-scale
2.5.1. Cocoa beans fermentation
By the end of the first cocoa harvesting season of 2021 (April), a large-scale pre-standardized industrial cocoa fermentation experiment was carried out at the Luker cocoa farm (Palestina-Santágueda, Palestina, Caldas), located at a latitude of 5° 4' N and longitude 75° 41' W and an altitude of 1050 MASL in Colombia, using a mixture of fine-flavor clones from Arauca as described in the experimental section in a previous report by Fernández-Niño et al. (2021) 29. Once harvested, the cocoa pods were transported from Arauca farms to Luker cocoa farm. For more information about clones, consult the international cocoa germplasm database (available at http://www.icgc.reading.ac.uk).
2.5.2. Luk 70 biomass production
The biomass of Luk70 was produced as stated in section 2.2. Once the fermentation was over, the concentrations of vegetative cells (UFC mL− 1), spores (UFC mL− 1), and contaminants, including fungi (UFC mL− 1) and bacteria (UFC mL− 1) were measured. The quality control of the fermentation broth involved serial dilution starting from 1 mL of the fermentation broth. A 0.85% w/v saline solution was used as diluent, and 100 µL of the selected dilutions were plated on the surface of nutrient agar (NA) for bacterial analysis. Sabouraud potato dextrose agar (SPDA) was used to quantify fungi concentration, while yeast malt extract agar (YM) was used for yeast. Both were plated in Petri dishes for analysis. NA and YM Petri dishes were incubated at 28 ± 2°C for 48 hours, while PDA Petri dishes were incubated at 25 ± 2°C for 120 hours.
2.5.3. In situ cadmium removal at lab-scale setup
A completely randomized factorial (3x3x4) design with three replicates per treatment was carried out in plastic containers, where each experimental unit consisted of 2.5 kg of fermented cocoa and 2.5 L of bacterial solution at any of the following concentrations (in UFC mL-1): B1 = 1x107, B2 = 5x107, and B3 = 1x108. Samples of cocoa beans and supernatant were collected from each container at 0, 12, 24, and 48 hours, and cadmium concentration, and pH of nibs were measured. The Cd concentration was determined using a flame atomic absorption spectrometer (Perkin Elmer Analyst 300). Cd removal was calculated following the UNE-EN 14084:2003 protocol 30. Removal coefficients were calculated using Eq. 4.
| \(\:removal\:coefficient=\frac{{Cd}_{tf}}{{Cd}_{t0}}\) | Equation 4 |
Where \(\:{Cd}_{t0}\) is the initial Cd concentration (mg Kg− 1) in cocoa beans, and \(\:{Cd}_{tf}\) is the Cd concentration (mg Kg− 1) in cocoa beans at the end of the experiment.
2.6. Cost analysis for pilot-scale fermentation
The production costs of Luk70 biomass at the pilot scale were estimated. A calculation base of 100 L for the bacterial fermentation production process was considered. The cost modeling methodology outlined by AGROSAVIA’s Bioproducts Department, as described in a previous study 31, was used to calculate the costs of raw materials, personnel, and other expenses encompassing both direct and indirect measures (Eq. 6).
Where DRM (direct raw material) corresponds to the consumption of each raw material required multiplied by its price; DLC (direct labor cost) is the cost of man-hours required for each level of staff training; ILC (indirect labor cost) is the cost of man-hours required for personnel performing supervision; IRM (indirect raw material) corresponds to 5% of the total raw material costs (general commodities); OIC (other indirect costs) corresponds to 10% of net costs (infrastructure, equipment, leases, and waste collection); DRM, DLC, and IRM are variable costs; and ILC and OIC are fixed costs.
2.7. Analytical methods
2.7.1. Quantification of vegetative cells, spores and Luk70 dry biomass
For vegetative cell quantification in medium broth during fermentation, serial dilutions were prepared from 1 mL of fermentation broth in 0.85% w/v saline solution as diluent. 100 µL of the dilutions were plated on Luria-Bertain (LB) agar surface in Petri dishes. The sample was homogeneously spread with a sterile Drigalsky spatula and then incubated at 30°C for 48 to 72 h. For the quantification of spores in liquid, a procedure similar to that described for vegetative cells was applied with the addition of a heat shock process, involving heating the sample to 80°C in a water bath for 10 minutes, then cooling it down with ice cold water. Subsequently, the colonies of Bacillus xiamenensis Luk70 were numbered, and the results were expressed as UFC mL− 1.
The fermentation broth aliquots were centrifuged at 4000 rpm for 15 minutes to quantify vegetative and spores in biomass samples. One gram of the precipitate was utilized to perform serial dilutions in 0.85% w/v saline solution, and the procedure mentioned above was applied to quantify vegetative cells and spores.
Dry biomass (g L− 1) was determined by dry weight, where wet samples of pellets from a known volume of fermentation broth were subjected to oven drying until a constant weight was achieved.
2.7.2. Sugar concentration quantification
Sugar analysis (sucrose, glucose and fructose) was performed on a UPLC Waters Acquity system equipped with an Aminex HPX-87P (Bio-Rad, Filadelfia, PA, USA) column (300 mm x 7.8 mm) and a Waters IR 2414 refractive index detector. All samples from culture mediums were centrifuged at 5000 rpm for 10 min and filtered through a 0.22 µm Durapore® PVDF membrane filter. Ultra-pure water was used as the mobile phase with a flow rate of 0.6 mL min− 1 and oven temperature was maintained at 85°C. The injection volume was 10 µL. Tests were conducted by duplicate.
2.7.3. Nib and testa pH determination
The nib pH was determined following a previously reported protocol with some adaptations 16. Briefly, 500 g of beans were manually separated from the cocoa nibs. Then, 20 g of nibs were mixed with 40 mL of deionized water in a blender. This preparation was filtered using a 60 µm mesh, and the filtrate was used for pH measuring using a potentiometric pH meter, previously calibrated with standard solutions at pH 4, 7, and 10. The same process was carried out for the pH analysis of the test.
2.7.4. Cocoa liquor production and sensory evaluation
Sensory evaluation was conducted following a previously reported protocol 32. The fermented cocoa beans were sun-dried, roasted, and processed into cocoa liquor by CASALUKER S.A. (Bogotá, Colombia) using their standardized protocols. Briefly, cocoa liquor was obtained by subsequent steps, including cleaning (i.e. separating nibs from testa), roasting, cooling, winnowing, grinding, tempering, molding, and storage for further analysis. A trained tasting panel performed a sensory evaluation of cocoa liquor. This is a standardized procedure performed several times a year at the company using liquor produced with material from these farms. Flavor descriptors included acid/sour, astringent, cocoa, bitter, fruity, floral, nutty, moldy, green, malt and other attributes. The analysis was performed in triplicate with an evaluation temperature between 45–50°C. The flavor intensity was quantified using numerical values ranging from 0 to 9.
2.7.5. Statistical analysis
Pairwise statistical comparison tests were carried out between the cadmium concentrations in nibs and testa across different treatments using a Games-Howell´s test, with the R package ´ggstatsplot´33 at a confidence level of 95%. Holm’s multiple comparison correction adjusted the p-values for comparisons involving more than two groups. Pearson correlation coefficients were calculated between the study variables using the R package ´ggstatsplot´33. All figures were generated using the R package ´ggplot2´ 34.
3.1. Effect of the type of biomass in the cadmium removal efficiency
The heavy metal removal process from solutions by Luk70 biomass has not been well studied, and it is unclear which structure is the most effective for metal ion removal. Nevertheless, Mohapatra et al. (2019) suggested that Pb(II) could be a passive mechanism such as adsorption, chelation, precipitation, and ion exchange, which directly depends on the type of biomass. Therefore, an experiment was set up to determine which type of biomass from Luk70 is the most efficient for cadmium removal in vitro. Four types of biomasses were tested: Biomass with a similar number of vegetative cells and spores (long fermentation), biomass with more vegetative cells than spores (short fermentation), biomass with only spores (spores) and dead biomass (dead). The experiment also had two controls, water only and the sterile medium without inoculation of bacteria. The efficiency of cadmium removal for the different structures is shown in Fig. 1a. The higher removal was achieved by the live biomass-long fermentation with 65% at 24h and 85% at 48h. The live biomass-short fermentation with more vegetative cells than spores reached a cadmium removal level of 20% at 24h and 48h. This result is close to the values achieved by the spores for cadmium removal of 22% at 24h and 48h. Finally, the control water reduced the cadmium concentration by 4.64% and the control sterile medium had a removal rate of 8.32%, whereas the dead biomass did not remove any cadmium within the 48 hours.
The spore concentration increased over the 48 hours for the long fermentation, contrary to the case of the short fermentation, where the spore concentration increased within 24 hours, and decreased in the last 24 hours (Fig. 1b). Meanwhile, cadmium did not affect the viability of vegetative cells of biomass from long or short fermentations (Fig. 1c). The long fermentation had almost a similar number of vegetative cells and spores (1:1), and the short fermentation had four times more vegetative cells than spores. A previous study by Mohapatra et. al. (2019) reported a higher removal efficiency of lead by live bacterial biomass of B. xiamenensis than dead biomass due to the intracellular metal accumulation mechanism 26. Therefore, the live biomass with the proportion of vegetative cells and spores close to 1:1 was chosen as the active ingredient for the bioproduct. This result could be explained as a summary of different ways of how B. xiamenensis survives in the presence of Cd using passive strategies such as biosorption, binding of Cd ions in the cell wall, in vegetative cells, spores or dead biomass. The biosorption mechanism for bioremediation of heavy metal components in bacterial cell walls are the cornerstone for the proper functioning of this mechanism. The presence of anionic groups such as alcohol, carboxylic, ester, sulfhydryl, amine, etc. 35 constitute the receptors where cationic heavy metals are bound and immobilized. Finally, how different species of microorganisms carry out the biosorption process depends on the uniqueness and concentration of the pollutants in the environment, the pretreatment of the cell biomass, and the experimental conditions 36. This could explain the low cadmium removal rates shown by dead biomass treatments, where the harsh conditions during biomass inactivation, obtained by sterilization in an autoclave at 121°C, 1.5 atm for 40 minutes, could affect the presence or disrupt the chemical structure of important cell wall components that are important for biosorption.
3.2. Luk70´s submerged fermentation (SmF)
3.2.1. Fermentation kinetics of Luk70
The trend of the dissolved oxygen kinetics of Luk70 in the two batches is shown in Fig. 2a. A typical behavior was observed within the first 20 hours of fermentation, with an initial phase of rapid oxygen consumption associated with the exponential growth of the microorganism without the presence of an adaptation phase. On one hand, the oxygen concentration decreased to a minimum value between 5–10% and increased in a reaeration process to a final value close to 100% saturation for batch 1. Batch 2, on the other hand, had an intermediate deaeration phase at around hour 42 of fermentation and a subsequent reaeration as in batch 1. These deaeration-aeration cycles are associated with vegetative cell death, simultaneous spore formation, and the final entry into a stationary growth phase due to the depletion of nutrients.
Meanwhile, the pH kinetics is presented in Fig. 2. The ranges obtained are very similar (6.20–9.55 for batch 1 and 6.40–9.53 for batch 2), which is an indirect indicator of the presence of similar fermentation processes corresponding to growth phases and metabolic behavior of the microorganism that is reproducible over time. Furthermore, Luk70 has a metabolism that basifies the medium and does not produce organic acids. A wide range of secondary metabolites can be produced, and genomic analysis will allow their elucidation in future works.
The pH was chosen as an indicator variable for the reproducibility of the Luk70 fermentation process since a tolerance interval of 95% was obtained, covering the entire experimental range with a normal distribution. By applying a tolerance interval analysis, long-term coefficients of variation were obtained for the whole process: LTCV (batch 1) = 9.29% and LTCV (batch 2) = 8.55%, respectively, with a variability of 2.36% in the response variable dry biomass, allowing us to confirm that the fermentation process of the Luk70 strain is reproducible over time under standardized operating conditions.
The concentration values for vegetative cells and spores of Luk 70 were in an order of magnitude of ten (10) times higher than those of the previous stage in Erlenmeyer (data not shown). In addition, there was a high reproducibility between batches, reflected in the dry biomass values with a coefficient of variation of 2.36% (Table 1). Finally, a high yield was obtained resulting from the rapid consumption of the entire substrate.
| Batch | Concentration in fermentation broth (UFC mL− 1) | Quotient (Spores: Vegetative cells) | Dry biomass-X (g L− 1) | Average yield* X DS− 1 | ||
|---|---|---|---|---|---|---|
| Spores | Vegetative cells | |||||
| 1 | 1.97E09 | 5.03E09 | 1.0: 2.55 | 1.52 | 1.38 | |
| 2 | 1.57E09 | 1.33E09 | 1.18: 1.0 | 1.47 | 1.34 | |
| *Based on sucrose consumption (DS) | ||||||
3.3 Effect of the biomass concentration, cocoa fermentation time, and contact time on in situ cadmium removal at lab-scale
3.3.1. Cocoa bean fermentation
Within the nine days of fermentation, the pH of the nib, the pH of the bean shells, and the temperature of the fermentation mass were monitored daily, whereas the percentage of fermented beans was determined from day 4 onwards (Fig. 3). The nib pH decreased over time, reaching a pH of around 4.5 on day 5, whereas the testa pH increased over time, reaching a pH of 4.5 on day 6, and then remained around this value for the rest of the fermentation for both testa and nib. The temperature of the fermentation mass increased over time from 26°C to 48°C until the 5th day, when the mass turns to allow oxygenation, changing from anaerobic fermentation to aerobic fermentation, and the temperature profile changes to an isothermal phase at around 48°C. The behavior of the pH and temperature variables measured during the spontaneous fermentation was similar to that previously reported in other studies, which are crucial for forming the precursors of the characteristic flavors of the chocolate 3,16,29,37.
3.3.2. In situ cadmium removal at lab-scale
Luk70 was produced according to the procedure described above. The bioreactor was harvested at the end of the fermentation, and the biomass was recovered by centrifugation. Quality control of the recovered biomass showed no significant presence of contaminants, neither bacteria, yeast, nor fungi. The values were below the limits of detection for bacteria (1x105 UFC ml− 1) and for yeast and fungi (< 100 UFC ml− 1) according to the Colombian regulations for bioproducts, which establishes a maximum of 5% of contaminants in relation to the active ingredient 38. Meanwhile, the concentration of spores was 1.11 X1011 UFC g− 1 and vegetative cells 9.83 X1010 UFC ml− 1, staying close to the ratio (1:1).
Once the cocoa beans had reached the desired point of fermentation, the quantity of fermented beans required for each treatment was taken and thoroughly mixed with the solution of Luk70 biomass at the set vegetative cell concentration (B1 = 1x107 UFC ml− 1, B2 = 5x107 UFC ml− 1, and B3 = 1x108 UFC ml− 1), which is similar to the spore concentration since a 1:1 ratio was kept. The nib Cd concentrations from 60%, 80% and 90% fermented cocoa beans were 2.60 ± 0.06 mg Kg− 1, 2.61 ± 0.04 mg Kg− 1, and 2.45 ± 0.06 mg Kg− 1, respectively, which are close to previously reported values. For example, Bravo et al. (2022) reported a nib cadmium concentration from cocoa-fermented beans for 5 days from the Arauca region of 3.28 mg Kg− 1, while, from Piura, Peru, a nib Cd concentration of 2.94 ± 0.11 mg Cd Kg− 1 fermentation 39 was obtained. Also, Vanderscueren et al. (2020) reported a cadmium concentration of around 2.5 mg Kg− 1 on nib from cocoa-fermented beans within 4 days from Guayas field 2-Ecuador 15.
The effect of the percentage of fermented beans, the treatment time, the Nib pH, and the concentration of bacteria applied on the reduction of cadmium concentration treatment was studied. Figure 4a shows that, with a longer fermentation time, lower nib cadmium concentrations were obtained with the application of Luk70 biomass. The lowest Cd concentrations were found in the treatments with 90% fermentation and a biomass concentration of 5x107 UFC ml− 1 at 12 h, which reached a value of 1.70 ± 0.05 mg Kg− 1, and a similar value of 1.43 ± 0.20 mg Kg− 1 was obtained by the treatment with 90% of fermentation and a biomass concentration of 1x108 UFC ml− 1, but in a longer period (48 h). Meanwhile, the lowest nib Cd concentration value for the treatment control was found at 90% of fermentation after 24 hours with 2.18 ± 0.06 mg Kg− 1. The Pearson correlation coefficients were calculated between the studied variables (Fig. 4b). It was found in the last fermentation that the longer the treatment time had a significant adverse effect on the Cd concentration (p < 0.05). In contrast, the higher pH significantly affected the Cd concentration (p < 0.05). Finally, a non-significant effect of Luk70 biomass concentration on Cd concentration was observed.
The Cd concentration in the testa increased over time for most of the treatment, showing the migration of Cd from the cocoa nibs to the testa (Fig. 4c). Treatments with longer fermentation times had higher cadmium concentrations. The control treatments showed no significant variation in testa cadmium concentration with some treatments, showing an increase over time, whereas others remained stable over time. Figure 4d showed the Pearson correlation coefficient, where only a significant positive effect between the percentage of fermentation and the cadmium concentration in the testa (p < 0.05) was found. Thus, the longer the fermentation, the more degraded the testa becomes, which probably allows an easy migration of cadmium from the nibs to the testa while also favoring water diffusion into the nib 13.
A pairwise comparison was made between the nib Cd concentration at different contact times for all treatments and time zero (Fig. 5a). It was found that the nib Cd concentration was statistically significantly higher at time zero compared to the samples at 12, 24, and 48 hours. Also, the cadmium removal occurred mainly within the first 12 hours with a decrease in cadmium concentration from 2.55 mg Kg− 1 to 2.25 mg Kg− 1, and thereafter, the cadmium concentration did not change significantly. The comparison of the Cd concentration in the testa showed that the concentration at time zero was statistically significantly lower than the values at 12 and 24 hours (Fig. 5b). As with the behavior of the Cd concentration in the nib, the main change in the Cd concentration in the testa occurs within the first 12 hours, but with some additional increase over 24 hours, and then a decrease in the Cd concentration in the testa.
During the contact time, the pH decreased below 4.5 in all treatments, and a greater decrease in pH was observed in the control treatment (Table 2). The bacterial treatment ended at 4.3 after 48h, whereas the pH of the water treatment ended on average at 4.1. This behavior could be a consequence of the metabolism of the bacteria. During the 48 hours of contact, Luk70 continues growing and producing metabolites that could help to maintain the pH in the treatments. Nevertheless, a positive correlation was found between the cadmium concentration in the nibs and the pH of the nibs (p < 0.05) (Fig. 4b). Thus, acidic pH favored Cd removal from the cocoa nibs. Previously, it had been shown that the nib pH < 5 induces migration of Cd from the cocoa nibs to the outside 14,15. Here, every treatment had a pH below 5 and showed migration of Cd from the nib to the testa, as previously discussed. Recent studies have shown that Cd is stored in the nib prior to fermentation by chelation with phytate 40, and during the fermentation conditions of pH 5.0 and high temperatures (45 and 55°C), phytase is activated in the nib and helps to hydrolyze the phytate bonds, thereby mobilizing chelated metals such as cadmium 13,17. However, our experiments were carried out without temperature control and under ambient conditions, with temperature variations between 24°C-40°C, with the lowest temperature around the 12 h (at night) and the highest temperature around the 24 h (at noon). Thus, the mobilization of cadmium in our experiments could be caused by a combination of factors such as the presence of contact solution, which helps to migrate cadmium from the nib to the solution 13. Also, the biomass of Luk70 could help to favor the diffusion of more cadmium by trapping it inside or outside the cells and increasing the concentration gradient. Furthermore, during the contact time, the bacteria could produce acids such as acetic acid or lactate, which are known to be able to penetrate the cocoa nibs and contribute to the release of cadmium 14. In the production process of the bacteria in the bioreactor, Luk70 had produced this type of acids (data not shown). Finally, the phytases could still be active even though they were not under optimal conditions.
| % fermented beans | Vegetative cells concentration | nib pH | nib Cd concentration ratio | |||||
|---|---|---|---|---|---|---|---|---|
| 0 h | 12h | 24h | 48h | 12h | 24h | 48h | ||
| 60 | Control | 5.21 | 3.7 | 3.8 | 3.9 | 1.10 | 1.08 | 1.04 |
| 1x107 UFC ml− 1 | 4.3 | 4.2 | 4.1 | 1.05 | 1.22 | 1.23 | ||
| 5x107 UFC ml− 1 | 4.2 | 4.2 | 4.1 | 1.05 | 1.09 | 1.12 | ||
| 1x108 UFC ml-1 | 4.2 | 4.2 | 4.1 | 1.15 | 1.08 | 1.10 | ||
| 80 | Control | 4.53 | 4.7 | 4.5 | 4.7 | 1.07 | 1.12 | 1.13 |
| 1x107 UFC ml− 1 | 4.3 | 4.2 | 4.2 | 1.20 | 1.06 | 1.13 | ||
| 5x107 UFC ml− 1 | 4.3 | 4.2 | 4.1 | 1.26 | 1.27 | 1.18 | ||
| 1x108 UFC ml-1 | 4.3 | 4.1 | 4.2 | 1.17 | 1.05 | 1.07 | ||
| 90 | Control | 4.50 | 3.9 | 3.9 | 3.8 | 1.04 | 1.12 | 1.11 |
| 1x107 UFC ml− 1 | 4.3 | 4.1 | 4.2 | 1.11 | 1.15 | 0.96 | ||
| 5x107 UFC ml− 1 | 4.2 | 4.1 | 4.4 | 1.45 | 1.12 | 1.16 | ||
| 1x108 UFC ml-1 | 4.2 | 4.1 | 4.3 | 1.19 | 1.23 | 1.77 | ||
The application of bacteria at a concentration of 1x108 UFC ml− 1 over 90% of fermented beans achieved the highest reduction in nib cadmium concentration at 48 hours by a factor of 1.77 (Table 2). However, the treatment with Luk70 at a concentration of 5x107 UFC ml− 1 and 90% of fermentation showed a close decrease in nib Cd concentration within the first 12 hours by a factor of 1.45. The control treatment also reduced nib Cd concentrations by a factor of 1.12. The lower value of nib Cd concentration achieved with the proposed treatment was 1.43 ± 0.20 mg Kg− 1, which could be used to produce chocolate with 55% cocoa solids under current EU regulations, as it has been reported that almost all the Cd on cocoa nib (depending on cocoa solids percentage) ends up in the chocolate 3.
So far, post-harvest Cd removal treatments have been studied focusing on the effect of the conventional fermentation process on the reduction of nib Cd concentration. Bravo et al. (2022) reported a Cd concentration ratio of 1.25 after fermentation, while the one from Piura, Peru, reported a Cd concentration ratio of 1.24 39, and Vanderscueren et al. (2020) reported a reduction in Cd concentration by a factor of 1.3 15. Nevertheless, the post-fermentation cadmium concentration ratios are contrasting and variable across Colombia, as are the cadmium concentrations. For example, the Casa Luker database on cocoa beans specifically fermented to 90% over the years (data not shown) shows samples from Santander with a reduction in post-fermentation nib Cd concentration by a factor of 1.20, and samples from other farms also from Santander with a Cd ratio concentration of 1.14. Meanwhile, samples from Nariño showed a Cd concentration ratio of 1.08 after the fermentation, but samples from another farm also from Nariño showed a decrease in nib Cd concentration by a ratio of 1.70. These results show the variable effects of the conventional fermentation process on the removal of cadmium, which is carried out under non-standard or controlled conditions of temperature, time, and pH. Also, the variability in the cadmium concentrations found in large fermented masses has been reported previously 14, as these masses came from different trees and even different fields. A nationwide survey of cadmium concentrations in Ecuador reported an average 39% variation in Cd concentration between pods harvested from different trees in the same field 13. Meanwhile, in Colombia, Bravo et al. (2021.) reported the highest variability (Cv 1.73) between samples from the Santander district 6.
Another proposed solution to improve the Cd reduction during the fermentation process is the addition of acids and/or more extreme temperature conditions. Camargo et al. (2024) used controlled temperature fermentation treatments to reduce nib cadmium concentration. The treatment with the higher mean temperature (44.25°C), lowered the nib pH to 4.66, and demonstrated a direct relationship between these variables and Cd migration from nibs to testa without affecting the organoleptic properties of the liquor 18. Vanderschueren et al. (2023) suggested adding 20 g L− 1 of acetic acid at a temperature of 45°C, achieving a reduction in nib Cd concentration by a factor of 1.3, whereas increasing the temperature up to 65°C and adding 40 g L− 1 of acetic acid increased the reduction in nib Cd concentration by a factor of 1.6. However, the effect of these treatments on chocolate quality was not researched 13,17 and the treatments were applied to small fermentation masses. They highlighted the importance of the temperature in cadmium removal and the need to fine-tuning fermentation conditions 13.
Colombian cocoa is recognized around the world for its flavor and aroma characteristics and is considered as “fine flavor” by the International Cocoa Organization. Therefore, the treatment chosen to reduce the cadmium concentration should not affect the quality of the chocolate. The treatment with 90% fermentation and a Luk70 concentration of 5x107 UFC ml− 1 was chosen as the best. As Cd migration occurs mainly within the first 12 hours, the cost of maintaining a longer treatment and a higher bacterial concentration was considered. Samples from this treatment were analyzed to determine its sensory profile. The chocolate was not affected by the treatment and did not experience any strange flavors. Its sensory profile showed higher bitter, astringent and green notes and lower cocoa notes than the cocoa without the application of Luk70 treatment for Cd removal (Supplementary Fig. 1).
3.4. Quantification of production cost at pilot-scale
The total cost of production was extrapolated to a pilot-scale production process (100L) of Luk70 biomass as shown in Table 3. The cost was calculated considering only the fermentation, separation, and quality control steps, resulting in a cost of USD 0.56 per kg of wet cocoa beans at a biomass concentration of 5x107 UFC mL− 1. This cost represents 14.4% of the average current price of wet cocoa beans in Colombia, which is a high investment in the process to reduce cadmium concentration, in addition to the fact that a better price for cocoa with lower cadmium levels is not currently offered in Colombia. There are also additional costs such as labor to apply the treatment, as well as water to dilute the bacteria during the treatment and residue management that have not been considered at this stage.
| Bacteria batch production | 100L |
|---|---|
| Bacteria dry biomass quantity (g) | 294.46 |
| Bacteria concentration (UFC g− 1) | 1 x1011 |
| Costs | |
| Direct raw material (DRM) | $ 112.29 |
| Direct labor cost (DLC) | $ 79.80 |
| Indirect labor cost (ILC) | $ 2.46 |
| Indirect raw material (IRM) | $ 5.61 |
| SUBTOTAL | $ 200.15 |
| Other indirect costs (OIC) | $ 20.02 |
| TOTAL | $ 220.17 |
| Price (USD) for bacteria at 5x107 UFC ml− 1 per kg of cocoa wet beans | $ 0.56 |
This is the first time that a bioremediation process for cadmium in post-harvest cocoa has been presented with an in-situ application at a lab scale. The application of Luk70 biomass to fermented cocoa beans can capture cadmium and reduce nib Cd concentrations by a factor of 1.45 or 1.77, which to the best of our knowledge are the highest reductions in nib Cd concentration reductions reported to date without affecting the chocolate quality. The reduction in nib Cd was influenced by the time of contact, the percentage of fermentation of the beans and the nib pH. The longer the contact time and the more fermented the cocoa beans, the better for the nib Cd reduction. However, the highest reduction was achieved within the first 12 hours of the treatment. Acidic nib pH also favors Cd reduction in cocoa nibs, as reported previously. However, the cost of producing the bacteria for Cd reduction is high. The proposed treatment could be improved by using special equipment to control the temperature, but this will increase the cost of the whole process. The proposed treatment could be part of a series of actions to reduce the cadmium concentration in cocoa from plantation to chocolate production, but the economic cost and the impact on the cocoa chain value and the producers have yet to be considered.
ADDITIONAL INFORMATION
The authors declared no competing interests.
Author Contribution
Conceptualization: DAM, AD, ABP, CMRL, DC, FJG, MJC, EJB. Data curation: DAM, AD, ABP, CMRL, DC, MJC, EJB. Formal analysis: DAM, AD, ABP, CMRL, DC, MJC, EJB. Funding acquisition: FJG, MJC, EJB. Investigation: DAM, AD, ABP, CMRL, DC, FJG, MJC, EJB. Methodology: DAM, AD, ABP, CMRL, DC, FJG, MJC, EJB. Project administration: MJC, EJB. Supervision & Validation: MJC, EJB. Visualization: MJC, EJB. Writing: DAM, AD, ABP, CMRL, DC, FJG, MJC, EJB. Writing ± review & editing: DAM, AD, ABP, CMRL, DC, FJG, MJC, EJB. All authors have approved the final version.
Acknowledgement
The authors acknowledge CasaLuker S.A for technical, scientific, and financial support. The Ministry of Environment and Sustainable Development (Colombia) is also acknowledged for granting sampling permissions according to Resolution no. 0523 of 2020. Thanks to AGROSAVIA and the Ministry of Agriculture and Rural Development (MADR, by its Spanish initials) for their funding. The study was funded through the co-development projects between AGROSAVIA-CASA LUKER (Agreement TV19-21 and TV20-06).
Data Availability
The datasets generated during and analysed during the current study are available from the corresponding author on reasonable request.
- Vanderschueren, R. et al. The contribution of cacao consumption to the bioaccessible dietary cadmium exposure in the Belgian population. Food Chem. Toxicol. 172, 113599 (2023).
- Hussain, B. et al. Cadmium Prevalence in Cacao (Theobroma cacao L.) and Potential Remediation Strategies. J. Soil Sci. Plant Nutr. 2023 1–17 (2023) doi:10.1007/S42729-023-01349-6.
- Bravo, D. et al. ‘From soil to chocolate bar’: identifying critical steps in the journey of cadmium in a Colombian cacao plantation. Food Addit. Contam. Part A 39, 949–963 (2022).
- Vanderschueren, R. & Pulleman, M. Cadmium in cacao: why it occurs, how it is regulated, and why it is a concern for producers. Clima-LoCa Policy Br. (2021).
- Blommaert, H. et al. Ca-oxalate crystals are involved in cadmium storage in a high Cd accumulating cultivar of cacao. Environ. Exp. Bot. 105713 (2024).
- Bravo, D. et al. The First National Survey of Cadmium in Cacao Farm Soil in Colombia. Agronomy (2021) doi:10.3390/agronomy11040761.
- Bravo, D. et al. Assessing the cadmium content of cacao crops in Arauca, Colombia. Environ. Monit. Assess. 196, 387 (2024).
- Liu, L., Li, W., Song, W. & Guo, M. Remediation techniques for heavy metal-contaminated soils: Principles and applicability. Sci. Total Environ. 633, 206–219 (2018).
- Vazquez, A., Zawoznik, M., Benavides, M. P. & Groppa, M. D. Azospirillum brasilense Az39 restricts cadmium entrance into wheat plants and mitigates cadmium stress. Plant Sci. 312, 111056 (2021).
- Bravo, D. & Braissant, O. Cadmium tolerant bacteria: Current trends and applications in agriculture. Lett Appl Microbiol Accepted, (2021).
- Deng, M., Li, K., Yan, Y.-J., Huang, F. & Peng, D. Enhanced cadmium removal by growing Bacillus cereus RC-1 immobilized on different magnetic biochars through simultaneous adsorption and bioaccumulation. Environ. Sci. Pollut. Res. 1–13 (2022).
- Yu, X. et al. Cadmium-absorptive Bacillus vietnamensis 151–6 reduces the grain cadmium accumulation in rice (Oryza sativa L.): Potential for cadmium bioremediation. Ecotoxicol. Environ. Saf. 254, 114760 (2023).
- Vanderschueren, R., Helsen, F., Doevenspeck, J., Delcour, J. A. & Smolders, E. Incubation tests mimicking fermentation reveal that phytate breakdown is key to lower the cadmium concentrations in cacao nibs. Food Chem. 398, 133899 (2023).
- Vanderschueren, R. et al. Cadmium migration from nib to testa during cacao fermentation is driven by nib acidification. LWT 157, 113077 (2022).
- Vandersypen, K. et al. The impact of fermentation on the distribution of cadmium in cacao beans. (2020).
- Vanderschueren, R. et al. The impact of fermentation on the distribution of cadmium in cacao beans. Food Res. Int. (2020) doi:10.1016/j.foodres.2019.108743.
- Blommaert, H., Sarret, G., Chavez, E., Smolders, E. & Vanderschueren, R. Cadmium speciation in cacao beans changes during a fermentation-like incubation. Food Chem. 431, 137068 (2024).
- Camargo, I. D., Rodriguez-Silva, L. G., Carreño-Olejua, R., Montenegro, A. C. & Quintana-Fuentes, L. F. High temperature and nib acidification during cacao-controlled fermentation improve cadmium transfer from nibs to testa and the liquor’s flavor. Sci. Rep. 14, 12254 (2024).
- Rodriguez López, C. et al. Colección de Microorganismos asociados al cacao de Casaluker S.A. CasaLuker S.A.. Dataset/Occurrence. https://ipt.biodiversidad.co/sib/resource?r=luker_coleccion-microorganismos (2022).
- Xia, Y. et al. Multi-stress tolerant PGPR Bacillus xiamenensis PM14 activating sugarcane (Saccharum officinarum L.) red rot disease resistance. Plant Physiol. Biochem. 151, 640–649 (2020).
- Eke, P. et al. Endophytic bacteria of desert cactus (Euphorbia trigonas Mill) confer drought tolerance and induce growth promotion in tomato (Solanum lycopersicum L.). Microbiol. Res. 228, 126302 (2019).
- Nepomuceno, R. A., Brown, C. B., Gargarino, A. M. P., Pedro, M. S. & Brown, M. B. Growth enhancement of rice (Oryza sativa L.) by zinc-solubilizing bacteria isolated from Vesicular-Arbuscular Mycorrhizal Root Inoculant (VAMRI). Philipp. J. Crop Sci 45, 34–40 (2020).
- Sharma, A. et al. Insights into the bacterial and nitric oxide-induced salt tolerance in sugarcane and their growth-promoting abilities. Microorganisms 9, 2203 (2021).
- Abu Hujier, N. S., Abdelsalam, S. M., Sharif, F. A. & Fahd, M. I. Improvement of Phaseolus vulgaris growth by inoculation with multifunctional native rhizobacteria isolated from rhizospheric soils in Gaza strip-Palestine. J. Sci. Res. Sci. 37, 1–29 (2020).
- Mukherjee, B., Roy, S., Dhara, A. & Dutta, S. Screening and optimization of indole-3-acetic acid production by bacterial strain isolated from rice rhizosphere and its effects on plant growth. Environ. Exp. Biol. 19, 141–149 (2021).
- Mohapatra, R. K. et al. Active and passive biosorption of Pb(II)using live and dead biomass of marine bacterium Bacillus xiamenensis PbRPSD202: Kinetics and isotherm studies. J. Environ. Manage. (2019) doi:10.1016/j.jenvman.2019.06.073.
- Díaz-García, A., García-Riaño, J. & Zapata-Narvaez, J. Improvement of sporulation conditions of a new strain of Bacillus amyloliquefaciens in liquid fermentation. Adv. Biosci. Biotechnol. 6, (2015).
- Wunderlin, T., Junier, T., Paul, C., Jeanneret, N. & Junier, P. Physical isolation of endospores from environmental samples by targeted lysis of vegetative cells. J. Vis. Exp. (2016) doi:10.3791/53411.
- Fernández-Niño, M. et al. Dissecting industrial fermentations of fine flavour cocoa through metagenomic analysis. Sci. Reports | 11, 8638 (2021).
- Certificación, A. E. de N. y. UNE-EN 14084: productos alimenticios: determinación de elementos de traza : determinación de plomo, cadmio, cinc, cobre y hierro mediante espectrometría de absorción atómica (EAA) tras digestión en microondas. (AENOR, 2003).
- Sánchez, L. et al. Selection of inductors for thermotolerant conidia production of the entomopathogenic fungus Metarhizium rileyi Nm017. Biocontrol Sci. Technol. 32, 891–898 (2022).
- Herrera-Rocha, F. et al. Dissecting fine-flavor cocoa bean fermentation through metabolomics analysis to break down the current metabolic paradigm. Sci. Rep. 11, 21904 (2021).
- Patil, I. Visualization with statistical details: The “ggstasplot” approach. J. Open Source Softw. 6, 3167 (2021).
- Villanueva, R. A. M. & Chen, Z. J. ggplot2: elegant graphics for data analysis. (2019).
- Ayangbenro, A. S. & Babalola, O. O. A new strategy for heavy metal polluted environments: A review of microbial biosorbents. Int. J. Environ. Res. Public Health 14, (2017).
- Park, D., Yun, Y. S. & Park, J. M. The past, present, and future trends of biosorption. Biotechnol. Bioprocess Eng. 15, 86–102 (2010).
- Santander Muñoz, M., Rodríguez Cortina, J., Vaillant, F. E. & Escobar Parra, S. An overview of the physical and biochemical transformation of cocoa seeds to beans and to chocolate: Flavor formation. Critical Reviews in Food Science and Nutrition (2020) doi:10.1080/10408398.2019.1581726.
- ICA. Resolución Bioinsumos 068370. Instituto Colombiano Agropecuario - ICA vol. 068370 1–37 (2020).
- Ramtahal, G. et al. Prediction of Soil Cadmium Bioavailability to Cacao (Theobroma cacao L.) using Single-Step Extraction Procedures. Commun. Soil Sci. Plant Anal. (2015) doi:10.1080/00103624.2015.1089262.
- Blommaert, H. et al. From soil to cacao bean: unravelling the pathways of cadmium translocation in a high Cd accumulating cultivar of Theobroma cacao L. Front. Plant Sci. 13, 1055912 (2022).
No competing interests reported.