Introduction

In recent years, numerous food-related trends have emerged. These trends reflect the rapidly evolving dynamics of food consumption, which in turn present significant challenges to the sustainability and security of the food supply chain. Factors such as climate change, the exponential growth of the world population, ongoing international conflicts and the outbreak of pandemics further exacerbate these challenges. As a result, all participants within the food supply chain are actively seeking innovative solutions to make the food system more sustainable. This effort is reinforced by the increasing awareness among consumers about their dietary habits, health and nutrition1,2.

A solution that is gaining global popularity is the trend of plant-based diets, including vegetarian and vegan food. These diets, which emphasize legumes, whole grains, vegetables, fruits, nuts and seeds consumption, and discourage most or all animal products, have multiple benefits3. They are not only more environmentally sustainable than meat-based diets, due to generating lower levels of greenhouse gases, but they also align with religious practices and address concerns regarding animal welfare4. Furthermore, plant-based diets offer a wide range of health benefits, including the prevention and management of chronic diseases, such as cardiovascular disease, obesity and diabetes5,6. However, vegetarians are advised to consume higher protein levels than what is typically recommended for omnivores. This is due to a lower protein quality and decreased digestibility of plant proteins7.

In plant-based diets, legumes play a key role due to their high protein content. Dietary fiber is recognized as another crucial component of a healthy diet. Legumes are considered high-fiber, as they contain a minimum of 6 g of fiber per 100 g. Studies suggest that supplementing plant-based diets with legume-derived products, such as plant-based protein powder, leads to similar results than a protein-matched mixed diet8.

Similar to other plant commodities, several fungi can infect the legume plants during growth, harvest and storage. Some of these fungi can produce mycotoxins: secondary metabolites produced by a wide diversity of toxigenic fungi9. The primary mycotoxin producers are species from Aspergillus, Fusarium, Penicillium and Alternaria. These low molecular weight compounds, are naturally ubiquitous and practically impossible to avoid10. Although pathogenic fungi can have toxic effects on plants and induce plant disease, the impact of mycotoxins on human and animal health is of greater concern worldwide. This is because mycotoxins accumulate in the edible parts of plants including legume-products11,12. They possess a wide range of toxic biological activities, such as hepatotoxicity, immunotoxicity, and carcinogenicity13. Moreover, mycotoxin contamination is a significant risk for producers, companies and export markets since it can lead to economic losses and instability in trade14. Although mycotoxins are not essential for fungi’s normal growth and development, they are produced in response to environmental changes15. For example, as temperature rises, there may be a general increase in fungi that grow in warmer conditions, including Aspergillus species that produce mycotoxins like the carcinogenic aflatoxins (AFs)16. Due to the combination of climate change and the increased sensitivity of analytical methods, the prevalence of mycotoxins has significantly increased worldwide17. The prevalence of mycotoxin contamination is higher in developing countries due to factors such as unfavorable climate, inadequate production technologies and insufficient crop storage facilities18.

Mycotoxins are stable compounds that can resist food processing. As a result, processed food remains vulnerable to contamination, even after undergoing various processing methods19. Researchers have been studying the behavior of mycotoxins during food processing in their quest to reduce human exposure to these compounds20. A range of processes may have an influence on mycotoxins, including cleaning, milling, brewing, cooking, baking, frying, roasting, flaking, alkaline cooking, nixtamalization and extraction. Although these processes generally result in a significant decrease in mycotoxin concentrations, complete elimination is not achieved21.

In addition to the reduction in mycotoxin levels, food processing can also cause mycotoxins to be released from or bound to matrix components. Furthermore, the chemical structure of mycotoxins can be influenced, potentially resulting in degradation or modification22. These modifications can lead to the formation of degradation products that may have different toxicokinetic and toxicodynamic properties than the parent mycotoxin23.

Therefore, having detailed knowledge about the fate of mycotoxins during food processing is essential. Thus, the aim of this work was the analysis of the fluctuation in mycotoxin concentration in different stages of the production process of legume-derived products from the raw materials to final products. This analysis aims to provide insights into the potential contribution of these raw materials to mycotoxin contamination in the final commercial products.

Results

Fate of multiple mycotoxins in the production of legume-derived products

The mycotoxins that were not detected in the 41 analyzed samples are either not mentioned in the tables or indicated by “not detected” (n.d.). Some mycotoxins were detected in concentrations below the method´s LOD or LOQ, but fulfilling the identification criteria of SANTE/12089/2016. They are indicated in the tables with an * and referred in the text as detected.

Figure 1 shows the flow diagram of the production of legume protein and other derived product and other intermediate products of the food processing company and indicates in which steps of the process mycotoxins were found.

Fig. 1: Flowchart representing the steps in the process to obtain product A, product B, product C, and product D from legume.
figure 1

Blue boxes: the raw materials, intermediate products, final products and side streams products that were analyzed. Gray boxes: the food-processing steps. Red squares: steps in which mycotoxins were found. AME alternariol monomethyl ether, ENN B enniatin B, DON deoxynivalenol, FB1 fumonisin B1, T2 T2-toxin.

Full size image

Raw materials and intermediate products

All the analyzed mycotoxins were below the method´s LOD in the two analyzed raw materials. Thus, a mass balance could not be established and concentrations in each step are informed.

The results of mycotoxin contamination of the intermediate steps of the processing of legumes are shown in Table 1. One sample of the hull contained alternariol monomethyl ether (AME). In one sample of the hydrated milled legume deoxynivalenol (DON), T2 toxin (T2) and enniatin B (ENN B) were found. The liquid fraction after the decantation sample (intermediate product) contained T2 and ENN B. Nivalenol (NIV), fumonisin B1 (FB1) and ENN B were found in one concentrate sample (intermediate product). At last, product C and D (intermediate product) was contaminated with T2 and ENN B.

Table 1 Presence and concentration of multiple mycotoxins in the different intermediate steps of the production of legume-derived products
Full size table

Legume-derived final products

Table 2 shows the contamination of AME in the final legume-derived products. One sample of product B tested positive for AME. In the other final legume-derived products, no mycotoxins were detected.

Table 2 Concentration of alternariol monomethyl ether in the different final products
Full size table

Side streams

Table 3 illustrates the results of mycotoxin contamination in the side streams. These are the steps that do not continue in the process line. Both dust samples were contaminated with DON and AME with concentrations ranging from 39 µg/kg to 46 µg/kg. Sterigmatocystin (STC) was also found in one dust sample. In one sample of the washing water, DON and ENN B were found. The same results were observed for one soluble C sample. One soluble D sample was also contaminated with DON and ENN B.

Table 3 Presence and concentration of multiple mycotoxins in the different side streams of the production of legume-derived products
Full size table

Discussion

Since literature regarding the fate of mycotoxins in the production and isolation of legume products is extremely limited, in this section, each mycotoxin and its fate is discussed individually.

ENN B was detected in certain intermediate products and side streams obtained during the processing of legumes, particularly following the process of milling and the alkaline solubilization. Moreover, since there is a loop in the wash water, ENN B could also be present due to a contamination from another batch. These processing steps can potentially cause the release of ENN B from its matrix. This results in redistribution and concentration within specific mill fractions, leading to its presence in the resulting intermediate products19. ENN B was found in legume based products in different countries24,25, but in our study ENN B was not detected in the final legume protein nor the other derived products. This absence could be due to either the sampled fraction not containing any ENN B or the influence of specific processing steps. Thermal treatment is a crucial factor in how processing can impact the mycotoxin levels in the final food product. Most mycotoxins are heat-resistant, and conventional food preparation methods involving temperatures up to 100 °C generally have minimal effects on them. However, higher temperatures used in processes like frying, roasting, toasting and extrusion have the potential to reduce mycotoxins levels in the remaining product22. Washing and alkaline solubilization were also frequently applied to obtain the final products and with soluble products as side streams. Water-soluble mycotoxins can be partially washed off26. Although ENN B is almost insoluble in water, it dissolves well in organic solvents. An effective method for solubilization involves the use of an alkaline solution. The solubilization step seems to be crucial for reducing the concentration of ENN B in the final products. It is known that ENN B has the ability to form complexes with alkaline ions, such as potassium and sodium, through the carbonyl oxygen atoms within its structure27. Therefore, when an alkaline solubilization process is employed, ENN B can bind to the alkali present in the solution, which may facilitate the selective extraction of ENN B to the soluble side streams.

DON was detected in the dust samples, which consists of small broken particles of the legumes and is generated by particle abrasion or friction any time the legume is moved or transferred. However, the whole legume samples taken prior to processing did not test positive for this mycotoxin. This absence may be attributed to a concentration effect in the dust particles. This was also observed in previous work, were there was a 13 fold accumulation of DON in dust compared to the wheat28. The sieving that excludes the dust seems to be an essential step to reduce mycotoxins in the final products. The dust is not utilized further in the processing, ensuring that the consumers are not directly exposed to it. In particular, this process uses a filter before releasing the dust to the environment to reduce the exposure to harmful compounds into the environment. DON, produced by Fusarium spp., was also detected at a much lower concentration in the hydrated milled legume sample. This presence can potentially be attributed to the release of DON from the matrix due to the hydration and milling steps, but further investigation is needed.

FB1 was detected in an intermediate product of the production of legume-derived products; specifically after undergoing thermal treatment and concentration processes. The aim of the concentration is to increase the total solid content of the food. However, during this process, certain contaminants such as FB1, have the potential to become more concentrated, leading to detectable levels in the resulting intermediate product21. Fumonisins (FBs) are known to exhibit relative heat stability and reduction occurs during processes involving temperatures exceeding 150 °C21. Consequently, the temperature employed was below this threshold, leading to the explanation of the presence of FB1 in the intermediate product. Nevertheless, the final products of the processing of legumes did not contain detectable levels of FBs. This observation suggests that further processing steps, such as thermal treatment and drying, effectively reduce the levels of some mycotoxins to undetectable amounts.

Similarly to DON, STC was found in the dust samples, yielding the same observation about DON. However, only one dust sample was contaminated and STC was not detected in any other steps, requiring the analysis of more batches to establish the risk of exposure to this mycotoxin.

AME, T2 and NIV were detected in the products obtained during various processing steps. AME, produced by Alternaria spp., was found in high concentration in the dusts and in one of the final products B. These fungi were reported in several legumes worldwide29,30,31 and AME was also reported in legumes25. AME was the mycotoxin that was more frequent along the process. This mycotoxin is known to resist conventional processing steps32 and was also reported in human biomonitoring studies, indicating exposure to this mycotoxin33. Even though AME is not regulated in legumes, and there are only recommended levels for this mycotoxin in some food products34, this mycotoxin should not be overlooked regulatory bodies since it can occur in a great variety of food products.

On the contrary, T2 and NIV were not detected in any of the final products. This suggests that the applied processing steps, such as thermal treatment and washing reduced these mycotoxins levels to undetectable. The overall mycotoxin contamination in these products seems to be lower than the reports in cereals35. Comparing the exposure to mycotoxins through legumes to that through cereals commonly consumed in vegan or vegetarian diets may suggest that the former is likely lower. Nonetheless, more studies and a comprehensive risk assessment are needed to evaluate the contribution of legume products to the mycotoxin exposure in vegan or vegetarian diets.

Notably, ochratoxin A (OTA), the only mycotoxin that is regulated in legumes36 was not detected in any step of the two batches that were analyzed. Nonetheless, the analysis of more batches is recommended to understand the fate of this mycotoxin in the production of legume based products. It is worth noting that this work was done on naturally low contaminated material, and thus, represents the natural contamination of these products. To gain further insights on the fate of other mycotoxins during this production process, it is recommended to artificially contaminate the raw material, or to infect it with toxigenic fungi to be able to establish a mass balance.

To our knowledge, this is the first study on the fate of multiple mycotoxins in the production of legume protein and other derived products. The findings of this study suggests that certain processing steps can impact mycotoxin levels in legumes and legume-derived products. Notably, the milling process has the potential to release mycotoxins from their natural matrix. On the other hand, the study also highlights the effectiveness of the thermal treatment and solubilization in reducing mycotoxin concentrations in the final products. These observations emphasize the importance of implementing effective processing techniques to minimize mycotoxin contamination in legume-derived products. More studies, including a higher number of samples, different seasons and regions are needed to analyze the presence of AME in the final product B.

Methods

Materials

OTA, ±10 µg/mL in acetonitrile (ACN), aflatoxin mix containing AFB1, AFB2, AFG1, AFG2, ±20 µg/mL in ACN, DON, ±100 µg/mL in ACN, zearalenone (ZEN), ±100 µg/mL in ACN, fumonisin mix containing FB1, FB2, ±50 µg/mL in 50/50 ACN/H2O, NIV, ±100 µg/mL in ACN, neosolaniol (NEO), ±100 µg/ml in ACN, DOM, ±50 µg/mL in ACN, T2, ±100 µg/mL in ACN, HT2, ±100 µg/mL in ACN, 3-ADON, ±100 µg/mL in ACN, 15-ADON, ±100 µg/mL in ACN, diacetoxyscirpenol (DAS), ±100 µg/mL in ACN, fusarenon-X (FX), ±100 µg/mL in ACN, STC, ±50 µg/mL in ACN, all the above components were acquired from Biopure, Food Risk Management B.V., RomerLabs agent (Oostvoorne Nederland). Roquefortine C (ROQ-C), 0,5 mg, was purchased at Enzo Life Sciences, (Axxora Platform, Brussels, Belgium), ZAN, 10 mg, at Sigma (Darmstadt, Germany), AME, 5 mg, and alternariol (AOH), 5 mg, at Sigma (Darmstadt, Germany), FB3, 1 mg, at Medical Research Council (Swindow, UK), ENN B, ±1 mg/mL in methanol was acquired at Fermentek (Jerusalem, Israel). Acetic acid was provided by Merck (Overijse, Belgium), Methanol absolute LC-MS (for mobile phase) by Biosolve Chemicals (Valkenswaard, The Netherlands), ammonium acetate by Merck (Overijse, Belgium), n-hexane by BDH HiperSolv CHROMANORM (VWR International, Leuven, Belgium), ACN HPLC by Biosolve Chemicals (Valkenswaard, The Netherlands). All used solvents were HPLC or MS-grade.

Samples

A total of 41 legume samples of 1 kg each from a legume processing company were included to examine the fluctuations in mycotoxin levels during processing. The samples originated from two production batches with different production dates and different raw material suppliers, and were followed from the raw material to the final products. Due to a non-disclosure agreement, the nature of the legume from the Fabaceae family and the company cannot be disclosed.

For quantification, matrix matched calibration curves (MMCC) were constructed on blank legume flour. The blank sample must be similar to the products being analyzed and the mycotoxin concentration should be lower than the limit of detection (LOD)37.

For the construction of the MMCC, five blank samples were spiked with a multi-mycotoxin mix and were stored in the dark for 15 min so the mycotoxins could be absorbed in the samples. Sample preparation was performed using a validated method as described by Chilaka et al.20. Briefly, each sample was thoroughly homogenized, and milled, from which a representative portion of 5 g was spiked with internal standards zearalanone (ZAN) and deepoxy-deoxynivalenol (DOM) at a concentration of 250 and 150 μg/kg, respectively. DOM was used as internal standard for DON, DON-3G, 3-ADON, and 15-ADON while ZAN was used for the other mycotoxins. The sample was extracted with 20 mL of extraction solvent [ACN/H2O/acetic acid (79/20/1, v/v/v)], vortexed, agitated and centrifuged at 4000 × g for 15 min. The supernatant was passed through a pre-conditioned C18 SPE column (Phenomenex, Utrecht, The Netherlands). The extraction was repeated by adding 5 mL of the extraction solvent ACN/H2O/acetic acid (79/20/1, v/v/ v) and the combined extract was made up to 25 mL with the extraction solvent (ACN/H2O/acetic acid (79/20/1, v/v/v)). The extract was transferred into the extraction tube and defatted using 10 mL of hexane. The defatted extract was divided into two parts, 10 mL of the first part was filtered through a glass filter, and 12.5 mL of the second part was mixed with 27.5 mL of ACN/acetic acid (99/1, v/v). Thirty milliliters of the mixture were further purified by passing through a MultiSep® 226 AflaZon+ multifunctional column (Romer Labs, Gernsheim Germany). The total Multisep 226 eluate was added to 2 mL of the glass filtered portion, and the mixture was evaporated to dryness under a gentle nitrogen flow in a thermostatic water bath heated at 40 °C. The residue was then redissolved in 150 μL of the injection solvent containing mobile phase A (H2O/MeOH/acetic acid (94/5/1, v/v/ v) +5 mM ammonium acetate) and mobile phase B (H2O/MeOH/ acetic acid (2/97/1, v/v/v) +5 mM ammonium acetate) mixed in the ratio of 3/2, v/v. The redissolved extract was centrifuged for 10 min at 14000 g, and further filtered using Ultrafree® PVDF centrifuge filters (Millipore Bedford, MA, USA) and analyzed using LC-MS/MS.

LC-MS/MS analysis

An LC-MS/MS Waters Acquity HPLC coupled with a Micromass Quattro Premier XE triple quadrupole mass spectrometer was used. The column used was a Symmetry C18 (150 mm × 2.1 mm i.d. 5 µm) column with a guard column (10 mm × 2.1 mm i.d.) of the same material (Waters, Zellik, Belgium), and was kept at room temperature. During the run, a gradient elution program was employed. This gradient program involved varying amounts of mobile phase A and mobile phase B. These were used at a flow rate of 0.3 mL/min. The total analytical run time was 28 min. The gradient started at 95% mobile phase A with a linear decrease to 35% in 7 min. Mobile phase A decreased to 25% at 4 min, and an isocratic period of 100%, mobile phase B started at 11 min for 2 min. Initial column conditions were reached at 23 min using a linear decrease of mobile phase B, and the column was reconditioned for 5 min before the following injection. An injection volume of 10 μL was used. The capillary voltage was 3.2 kV and nitrogen was used as the desolvation gas. The mass spectrometer was operated with selected reaction monitoring (SRM) channels in the positive electrospray ionization (ESI+) mode. Source and desolvation temperatures were set at 150 and 350 °C, respectively. The instrumental control and data processing were performed with Masslynx™ version 4.1 and Quanlynx® version 4.1 software (Micromass, Manchester, UK). Prior to the study, the cone voltage at which the precursor ion is most abundant for each component was determined, as well as the corresponding precursor ion. Then, the collision energy and fragment ions formed, and their corresponding ionization mode were determined. More information about the MS/MS transition of each mycotoxin and the validation parameters according to Commission Regulation (EC) No. 401/2006 (68) and Commission Decision (EC) No. 2002/657 are described in refs. 20,37,38,39 and in the supplementary material. The performance of the method, including the LOD, LOQ, accuracy and precision, expressed as apparent recovery (Rapp), repeatability (RSDr), and intermediate precision (RSDR) are included in the Supplementary Material 2 and 3. The Waters QuanLynx software was used to determine the mycotoxin concentration in the food samples. For each set of analyses, a MMCC was constructed. When the identification criteria for mycotoxins in food and feed stated in the Guidance document SANTE/12089/201640 were met, but the concentration was lower than the LOD or LOQ, the sample was considered positive, but no concentration is given.