MONITORING OF MYCOTOXINS AND PESTICIDES IN WINEMAKING

This study monitored concentrations of both pesticides 2,4-dichlorophenoxyacetic acid (2,4-D) and procymidone, and mycotoxin ochratoxin A (OTA) in stages of the winemaking process. Sampling was carried out in the usual vinification process of red wine in a winery between the steps to obtain must and alcoholic fermentation. The highest transference of contaminants in the process occurred in the crushing step to 2,4-D (100%) and maceration to OTA and procymidone (100%). Removal of contaminants in the winemaking process corresponded to 100%, with a half-life (T 1/2 ) longer for procymidone (216.5 h) and shorter for 2,4-D (38.5 h) and OTA (96 h). The processing factors (PFs) (0) for the contaminants, together with the data obtained, characterize winemaking as a process of reducing mycotoxin and pesticides. Results highlight the importance of fermentation to reduce contaminants and that yeasts promote detoxification RESUMO Este estudo monitorou, nas etapas do processo de vinificação, a transferência e a redução dos pesticidas 2,4-D, procimidona e da micotoxina ocratoxina A (OTA). A amostragem foi realizada durante processo usual de vinificação em vinho tinto em vinícola para o acompanhamento das concentrações de OTA e agrotóxicos entre as etapas de obtenção do mosto e fermentação alcoólica. A maior transferência dos contaminantes no processo ocorreu nas etapas de esmagamento para 2,4-D (100%) e maceração para OTA e procimidona (100%). A remoção dos contaminantes durante o processo de vinificação correspondeu a 100%, tendo tempo de meia vida (T 1/2 ) maior para procimidona (216,5 h) e menor para 2,4-D (38,5 h) e OTA (96 h). Os fatores de processamento (PFs) (0) para os contaminantes, juntamente com os dados obtidos caracterizam a vinificação como processo de redução de micotoxina e agrotóxicos. Os resultados destacam a importância da fermentação na redução de contaminantes, indicando que as leveduras promovem a detoxificação


INTRODUCTION
The world's wine production was estimated to 260 million hectoliters (hL) in 2020. Brazil produced 3.6 million hL (OIV, 2021), mainly in Rio Grande do Sul (RS) state, which accounts for 62.51% of the national production (Mello and Machado, 2021). Moderate wine consumption has shown beneficial health effects associated with phenolic compounds, such as anthocyanins, quercetin and resveratrol (Giovinazzo and Grieco, 2015;Gabrielyan and Kazumyan et al., 2018;Freire et al., 2020;Ulrih et al., 2020;Tıraş et al., 2022). Despite its benefits, some contaminants, such as pesticides and mycotoxins, can be detected in this beverage sold worldwide and may cause negative effects on human health (Čepo et al., 2018;Freire et al., 2020;Čuš et al., 2021).
The occurrence of pesticides is related to their use in vineyards to prevent fungal colonization. In addition, cross-contamination by pesticides from other agricultural crops (e.g., rice and soybeans), such as 2,4-dichlorophenoxyacetic acid (2,4-D), can also cause negative impacts on the wine sector (IBRAVIN, 2019). Even though this auxin-type herbicide is commonly used for exterminating weeds in grain and cereal crops (Rossouw et al., 2019), effects of its spray particles may be seen several kilometers away from the target, particularly under conditions of high temperature, low relative humidity and prevailing winds (Felsot et al., 2010).
In addition to the toxic effect on humans, contaminants, such as fungicides in beverage production, may also delay alcoholic fermentation (Briz-Cid et al., 2018), and influence the composition of volatile compounds in wines (Oliva et al., 2015). Among fungicides, procymidone has been used for fighting diseases that affect vineyards worldwide, as shown by its occurrence in wines described by Čuš et al. (2010a), Doulia et al. (2016Doulia et al. ( , 2017 and Romanazzi and Feliziani (2014). Pesticides application in order to protect vineyards may also be linked to the mycotoxin biosynthesis often related to ochratoxin (A) (OTA) (Gil-Serna et al., 2018;Costa et al., 2019), mainly by Aspergillus carbonarius, the main fungus contaminant in vines Medina et al., 2007;Magistà et al., 2021). This mycotoxin is described as harmful for the fermentative capacity of yeasts during alcoholic fermentation (Esti et al., 2012), and causes changes in wine composition. Thus, it affects its flavor and color (Freire et al., 2020).
Therefore, methods of pesticide and mycotoxin decontamination have become a unique remediation alternative and must be investigated in the wine production chain. Emphasis should be given to biological conditions, with the use of yeasts, such as Saccharomyces cerevisiae, and usual conditions which compose usual winemaking processes (Tempère et al., 2018). The use of this yeast is essential to the production process, since studies have shown its potential to remove OTA and pesticides, i.e., rates above 90% of contamination observed in wine (Csutorás et al., 2013;Pan et al., 2018).
Therefore, this study aimed to monitor concentrations of both pesticides 2,4-D and procymidone and mycotoxin OTA in stages of the winemaking process. Besides, this study, which was carried out on an industrial scale, contributed to the characterization of the winemaking as a process to reduce contamination of this widely consumed beverage.

Pesticides and OTA
Both pesticides 2,4-D and procymidone and mycotoxin OTA with purity above 98% were purchased from Sigma-Aldrich Brazil. Stock solutions of every pesticide (1000 µg/mL) were prepared in acetonitrile (MeCN). OTA stock solution was prepared by dissolving 1 mg of mycotoxin in benzene: MeCN (98:2 v/v) to a concentration of 100 µg/mL (AOAC, 1995).

Sampling
On an industrial scale (1,200 L), sampling was carried out in the usual vinification process of red wine produced by a winery. Initially, grapes were stemmed. Grains were slightly broken and transported to oak barrels with the addition of pectolytic enzyme (3 g/hL) and sulfur dioxide (10 g/hL). Maceration and alcoholic fermentation took place in oak barrels under temperature control (22 °C) with inoculation of non-Saccharomyces yeast Torulaspora delbrueckii (3 g hL-1), along with S. cerevisiae (10 g/hL) (Zymaflore, Laffort, Bordeaux, France). 48 h after the beginning of alcoholic fermentation, 1 g/hL of Oenococcus oeni lactic acid bacteria was added to promote malolactic fermentation. Devatting (separation of solid and liquid parts) was carried out one week after the beginning of malolactic fermentation, which was completed within one week after devatting while alcoholic fermentation ended three weeks after the end of malolactic fermentation.
Samples were collected at the end of every stage of the process, corresponding to crushing of grapes to obtain must (0 h time), pre-fermentative maceration with addition of non-Saccharomyces yeast (24 h), addition of S. cerevisiae yeast (120 h), devatting (480 h), stop of alcoholic fermentation (1062 h), and end of alcoholic fermentation (1230 h). Samples were placed in 250 mL vials and kept frozen until analysis.

Determination of 2,4-D and procymidone
Extraction of herbicide 2,4-D and fungicide procymidone from must and wine samples in stages of the winemaking process were carried out by the Quick, Easy, Cheap, Effective, Rugged and Safe (QuEChERS) method described by Payá et al. (2007) with modifications, that is, elimination of sesquihydrated and dehydrated citrate salts. Extraction was performed with the use of MeCN while the extract was cleaned with magnesium sulfate (MgSO4) and sodium chloride (NaCl) salts. The resulting extract was used for identifying and quantifying pesticides by liquid chromatograph with a diode array detector (LC-DAD) and gas chromatograph coupled to a mass spectrometer (GC-MS).
An RTX-5MS column (30 m x 0.25 mm ID, 0.25 μm) was used in a Shimadzu Gas Chromatograph (Kyoto, Japan) QP2010 Plus equipped with an automatic sampler (AOC-20i) coupled to a spectrometer mass with quadrupole mass filter (GC-MS) to separate the fungicide procymidone. The source was electron ionization (EI) with energy of 70 eV. Helium was used as the carrier gas and the injection volume was 1 µL, as described by Barbosa et al. (2020).

Determination of OTA
OTA extraction from samples was carried out by the QuEChERS method described by Fernandes et al. (2013). Extraction was performed with acidified MeCN (1% acetic acid) while partitioning was done with the addition of a mixture of salts containing MgSO4, NaCl, tribasic sodium citrate dihydrate and sodium citrate dibasic sesquihydrate. The supernatant (1.5 mL) was transferred to an amber flask and the solvent was evaporated under a stream of nitrogen. The dry extract was resuspended in 1 mL of mobile phase (60% MeCN, 40% Mili-Q water acidified with 1% acetic acid), followed by chromatographic analysis.
In separation, identification and quantification, the analytical chromatographic column Supelco® -Kromasil C18 (5 µm and 150 mm x 4.6 mm) was used, kept at 25 ºC in a Shimadzu Liquid Chromatograph (Kyoto, Japan) with a fluorescence detector (LC-FL). Wavelengths were 333 nm and 460 nm for excitation and emission, respectively; the mobile phase was composed of MeCN: 1% acidified water with acetic acid (60:40, v/v), and flow rate was 0.8 mL/min, as described by Garcia et al. (2020).

Method Validation of 2,4-D, procymidone and OTA
Methods of 2,4-D, procymidone and OTA determination were validated according to the guidelines issued by the European Commission -EC (EC, 2006; and the National Health Surveillance Agency -ANVISA (ANVISA, 2017). Concentrations used for constructing the analytical curves in the matrix extract (must and wine) ranged from 0.05 to 2.5 µg/mL, resulting in seven levels of 2,4-D; from 0.0075 to 1.5 µg/mL, totaling twelve levels of procymidone; and from 0.1 to 3.0 ng/mL, totaling eleven levels of OTA. Based on analytical curves, linearity of the methods was evaluated, considering the coefficient of determination (R 2 ). Recovery was determined by fortification in both matrices (must and wine) at three concentrations: 0.07, 0.5 and 1.0 µg/mL for 2,4-D; 0.075, 0.5 and 1.0 µg/mL for procymidone; 0.13, 1.0 and 3.0 ng/mL in must and 0.13, 1.0 and 2.5 ng/mL in wine for OTA.

Data treatment
Dissipation of contaminants in the winemaking process was estimated by the first-order kinetics equation and calculated by Equation 1, where C (µg/L for mycotoxin and mg/L for pesticides) is the residual amount of the contaminant at time t, C0 is the initial residual amount of the contaminant, and k (h-1) is the dissipation constant. Half-lives (T1/2) of every contaminant were calculated by Equation 2, (Pan et al., 2018;Hou et al., 2020).
Processing factors (PFs) were calculated according to the recommendations issued in the Joint Meeting on Pesticide Residues (JMPR, 2006) by the Food and Agriculture Organization of the United Nations (FAO) and the World Health Organization (WHO) to show residue disposal in processed products -Equation 3. PFs values < 1 show that there is decrease in contaminant levels after raw material processing, while values > 1 mean that the contaminant level increased on processing (Pan et al., 2018;Hou et al., 2020). PFs = residuale level in processed product residuale level in commodity to be processed

Eq. 3
The removal and transference of contaminants in the winemaking process were evaluated by Equations 4 and 5 (Hou et al., 2020). PAST 2.04 software (Hammer et al., 2001) was used for the principal component analysis (PCA) of the data set (triplicate) on concentration, removal and transference of OTA, 2,4-D and procymidone.

RESULTS AND DISCUSSION
Validation of the method for 2,4-D, procymidone and OTA determination In this study, quality parameters of determination of 2,4-D, procymidone and OTA exhibited results that showed the analytical quality of the proposed methods (Table I). Mean recoveries of herbicide 2,4-D ranged between 88 and 90% in must and from 72 to 100% in wine at three concentration levels (0.07, 0.5 and 1.0 µg/mL) with relative standard error (RSD) varying from 1.0 to 5.6% in must and from 1.4 to 8.5% in wine. Fungicide procymidone showed recoveries ranging from 99 to 104% in must and from 78 to 86% in wine with RSD from 1.5 to 4.9% in must and from 2.5 to 7.6% in wine. The proposed method of OTA mycotoxin showed recoveries between 98 to 112% in must (RSD between 2.5 and 2.6%) and from 104 to 108% in wine (RSD between

Levels of OTA and pesticides in the winemaking process
Both pesticides procymidone and 2,4-D and mycotoxin OTA were detected in steps of the winemaking process, as shown in Table II. However, for each contaminant there is a significant transference step. The highest levels of OTA and procymidone were detected in pre-fermentative maceration (contact of must with grape skin). In this step, OTA concentrated 2.0 times and procymidone 1.5 times. Some studies suggest that the highest transference of OTA and fungicides in the winemaking process occurs during maceration. This stage promotes extraction of anthocyanins from grape skin, thus, generating the desired color of red wines, but it also promotes extraction of other compounds, such as contaminants found in the matrix (González-Rodríguez et al., 2009;Čepo et al., 2018;Kochman et al., 2021).
Regarding herbicide 2,4-D, the highest transference was observed in the crushing step (Table II), in which its concentration was 4-fold higher than in the other steps. In this step, grape skin breaks to facilitate maceration (Kochman et al., 2021). This herbicide is not recommended for growing grapes since its contamination in vineyards is due to drift (Felsot et al., 2010;Rossouw et al., 2019). High transference of this herbicide may also be related to polarity. Besides, it has high solubility in water (24,300 mg/L) and low octanol-water partition coefficient (log Kow -0.82) (Lewis et al., 2006;FAO, 2020).
Some compounds tend to be more easily detected than others in liquid samples due to the wide range of physicochemical properties of pesticides (polarity, solubility in water, Kow) (Pichon et al., 1998). According to Manjarres-López et al. (2021), the more soluble pesticides are more easily detected in water; besides, the less soluble they are, the more easily detected in the soil. Therefore, more apolar and less water-soluble pesticides, such as procymidone (2.46 mg/L of water solubility and 3.3 log Kow) Lewis et al., 2006), require the maceration step to be transferred from grapes to must in the winemaking process (Table II).
Removal of the three contaminants (OTA, 2,4-D and procymidone) was observed in alcoholic fermentation with the addition of S. cerevisiae yeast.
They can activate their own mechanisms of detoxification and produce molecules that have  (Bangar et al., 2021;Yousefi et al., 2021). In our study, the microorganisms (lactic acid bacteria and yeasts) were used in the winemaking process, which can also be characterized by the total elimination of the three contaminants at the end of the process.

Dissipation of OTA and pesticides in the winemaking process
Results of dissipation of the three contaminants in the winemaking process (Table III) showed that half-life of procymidone was 216.5 h longer than half-lives of other contaminants, 2,4-D and OTA (T1/2= 38.5 and 96 h, respectively). Procymidone has low acute oral toxicity; oral LD50 in rats and mice has been reported in the literature as > 5000 mg/ g body weight Rifai et al., 2013). Acute oral toxicity of 2,4-D in rats is represented by an LD50 value of 699 mg/kg body weight (Munro et al., 1992;Bus and Hammond, 2007;FAO, 2020). Oral LD50 values of OTA in rats and mice are 20-30 and 46-58 mg/kg of body weight, respectively (FAO/WHO, 2001).
Thus, OTA and 2,4-D are classified as possible carcinogens to humans (Group 2B) by the International Agency for Research on Cancer (IARC, 1993;. Effectiveness of detoxification of different toxic compounds, such as pesticides and mycotoxins, due to the activity of yeasts, depends on several factors, such as the strain type, concentration, pH of the medium and incubation time in the process (Yousefi et al., 2021). 2,4-D and OTA have high toxicity than procymidone and may be metabolized in shorter process time due to their effect on microorganisms in the fermentation medium (120 h or 5 days), as shown in Table III. There was removal of 12 and 54% to both contaminants between the steps of pre-fermentative maceration and addition of yeasts. In contrast, in the case of fungicide procymidone, removal was 35% between these steps, suggesting that low toxicity of the fungicide at high residual concentrations can be observed in other steps of the winemaking process (1062 h or 44 days) (Table III). These data show that S. cerevisiae yeast has medium detoxification power, which may be 15 associated with its metabolism or adsorption capacity, in addition to the contaminant toxicity (Petruzzi et al., 2015).

Table III
Process time (T), first-order constant (k), half-life time (T½) and correlation coefficient (r) of dissipation of compounds in the winemaking process

Transference and removal of contaminants
Removal and transference of OTA, procymidone and 2,4-D and PFs that demonstrate the effect of the winemaking process on the residual levels of these contaminants are shown in Table IV. PFs values > 1 mean that the contaminant level increased on processing (Pan et al., 2018;Hou et al., 2020). Thus, results demonstrate that the highest transference occurred in the maceration step in the cases of OTA and procymidone, 100% (PFs = 1), that is, contaminants were concentrated in this step. However, the highest transference 100% (PFs = 1) of 2,4-D occurred at the beginning of the process (grape crushing) ( Table IV).
Transference of contaminants from grapes to must and from must to wine depends on the winemaking process and on physicochemical properties of contaminants, mainly lipophilicity (log Kow) and solubility (Alister et al., 2014). As previously reported, pesticides with high water solubility and low lipophilicity, including herbicide 2,4-D (solubility in water 24,300 mg/L and log Kow -0.82) (Lewis et al., 2006;FAO, 2020), may be less soluble in the system, that is, constituents found in grape skins and seeds, such as lignin, cellulose, hemicellulose, sugar, pectin, protein, polyphenols, minerals and lipids (Jiang et al., 2011;Gowman et al., 2019;Yoon et al., 2021), can adsorb them and contribute to the decrease (Yoon et al., 2021) in the crushing step, differing from other contaminants that showed higher transference in the maceration step (Table IV).
Other results observed in the winemaking process, mainly in the devatting step are shown in Table IV. In this step, both pesticides procymidone and 2,4-D and mycotoxin OTA were removed between 99 and 100%, respectively, with PF values below 1 (0 for the three contaminants). These values suggest that the winemaking process reduces concentrations of contaminants in red wines.
Similar results were found by Pan et al. (2018) and Hou et al. (2020), who demonstrated on a laboratory scale that PFs of different pesticides (zoxamide, boscalid, picoxystrobin, fluopicolide, 2,6dichlorobenzamide, pyraclostrobin and mandipropamid) were below 1 (factor of removal) after the winemaking process, thus, highlighting that the process may reduce pesticide residues in wines.
Results of great relevance were found in the case of different contaminants with 100% elimination at the end of alcoholic fermentation (Table IV). Regarding mycotoxin OTA, the highest transference (100%) was observed from grape to must in maceration, in comparison with the other contaminants under investigation (Table IV). Some studies reported that OTA contents detected in red wines are more frequent, mainly due to mandatory maceration of grape skin, which can enhance OTA extraction, as observed in the following ranges: from 0.02 to 0.134 µg/L (Torović et al., 2020); 0.29 µg/L (Freire et al., 2017); 0.45 µg/L (Remiro et al., 2013); and 2.47-2.77 µg/L (Abreu et al., 2016). In the present work, Table IV shows that there was neither detection of 2,4-D and OTA contaminants in the devatting step nor detection of procymidone at the end of the alcoholic fermentation (100% elimination).
For a wine to be considered dry, it must contain less than 4 g/L of sugar (expressed in grams of glucose per liter) (Brasil, 2014). When the yeasts stopped fermenting, the wine had 7.5 g/L of sugar. Stop in fermentation can harm ethanol production, generate costs with oenological inputs to carry out refermentation and lead to losses in organoleptic characteristics of wines (González-Álvarez et al., 2012;Briz-Cid et al., 2018).  Weak acids, such as 2,4-D, are known to have a negative impact on yeasts performance, thus, restricting efficiency in ethanol production and other products derived from their fermentative activity (Cabral et al., 2003;Ndukwe et al., 2020). Cabral et al. (2003) evaluated, on a laboratory scale, the inhibitory effect of 2,4-D on S. cerevisiae growth. The authors reported that this effect is strongly dependent on pH of the fermentation medium (from 2.5 to 6.5); the yeast strain (W303.1b) grown at 30 °C in MM2 growth medium exposed to increasing concentrations of 2,4-D was affected at pH 3.0, leading to a reduction in growth and loss of cell viability. Weak lipophilic acids, including 2,4-D, exert negative effects on yeasts when they diffuse into cells in their non-ionized state, because their pH is lower than the pKa of the yeast growth medium (Cabral et al., 2003;Ndukwe et al., 2020).
In the present study, alcoholic fermentation conducted by S. cerevisiae occurred at pH 3.1-3.2, close to the pKa (2.73) of 2.4-D (Cabral et al., 2003) on an industrial scale. Therefore, stop in fermentation may also have been influenced by pH and toxicity of contaminants in the medium. It should be mentioned that herbicide 2,4-D is not recommended for growing vines. The study reported by this paper evaluated its total elimination (Table IV), highlighting the importance of reducing levels of contaminants in the winemaking process.
Briz-Cid et al. (2018) assessed the influence of fungicides on sensory properties and flavonoid composition of red wines. The authors observed that fermentation kinetics was influenced by fungicides. Wine grapes treated with boscalide (200 mg/mL) combined with kresoxim-methyl (100 mg/mL) exhibited initial delay in alcoholic fermentation. Furthermore, the authors described that residues of these fungicides may affect sensory quality of wine and decrease intensity of fruity aromas in wine by 42% and 59%, respectively.
The wide range of physicochemical (polarity, water solubility) and toxicological properties of the product is a consequence of the different levels of removal and transference of these contaminants in the winemaking process (Table IV). This study is extremely relevant because it enables to observe and describe the winemaking process related to the elimination/removal of contaminants so as to contribute to and strenghten the wine sector.
To better understand the relationship between removal of contaminants and stages of the winemaking process, the PCA was performed (Figure 1), highlighting three regions. The first component (PC 1) explains 68% of total variance. Therefore, almost 2/3 of information found in 9 database variables can be encompassed by this component. The second component (PC 2) explains 28% of total variance. Therefore, 96% of total variance is explained by only two components.
The circled region encompasses PFs of OTA, 2,4-D and procymidone. The Pearson's correlation shows positive and significant relation (R = 0.99997, p = 1.7x10 -5 ) between PFs of 2.4-D and PFs of OTA and (R = 0.98542, p = 0.0003) between PFs of procymidone and PFs of 2,4-D. It shows that, in grape processing, there is removal in contaminant levels (Table II), mainly related to S. cerevisiae yeast, and points out that alcoholic fermentation reduces contamination of wines.
Another region framed by a square encloses the removal of OTA, procymidone and 2,4-D in the winemaking process, and shows that, from the devatting step onwards, removal of procymidone was 99% while removal of OTA and 2,4-D was 100%. At the end of the alcoholic fermentation, the three contaminants (100%) were completely removed (Tables II and IV). Based on these results, winemaking can be characterized as a process of reducing contaminants, mycotoxins and pesticides.

CONCLUSIONS
The occurrence of three contaminants -mycotoxin OTA, fungicide procymidone and herbicide 2,4-Dwas observed in the winemaking process. The highest transference of procymidone and OTA were detected in the maceration step (100%) and of 2,4-D, in the crushing step (100%). In the winemaking process, the highest dissipation was for procymidone (T1/2 = 216.5 h). Removal of contaminants at the end of the process was 100% and PFs were below 1, thus, showing that the winemaking process can reduce their residues in wine. Results highlight the importance of fermentation to reduce contaminants and that yeasts promote detoxification.