Thermal processing-induced changes in volatilome and metabolome of Brussels sprouts: focus on glucosinolate metabolism

This study aimed to investigate the influence of different thermal treatment methods (boiling, steaming) on the metabolome and volatilome of Brussels sprouts. Targeted, quantitative analysis of primary metabolites (amino acids and sugars), glucosinolates as well as untargeted analysis of volatile organic compounds (VOCs) were performed. The composition of primary metabolites was affected only by boiling, where the loss of nutrients was observed. Glucosinolate content increased after cooking, which can be explained by the loosening of the plant tissues and its better extractability. Most marked changes were noticed in the volatilome. Among the degradation products of glucosinolates, isothiocyanates were the main chemical group in the raw and boiled vegetable. In contrast, steaming favored the formation of nitriles. Another dominant group of VOCs were aldehydes, which were the most abundant in raw Brussels sprouts and their level significantly decreased after thermal treatment. It is well known that isothiocyanates are substances with widely proven bioactive properties. Therefore, their formation in the glucosinolate degradation process is favorable. Based on the obtained results, it was concluded that boiling seems to be the most beneficial form of Brussels sprouts’ processing.


Introduction
In recent years, consumers have become more aware of the nutritional value of food products, as well as the influence of food processing on the quality of the final product. Many products, including vegetables, require thermal processing prior to consumption, which can affect both their nutritional quality and palatability. Thus, the influence of thermal processing on the quality of food has recently gained considerable scientific attention [1]. It is known that elevated temperature causes the degradation of several compounds, resulting in the formation of new, frequently occurring volatile organic compounds (VOCs), which can contribute to the aroma of the product. Thermal treatment is one of the factors enhancing palatability of many food products, through flavor and structural changes. Therefore, the acceptance of thermally processed food is greater than that of the raw one; however, many questions arise regarding the impact of the available thermal processing techniques on the bioactive potential of the food. During thermal processing, several reactions occur which contribute to the degradation of thermally labile compounds and the formation of new, often volatile metabolites.
Brussels sprouts are one of the richest sources of glucosinolates (GSL), unique compounds found mostly in Brassica plants. GSL, and more precisely their breakdown products, isothiocyanates (ITC), are considered chemopreventive agents with anti-cancer properties [2]. Brassica vegetables are eaten after boiling, steaming, frying and fermenting as well as in their raw form. Thermal processing in an aqueous medium might cause leaching of many phytochemicals into the water, as well as thermal degradation of compounds, influencing their nutritional value and flavor [3]. In the intact plant tissue, the GSL are separated from the enzyme myrosinase, which has the ability to hydrolyze them into certain breakdown products. After disrupting the structure of the plant tissue, the enzyme comes into contact with the substrate and catalyzes the reaction. When these vegetables are processed, myrosinase can interact with GSL and depending on the reaction condition, such as pH, temperature and the 1 3 presence of cofactors, hydrolyzes them into ITC, nitriles or epithionitriles [4]. It was well proven that the gut microbiota can hydrolyze GSL into their bioactive forms, thus eating vegetables even after myrosinase inactivation is still beneficial for our health [5]. Apart from the health-beneficial properties, ITC together with other sulfur compounds are responsible for the typical, sulfurous flavor of Brassica vegetables [4]. Moreover, Brassica vegetables are also a source of other substances influencing their flavor and bioactive potential, such as amino acids, sugars, and vitamins. Many of these compounds are water-soluble and heat-sensitive, therefore, they might be lost during food processing [6].
Previously published research indicated that thermal processing results in a significant loss of GLS in rutabaga and cauliflower (Kapusta-Duch et al., 2016). However, the less dense structure of these plants might result in the leaching of these components into the water, which was not examined in that study. The research focused on Brussels sprouts showed that their content of GSL even when the vegetables had been cooked for 30 min was still relatively high, which can be explained by their more compact form [7]. Importantly, the content of GSL in cooking water increased with prolonged cooking time, but it was not equal to the loss of GSL from the vegetable.
Food metabolomics is a valuable and promising tool for food scientists to understand the metabolome of food, including its biochemistry and composition [8]. Food metabolomics as one of the "omics" technologies, offers enormous opportunities to obtain detailed information that can be correlated to the functional and nutraceutical composition of foods. The use of high throughput, hyphenated analytical techniques together with advanced statistical approaches constitutes a powerful tool providing the information on the nutritional quality of foodstuffs and changes caused by food processing, which can contribute to their overall acceptance. A specific branch of metabolomics is volatilomics, which focuses of the VOC profile [9].
To date, extensive studies on the effect of thermal processing on the GSL metabolism in Brassica vegetables have been conducted [7,10,11]. However, there is a gap in the knowledge regarding flavor formation, metabolic interactions and formation of VOCs during thermal processing. Until now, no studies were performed to comprehensively track changes in both the volatile and non-volatile (primary and secondary) metabolites in Brussels sprouts during processing. Knowing that thermal processing has a great influence on the stability of phytochemicals and leads to the formation of various VOCs, it can be expected that substantial changes in the volatilome and metabolome occur during the thermal processing of Brussels sprouts. However, the directions of changes depending on the type of thermal processing, especially the VOC formation, are unknown. Therefore, the aim of this study was to determine the concentrations of primary (sugars, amino acids) and secondary (GSL) metabolites and to combine them with the formation of VOCs analyzed using comprehensive, two-dimensional gas chromatography (GC × GC) in Brussels sprouts during thermal treatment, using the foodomics approach in order to understand the influence of different processing methods on the formation of VOCs in Brassica vegetables.

Plant materials
Brussels sprout was purchased in a local grocery store (Poznań, Poland) in the summer season. Samples for GC × GC-ToF analysis as well as the methanol extracts were prepared on the day of purchase. For the boiling process, around 200 g of vegetables were placed in 1 L of boiling water and cooked for 8 min. The same amount was placed in the steamer basket over boiling water and the vegetables were steamed for 10 min.

Non-volatiles analysis
The raw/boiled/steamed vegetables were frozen in liquid nitrogen and homogenized in a laboratory mill. Then, 30 mL of MeOH were added to 5 g of still frozen Brussels sprout powder, and then the ultrasound-assisted extraction (40 kHz) was performed for 15 min at 30 °C. The extraction was repeated two times to assure the exhaustive character of the procedure. The extracts were combined, filtrated through Nylon filter and analyzed by HPLC with adequate detectors.

HPLC analysis of amino acids
Determination of amino acids was carried out on an Agilent Technologies 1100 series HPLC system with a diode-array detector (DAD) (Agilent Technologies, Waldbronn, Germany). Automatic pre-column derivatization with orthophthalaldehyde/ 9-fluorenyl-methyl chloroformate (OPA/ FMOC) reagents was performed at room temperature. After the derivatization, 0.5 μL of the mixture was injected for each chromatographic separation into the AdvanceBio Amino Acid Analysis (AAA) column (Agilent Technologies). The flow rate was 1.5 mL/min. The mobile phase A was 10 mM Na 2 HPO 4 , and phase B 10 mM Na 2 B 4 O 7 pH 8.2 (phase A) and acetonitrile:methanol:water (45:45:10, v:v:v) (phase B). The gradient was 2% for 1 min, then linearly increased to 35% until 13.4 min, then to 15.8 min to 2% (of phase B) and kept until 18.0 min with the flow of 1.5 mL. The eluent was monitored at 338 nm for OPA derivatives (all aminoacids but proline and hydroxyproline, which were monitored as FMOC derivatives at 262 nm). Amino acids were identified and quantified using external standards based on retention times and prepared six points calibration curves. Were prepared to facilitate quantitation.

HPLC analysis of sugars
Determination of carbohydrates was carried out as described previously [12] on an Agilent Technologies 1100 series HPLC system with a refractive index detector (Agilent Technologies, Waldbronn, Germany). The analysis was performed isocratically at a flow rate of 0.6 mL/min at 80 °C on the Rezex RPM-Monosaccharide Pb+2 300×7.8 mm column (Phenomenex, Torrance, CA). Water was used as the mobile phase. Commercial standards were used for the identification and quantification of individual sugars.

HPLC analysis of glucosinolates
Glucosinolates were analyzed as described previously [13]. Individual glucosinolates were analyzed using an Agilent Technologies 1100 series HPLC system with a diode-array detector (Agilent Technologies, Waldbronn, Germany). Separation was performed using a Kinetex® C18 column (250 × 4.6 mm; 5 um). Desulpho-glucosinolates were separated with a gradient of water (phase A) and 20% acetonitrile (phase B) at a flow rate of 0.6 mL/min. Detection was performed at 229 nm. The injection volume was 20 uL. The quantification was performed based on calibration curves prepared with external standards for each glucosinolate separately.

GC × GC-ToFMS analysis
The VOCs were isolated using headspace solid-phase microextraction (HS-SPME) with divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) fiber and analyzed using comprehensive two-dimensional gas chromatography-time of flight mass spectrometry (GC × GC-ToFMS) for the comparative, semiquantitative (peak area percentage) analysis only. The vegetables were homogenized in liquid nitrogen, then 2 g of sample were immediately transferred to a 20 mL headspace vial, and then 5 mL of saturated sodium chloride solution was added. The samples were pre-incubated for 5 min at 50 °C, then the fibre was exposed for 30 min at the same temperature to adsorb VOCs. The VOCs isolated by SPME were desorbed in the injector port on the GC × GC-ToFMS system (Pegasus 4D LECO, St. Joseph, MI, USA). The GC was equipped with a DB-5 column (25 m × 0.2 mm × 0.33 µm, Agilent Technologies, Santa Clara, CA, USA) and a Supelcowax 10 (1.2 m × 0.1 mm × 0.1 µm, Supelco Bellefonte, PA, USA) as the second column. The injector temperature was set at 250 °C and the injection was performed in a splitless mode.
The gas used in the analysis was helium and its flow rate was set at 0.8 mL/min. The primary oven temperature was programmed as follows: 40 °C for 1 min, then rising at 6 °C/ min to 200 °C, then it was increased at 25 °C/min to 235 °C, where it was held for 5 min. In the secondary oven, the following program was used: 65 °C for 1 min, then rising at 6 °C/min to 225 °C, followed by a decrease at 25 °C/min to 260 °C, where it was held for 5 min. The transfer line temperature was 260 °C. The ToF mass spectrometer was operating at a mass range of m/z 38-388 and detector voltage − 1700 V at 150 spectra/s. The modulation time was 4 s.
Data were collected using the LECO ChromaTOF v.4.44 software (St. Joseph, MI, USA). The identification was accomplished using the National Institute of Standards and Technology (NIST) library (version 2.0) of mass spectra obtained from highly sensitive ToFMS. For the analyte match criteria, the following parameters were used: minimum similarity = 70%, mass threshold = 10, and signal to noise ratio = 1000. Compounds detected in the extract were identified tentatively. Additionally, retention indexes were calculated and compared with respective literature data.

Statistical analysis
All analytical measurements were performed in three biological replicates. The content of individual compounds was compared between the processing types using a one-way analysis of variance (ANOVA). The significance of differences between the samples was determined by Fisher 's LSD test at a p-value < 0.05. All statistical analyses were performed using STATISTICA version 13.3 (Statsoft, Tulsa, OK, USA) and GraphPad Prism version 8.0.0 for Windows (San Diego, CA, USA) software. To identify overall differences between the groups, principal component analysis (PCA) was performed using the XLSTAT software package (version 2022.1.New York, United States).

Amino acids concentration
In Brussels sprouts, the presence of 19 free amino acids (FAA) was determined (Table S1). The total FAA concentration was similar in raw and steamed Brussels sprouts (1722.95 ± 34.60 vs. 1714.09 ± 22.73 mg/100 g FW), while the total FAA in boiled vegetables was significantly lower (1520. 16 ± 39.19). The main FAAs in Brussels sprouts were glutamine, hydroxyproline and proline. According to our knowledge, this is the first time that FAA composition is reported in Brussels sprouts. The content of amino acids was previously described in other members of the Brassica family, such as cauliflower [13] and cabbage [14]. Similarly to our study, glutamine was previously reported as the main FAA in all aerial parts of cauliflower [13]. The content of dominant FAA was significantly lower in boiled Brussels sprouts compared to other treatments. Similarly, a significant reduction in levels of glutamic acid, valine, methionine and isoleucine was observed irrespective of the type of processing. The largest changes were observed for methionine, which concentration decreased approx. five-fold in Brussels sprouts after thermal processing. In contrast, contents of glycine, cysteine, phenylalanine and proline were not affected by thermal processing. The content of histidine was the highest in steamed Brussels sprouts, while leucine and lysine were most abundant in boiled vegetables. Interestingly, only the content of histidine and alanine was higher in thermally processed vegetable compared to the raw one.
The observed decrease in the amino acid level in the boiled vegetable could be accounted for as leaching into the cooking water. The leaching may have been facilitated by heat, which was found to increase solubility of nutrients in the processing water [15]. It has been reported that the content of FAA decreased with the prolonged cooking time with an almost equal increase in its content in the cooking water [16].
Our results are in agreement with the study comparing the effect of different cooking methods on the FAA content in the bamboo shoot [17]. Similarly, those authors reported that boiling resulted in a significant loss of FAAs, while steaming had no significant effect on FAAs. In another study, in cabbage cooked for only 2 min, the content of FAAs decreased by 30% [18]. In our study, the loss was much smaller (approx. 12%), which can be explained by the more compact structure of the Brussels sprouts, limiting the passage of nutrients from the inner parts into the cooking water.

Sugar concentrations
Sweetness is considered the most important factor influencing flavor perception [19]. Therefore, sugars, as substances responsible for the formation of sweetness, are important to be evaluated if quality changes after processing are considered. The major sugar detected in raw Brussels sprouts in this study was glucose (Table S2).
This finding is the opposite of the previous report [12], where sucrose was the most abundant one in Brassica vegetables. However, in that study, broccoli, cauliflower and kohlrabi were analyzed. The main difference between these two studies, except for the variety of the plant, was the harvesting season. In the present study, Brussels sprouts harvested in the summer season were analyzed, while in the previous one, vegetables collected during wintertime were examined. Another important observation from the above-mentioned study was the significant decrease in carbohydrate concentration after thermal processing of Brassica plants, such as broccoli, cauliflower or kohlrabi [12]. In contrast, in Brussels sprouts, these changes are relatively small, which might again result from the compact structure of this plant and limited leaching of sugars into the water during processing. In the presented study, processing, such as boiling and steaming, caused the decrease of glucose, fructose and to some extent stachyose, while sucrose was the most abundant. It was observed that changes caused by boiling were greater than by steaming, similar to FAAs, which can be explained by the fact that the exposure to the polar solvent (water) is greater in the traditional cooking method. The impact of thermal processing correlates with previously published data and indicates small changes after thermal processing, such as boiling or steaming. The major differences were observed for sucrose, which concentration was higher after thermal processing, and glucose, which level was higher in the raw plant. It might be related to the activity of invertases, which might be responsible for the degradation of sucrose into glucose and fructose [20].

Glucosinolate concentrations
The content of GSL is presented in Table 1. Among the ten GSL detected in Brussels sprouts, sinigrin was the dominant, followed by glucoiberin, glucobrassicin and progoitrin. It is in agreement with previous reports on Brussels sprouts [7,12]. Notably, the concentration of the majority of GSL was higher in processed vegetables when compared to raw. This phenomenon was explained by Ciska et al. [7], who hypothesized that the increment in the GSL level results from the degradation of tissue after heat treatment. It is because GSL might be partially bonded to the cell walls and be released only after tissue structure disintegration. As a result, the extraction performance from processed vegetables was more effective than from raw ones. Importantly, it suggests that the increase in the GSL content results from the underestimation, not from de novo formation. This theory seems to apply in this research as well, as it was observed that the concentration in boiled and steamed plants was higher than in raw vegetables. Comparing the GSL in terms of their side chain, the content of a majority of aliphatic GSL was much greater in thermally processed vegetables, while for indole GSL these changes were minimal. On the other hand, despite the relatively short processing time, the largest loss in concentration was observed for 4-hydroxyglucobrassicin, which content was significantly reduced in boiled vegetables and not detected in steamed ones. It is in agreement with previous studies showing that 4-hydroxyglucobrassicin is the most thermally labile GSL [7,21].
Several studies demonstrated that steaming is the most efficient treatment to retain GSL in Brassica vegetables compared to other thermal methods, such as blanching or boiling [22]. It is mainly related to the leaching of GSL into the water, which is limited in steaming. However, in our study, there was no significant differences in total GSL levels between steaming and boiling. Again, it can be explained by the more compact structure of the Brussels sprouts, which prevents leaching into the cooking water, as well as the lower temperature inside the sprout, which limits thermal degradation of GSL. However, it does not mean that the thermal degradation of GSL does not take place during thermal treatment. In the aforementioned study by Ciska et al. [7], in which while the content of GSL increased after a short cooking time, the formation of GSL breakdown products was observed from the beginning of processing. Maybe the prolongation of cooking time would cause loosening of the plant tissue and a more drastic reduction of GSL in Brussels sprouts, similar to those observed by Ciska et al. [7]. The lower content of GSL in the raw Brussels sprouts can also be related to the protocol of extraction. In our study, vegetables were frozen with liquid nitrogen and homogenized, which resulted in the disruption of plant tissues and releasing of myrosinase. In turn, during thermal processing, the activity of the enzyme could be reduced due to the elevated temperature, whereas in the raw vegetable, the enzyme was still active and could cause partial degradation of GSL.

Volatilome of Brussels sprouts
In total, 84 VOCs were detected in processed Brussels sprouts. According to their chemical class, 19 aldehydes, 6 alcohols, 14 ITC, 5 furans, 6 ketones, 15 nitriles, 3 pyrazines, 10 sulfur-containing compounds (other than ITC) and 6 miscellaneous compounds were detected (Table S1). The greatest abundance of VOCs was detected in raw Brussels sprouts, mainly due to the presence of aldehydes, ITC and nitriles (Fig. 1). In all analyzed vegetables, GSL breakdown products accounted for more than half abundant VOCs.

GSL hydrolysis products
ITC are compounds of special interest, not only because of their bioactive properties (Palliyaguru et al., 2018), but also due to their specific aroma [23], which might affect the flavor of Brassica vegetables. The summarized abundance of ITC was the highest in the raw Brussels sprouts mainly due to the presence of allyl ITC-the compound derived from the degradation of sinigrin, dominant in Brussels sprouts (Fig. 1). Similarly, cyclopropyl ITC, 4-methylpentyl ITC and isopropyl ITC were the most abundant in raw compared to processed vegetables ( Fig. 2A). A relatively high abundance of ITC was noted also for boiled Brussels sprouts, with the dominant allyl ITC and 3-butenyl ITC derived from sinigrin and gluconapin, respectively. Notably, the presence of most minor ITCs was the highest in the boiled vegetable. Similarly, the previous study showed that the formation of ITC is promoted after boiling of Brussels sprouts [24]. In steamed Brussels sprouts, the presence of only eight ITC  Fig. 1 Contents of volatiles (expressed as peak areas), collected according to the chemical group in raw, boiled and steamed Brussels sprouts. Based on data presented in Table S3 was detected with a much lower abundance compared to the other cooking treatments. Notably, the low amount of allyl ITC in the steamed Brussels sprouts correlates with the high concentration of sinigrin, which was not hydrolyzed.
In contrast, in the study of Ciska et al. [7], no ITC was detected either in Brussels sprouts or in water after boiling. It can be explained by the different extraction methods used in that study (solid-liquid extraction with methylene chloride vs. SPME in our study). Moreover, in our study, a different protocol was applied with homogenization with liquid nitrogen directly after cooking, which could prevent the evaporation of highly volatile ITC. In general, in raw vegetables, the plant tissue should be intact and thus GSL and myrosinase separated. A very high abundance of ITC in the raw plant in our study could be explained by enzymatic hydrolysis of GSL by myrosinase during homogenization. The reduction of ITC in boiled Brussels sprouts can suggest a loss of these compounds due to the leaching of GSL and ITC into the cooking water, similar to other studies [7,25], rather than by the thermal degradation, which can be expected only after minimum 10 min of boiling [25]. The very low abundance of ITC in steamed Brussels sprouts is very difficult to explain. One possible explanation may be the limited degradation of GSL during steaming, which can be assumed based on the highest content of intact GSL. Similarly, Song and Thornalley [26] reported that steaming did not cause a significant reduction in GSL. On the other hand, nitrile compounds were much more abundant in Brussels sprouts after steaming, being the dominant chemical class of VOCs in the vegetable processed this way (46% of the total peak area) (Fig. 1). Another possible explanation for this phenomenon is the amount of water in the medium, essential for the enzymatic reaction. Since the enzymatic reaction favors ITC formation, the limited water availability during steaming could shift the reaction into nitrile formation. Comparing the total peak areas, more nitriles were detected in raw Brussels sprouts than in steamed ones (2.35 vs 1.99 × 10 8 ); however, in raw vegetables, nitriles were not the dominant class (only 18% of the total peak area), and there was the biggest contribution of ITC. The predominant nitrile in steamed vegetables was 3-butenenitrile (Fig. 2B), derived from the dominant sinigrin, which level was almost two times higher than in raw and boiled vegetables. At the same time, the ITC degradation product of sinigrin was detected in a very low abundance. It suggests that the direction of GSL hydrolysis in steamed Brussels sprouts was shifted into nitriles. Considering that nitriles are suggested to be toxic [27], it raised a question of whether steaming, which is often described as the most preferable cooking method, is in fact the best way to prepare Brussels sprouts. Previous studies also suggested that thermal processing of vegetables shifts hydrolysis of GSL into nitrile formation [7,25], which is consistent with our study only for steaming. In boiled Brussels sprouts, the abundance of nitriles was the lowest among all the analyzed treatments, which can suggest that it is the best form for Brussels sprouts to be consumed,

Fig. 2
Heatmaps presenting the abundance of isothiocyanates (A) and nitriles (B), expressed as peak areas, from raw, boiled and cooked Brussels sprouts. Created based on data presented in Table S3 taking into account the compromise between the beneficial properties of ITC and the toxic potential of nitriles. The lowest content of nitriles in boiled Brussels sprouts was reported previously [28]. Interestingly, the abundance of degradation products of glucoiberin, the second most dominant GSL in Brussels sprouts, namely 3-(methylthio)propyl ITC and 4-(methylthio)butylnitrile, were small, compared to those of other ITC and nitriles. It can suggest that glucoiberin was more thermally stable than the other GLS. Notably, while the abundance of nitrile was similar in thermally processed vegetables, the content of ITC was much higher in boiled Brussels sprouts, with no presence in steamed vegetables, which again can confirm the different direction of GSL hydrolysis dependent on the type of treatment. Nevertheless, it should be also highlighted that the logP value of 4-(methylthio) butylnitrile is low (-0.4), which indicates its strong polar character. Since it is still detectable by the SPME technique performed in the headspace mode, which is less sensitive for polar compounds [29], it might mean that the concentration of this compound was very high.
Indole GSL, mainly glucobrassicin, were also detected in relatively high contents in analyzed Brussels sprouts; however, their breakdown products were not detected in this study. It is related mainly to the analytical approach used here. Indole derivatives, including indole-3-acetonitrile, indole-3-carbinol, ascorbigen and 3,3'-diindolylmethane, are compounds exhibiting relatively low volatility; therefore, they can be analyzed using targeted HPLC analysis. In this study, the main focus was put on the volatilome, therefore, these compounds were not specifically analyzed.
An important observation in this study was that it was not possible to match all the ITC and nitriles with their corresponding precursors. A similar problem was reported previously [30] with several possible explanations for this phenomenon. We agree with the authors that the most likely explanation is the existence of unknown mechanisms of post-hydrolysis modifications of GSL derivatives. Similarly, in a study by Marcinkowska et al. [23], more ITC and nitriles were identified in kohlrabi than GSL reported to date in this vegetable, and the origin of some of them is difficult to explain, likewise in turnip, ten various ITC and nitriles were detected. To date, some studies already showed possible modifications of ITC, such as tautomeric rearrangements [11]; however, the understanding of the formation of ITC and nitriles, which do not correspond to known GLS, requires further elucidation (Fig. 3).

Other VOCs
The second most abundant chemical class in raw Brussels sprouts were aldehydes (Fig. 1). Among them, (E)-2-hexenal was the dominant one, followed by hexanal and (Z)-3-hexenal (Table 1). Thermal processing resulted in a dramatic reduction in aldehydes. The level of only pentenal increased after thermal treatment, irrespective of its type. Interestingly, the abundance of some aldehydes increased in boiled vegetables, including (Z)-2-heptenal, (E)-2-octenal, benzaldehyde, 2-methylbutanal, nonanal and octanal. The highest abundance of alcohols was noted in raw Brussels sprouts, almost two-fold higher than in thermally processed ones (Table S3). The main alcohols in the raw plant were 1-penten-3-ol and (Z)-2-penten-1-ol, which levels decreased significantly after thermal processing. On the other hand, cooking resulted in the formation of 1-octanol, 1-octen-3-ol, and to the highest extent, 1-pentanol. In a previous study with broccoli, also alcohols were found to be the most abundant in raw vegetables, which was associated with their formation in the lipoxygenase (LOX) pathway from the reduction of aldehydes catalyzed by alcohol dehydrogenase [3]. Similarly, in that study in the vegetable treated at low temperature, the number of alcohols was reduced, which was explained by the inactivation of alcohol dehydrogenase. It can be assumed that due to the thermal treatment the activity of this enzyme was reduced and contributed to the lack of transformation of aldehydes into alcohols also in our study. The majority of C6 and C9 aldehydes, alcohols and esters, called "green leaf volatiles", are products of oxidation of fatty acids in the LOX pathway (Fig. 3), which are formed during the disruption of the plant tissue [31]. This pathway can be suggested as the main source of aldehydes and alcohols in the raw plant, where the disruption of plant tissue could occur during freezing with liquid nitrogen. After thermal processing, whether it was boiling or steaming, the inactivity of enzymes could be expected. Knowing that in the LOX pathway four different enzymes are involved (lipoxygenase, hydroperoxide lyase, 3Z,2E-enal isomerase and alcohol dehydrogenase [32], their limited activity in cooked vegetables could explain the lower abundance of aldehydes and alcohols. It can be confirmed also by changes observed in ITC intensities and the possible inactivation of myrosinase. The level of ketones was the highest in the boiled Brussels sprouts being over fivefold higher compared to that in raw and steamed vegetables (Table S3). The summarized peak area differed between the studied processing types. The lowest abundance of ketones was observed in raw Brussels sprouts. The main ketone in raw vegetables was 2,3-octanedione, which was significantly reduced after thermal processing. On the other hand, after steaming, the level of ketones was almost six times higher than in raw vegetables with the dominant ketone 3-methyl-2-pentanone, which was detected only in this sample. Methyl ketones are usually formed as products of α-and β-oxidation of lipids [32]. In boiled Brussels sprouts, the abundance of ketones was drastically higher compared to other treatments. Compared to raw and steamed vegetables, in boiled Brussels sprout, the levels of ketones were almost 34 and 6 times higher, respectively. The high abundance of ketones in boiled vegetables was caused by the presence of the main ketone, 6-methyl-5-hepten-2-one, which accounted for 98% of all ketone abundance. One of the possible origins of 6-methyl-5-hepten-2-one is the degradation of lycopene [33]. Moreover, only in boiled vegetables, the presence of 1-octen-3-one was detected, which is considered one of the possible products of oxidation of linoleic acid [34].
Another important group of VOCs in Brassica vegetables, contributing to the typical cabbage-like aroma, are sulfurcontaining compounds. In raw and boiled Brussels sprouts, dimethyl trisulfide was the most abundant. However, in steamed vegetables, the dominant compound was dimethyl sulfide. Derivatives of methionine, such as S-methylmethionine, release dimethyl sulfide, which is responsible for the aromas of fish, canned sweet corn, tomato juice, and others [35], being responsible for aromas described as asparagus-like, cheese-like, moldy, sulfur and cabbage-like [36]. Furthermore, the level of dimethyl disulfide, a secondary product of primary C-S lyase action on S-methyl-L-cysteine sulfoxide [37], also increased after thermal treatment (Fig. 3). The greater abundance of sulfur-containing compounds in thermally treated Brussels sprouts can be associated with the metabolism of methionine, which content was significantly reduced after thermal treatment, irrespective of the treatment type. Importantly, sulfides have very low odor thresholds, thus even a small amount of these compounds can significantly affect sensory perception [38]. Our results are in agreement with the previous study, which showed that in Brussels sprouts, the level of sulfides increased after cooking [24].
The level of furans was similar in the analyzed forms of Brussels sprouts. Furan can be formed via various pathways, including thermal degradation of carbohydrates alone, or in presence of amino acids, thermal degradation of amino acids, oxidation of ascorbic acid at high temperature and oxidation of polyunsaturated fatty acids and carotenoids [39]. Two furan derivatives were formed after cooking in our study, which was 2-propylfuran and 3-methylfuran in boiled and steamed Brussels sprouts, respectively. However, generally, changes in furan levels were not that prominent in processed Brussels sprouts.
To evaluate the similarity between the types of treatments based on VOCs, PCA analysis as an unsupervised clustering method was performed. Two first principal components (PC) represented 85.27% of the total variance. Based on the PCA plot (Fig. 4), the raw and boiled vegetables were placed in the same quadrant revealing their high similarity. In contrast, the profile of VOCs in steamed Brussels sprouts formed a separate cluster in the second quadrant. It could also be concluded that most of the analyzed VOCs were concentrated near the origin of coordinates and that they did not exhibit any specificity. However, some of the compounds could be clearly associated with the type of treatment. (E)-2-Hexenal and cyclopropane ITC were the main dominating VOCs present in raw Brussels sprouts. Dimethyl trisulfide and 3-butenyl ITC were compounds closely associated with boiled vegetables. The steamed Brussels sprouts, which formed a separate group on the PCA plot, were rich in nitriles, dimethyl sulfide, pentanal, 1-pentanol and 2-ethylfuran (Fig. 4.).

Conclusions
In summary, the present study showed a successful application of metabolomic fingerprinting in food science. The results of this study showed that thermal processing has an important influence on the volatilome of Brussels sprouts. Although the contents of primary metabolites were similar irrespective of the processing method, the contents of GSL and their breakdown products were clearly affected. Interestingly, the content of GLS was higher in processed vegetables, mainly due to the loosening of the plant tissue and better extractability. In the raw Brussels sprouts the richest profile of VOCs was determined, with aldehydes and ITC being the dominant groups. In contrast, in steamed vegetables, the formation of nitriles was favored, which accounted for around 50% of all VOCs detected. In boiled Brussels sprouts, the compromise between the high level of ITC and the lowest abundance of nitriles was achieved, which suggests that this form of consumed Brussels sprouts is the most beneficial for health. In this report the presence of ITC, which do not correspond to GLS detected in Brussels sprouts was recorded, which can suggest other, unknown rearrangements and/or degradation of GLS breakdown products, requiring further in-depth elucidation. In summary, the result of this study can help to identify the most beneficial method of processing not only Brussels sprouts, but also other cruciferous vegetables as well as enhance our understanding of the formation of VOCs, of which many are responsible for aroma properties of Brussels sprouts.  Table S3