2.1. The influence of freezing and thawing on the profile of volatile compounds
The results of the GC×GC-ToFMS analysis of fresh broccoli and the same vegetable after freezing and thawing are shown in Fig. 1. The comparison of profiles of volatile compounds obtained from fresh broccoli with that of frozen and later thawed broccoli showed significant differences. In freshly cut florets, 3-hexen-1-ol, 1-hexanol dominated and in the sample analysed after freezing and thawing, the most abundant compounds were: hexanal, 2-hexanal, 2,4-heptadienal, 1-penten-3-one, 1-pentanol and also 3-hexen-1-ol (however in smaller amount than in the fresh sample) and furan 2-ethyl (Table S1. Supplementary material). The meaningful difference was connected with the LOX pathway metabolites (Fig. 1A). There was a change in the proportion between aldehydes and alcohols in the volatile fraction. (Fig. 1A.) The experiment showed that in fresh vegetable, the main products were alcohols, which were metabolites in the LOX pathway, resulting from aldehydes reduction catalysed by alcohol dehydrogenase. While in frozen/thawed samples, the main products from the LOX pathway were aldehydes: hexanal, 3-hexenal, 2-hexenal among others (Tab. S1) which suggest that alcohol dehydrogenase is effectively stopped by low temperature and it is not catalysing reduction of aldehydes after thawing. According to applied statistics – a t-test was performed and P = 0.00005 was observed for alcohol concentration, no meaningful probability (P = 0.18) was noted for aldehydes which could be caused by a high standard deviation value (Fig. 1, Tab.S4). The profile of compounds was the same regardless of the use of liquid nitrogen or freezing at -20oC.
The difference between samples was significant also in case of glucosinolates – the myrosinase system products, as presented in Fig. 1B. In fresh vegetables, the isothiocyanates were the main products resulting from glucosinolates hydrolysis, while in frozen vegetables, the number of nitriles - was growing rapidly, compared to fresh tissue. The differences between isothiocyanates and nitriles in fresh and freeze-thawed vegetable was significant – in both cases P < 0.05 (isothiocyanates: P = 0.02; nitriles: P = 0.003, according to t-test results) (Fig. 1. Tab.S4)
Our results indicate that there are broader consequences of freezing-thawing on broccoli volatilome. Fig. S1 (supplementary material), together with supplementary table S1, shows a more detailed insight into the profile of volatile compounds extracted from fresh and frozen-thawed broccoli florets. Some characteristic compounds for fresh and frozen-thawed broccoli tissues can be elucidated from the graph regarding peak areas (intensities) of these metabolites. For fresh broccoli there are (Z)-3-hexene-1-ol (168), 1-hexenol (42), 3-ethyl-1,5-octadienone (164), thiocyanic acid methyl ester (478), 1-butene, 4-isotiocyanate (36), 3-hexenol-1-acetate (170). For frozen and thawed florets, the most abundant were 2-ethylfuran (354), 1-pentene-3-one (65), 1-pentene-3-ol (64). The detailed list of the compounds differentiating both samples based on GC×GC-ToFMS results was listed in supplemental materials as Table 1S. 489 compounds are listed alphabetically and colour-coded, where the heat scale ranges from deep blue (0 intensity) to red (highest intensities). From comprehensive GC×GC-ToFMS data initially, 768 features were compared for frozen-thawed and fresh broccoli. First compounds, including siloxanes, TMS derivatives, perfluorinated derivatives, were discarded on the assumption that they were column artifacts. Then some of the most abundant and tailing peaks were combined and unidentified (89) analytes were removed.
2.2. The confocal microscopy scanning results
In order to determine the reasons for changes caused by freezing in liquid nitrogen and storage at -20oC, at first the confocal microscopy was applied. It enabled evaluating the state of cells in broccoli tissue. As presented in Fig. 2 and Fig. 3, the use of liquid nitrogen freezing did not cause significant changes in the cell structure, while in slow freezing (-20oC) those changes were visible. As presented in Fig. 3, geometrical parameters in both fast and slow freezing samples were similar to the control sample – fresh untreated broccoli (statistical data were presented in supplementary materials in Fig. S2, S3, Tab. S2, Tab. S3). Based on obtained results, it is difficult to support the thesis concerning changes in profile of volatiles due to tissue deterioration.
2.3. The pH measurement
The pH value was measured in fresh broccoli, broccoli comminuted in liquid nitrogen and frozen at -20oC. The pH value did not change significantly after freezing. Differences between samples were minimal. In fresh broccoli the pH value was 6.6 ± 0.12, while in case of the sample comminuted in liquid nitrogen it was 6,65 ± 0.17 and for the one frozen at -20oC − 6.56 ± 0.09. As presented, the changes in pH value after freezing did not occur, thus this parameter did not influence the enzymatic activity.
2.4. Enzyme activity assays
The changes in volatile composition after freezing in liquid nitrogen or at -20oC could be caused by the activity of selected enzymes. In the case of broccoli, those changes were mainly related to nitrile and aldehyde concentrations. Therefore, special attention was placed on myrosinase and lipoxygenase. To check the myrosinase activity in broccoli after freezing, two trials were compared: fresh broccoli and broccoli after freezing in liquid nitrogen. The sinigrin solution was added to the enzyme extract from the vegetable, because broccoli does not contain this glucosinolate – the detailed scheme is presented in supplementary materials, namely Fig. S4. That is why broccoli could be a model matrix in which the hydrolysis of sinigrin (added to it) might be observed. The amount of glucosinolate hydrolysis products formed was measured using SPME. The zero tests were extracts without the addition of sinigrin, in which no allyl isothiocyanate or alternative breakdown products of 3-butenonitrile sinigrin or the corresponding epitionitrile were noted.
A significant amount of allyl isothiocyanate was found in the test with the addition of sinigrin, to the extract from fresh broccoli (Fig. 4A). In the sample after freezing in liquid nitrogen, the amount of this isothiocyanate was as high as in the non-frozen sample. In the next stage, in order to separate myrosinase from epitiospecific and nitrile-specifier proteins whose mass does not exceed 50 kDa [13]. AMICON filters were used to separate 100 kDa proteins enabling separation of myrosinase (whose mass is about 200 kDa) from epithiospecific and nitrile-specifier proteins. A significant amount of allyl isothiocyanate was formed after the addition of sinigrin to the upper layer containing myrosinase (Fig. 4B), the larger peak on chromatogram was probably caused by a higher myrosinase concentration after filtration. On the other hand, after the addition of sinigrin to the lower layer in which the nitrile-specifier proteins were present, no degradation products of glucosinolate were observed (Fig. 4C).
2.4. The influence of cutting on the profile of volatile compounds
Samples of broccoli were prepared in the following three ways prior to sampling: cutting, blending and shredding. The method of broccoli tissues fragmentation had a substantial influence on the profile of the volatile compounds extracted using the SPME technique (Fig. 5). After comparing the abundances of compounds in tested main groups of volatiles, it was evident that tissue cutting influenced the overall profile of volatiles and their total amount. Though alcohols were the prevailing group in all methods, the number of aldehydes and the ratio of aldehydes to alcohols varied from 0.04 in cutting to 1.44 in blending (P = 0.000095, ANOVA, Tukey’s test). Shredding resulted in the ratio of aldehydes to alcohols of 0.07 (shredding-blending P = 0.0006; shredding-cutting P = 0.06, ANOVA, Tukey’s test). The number of released sulphides was the lowest for cutting and highest for shredding, however, no significant difference was noted according to the ANOVA test (P > 0.05). The highest differences in amounts of extracted compounds were noted for isothiocyanates. Their amount obtained after blending was 31 times higher than in case of cutting (P = 0.00081, ANOVA, Tukey’s test) and almost 3 times higher than after shredding (P = 0.005, ANOVA, Tukey’s test), however, the number of the identified components remained almost the same in all analysed samples. Based on the ANOVA analysis and post-hoc Turkey test, the most important differences were noted in alcohol, aldehyde and isothiocyanate concentrations, and no statistically relevant difference for sulphide concentration was noted (see Tab S5. and S6. Supplementary materials).
2.5. The influence of enzyme reaction quenching on the profile of volatile compounds
As presented in Fig. 6. and Tab. S7, S8 (supplementary materials), different inhibition techniques gave variable effects. Main groups of volatiles were selected to visualise differences between treatments. EDTA led to an increase in aldehyde concentration and at the same time a visible reduction in alcohol amount (comparing to cooked sample and control sample, P values were below 0.0001, and in case of sample treated by liquid nitrogen < 0.001, according to Tukey’s Multiple Comparison Test – see Tab S8). There was also a significant growth of 2-ethyl furan concentration in samples treated with liquid nitrogen and EDTA compared with fresh vegetable (P < 0.00001 in both cases, Tukey’s test). In samples treated with liquid nitrogen, sulphur and alcohol levels decreased in comparison to the fresh tissue (P < 0.0001, post-hoc Tukey’s test), however, the concentration of aldehydes was higher than in the untreated sample (P < 0.0001 post-hoc Tukey’s test). However, it is important to realise that the inhibition agent EDTA was introduced to the sample with water, which may also result in a lowered partition coefficient of some analytes. The important observation is a low standard deviation in samples treated with EDTA which is particularly relevant in case of working with fresh plant tissue. Liquid nitrogen treated profile of volatiles compared to fresh tissues confirm our earlier results.
2.6. Aspects related to SPME extraction
SPME is a predominant extraction method used for the isolation of flavour compounds and profiling volatiles. The initial step in the elaboration of the SPME extraction parameters was fibre coating selection, as this parameter significantly influences the extraction efficiency. Four most popular fibres (DVB/PDMS, CAR/PDMS, PDMS and CAR/DVB/PDMS were tested for isothiocyanate extraction (allyl isothiocyanate, benzyl isothiocyanate, and isobutyl isothiocyanate) and both Carboxene - containing fibres showed the highest extraction efficiency (Fig. S5A. – supplementary materials). Then the same fibres were used to extract volatiles from blended broccoli florets to evaluate their usefulness for extraction of other important groups of broccoli volatiles. The results are shown in Fig. S5B (supplementary materials). Four groups of metabolites were taken into consideration in experiments involving broccoli florets – alcohols, aldehydes, sulphides, and isothiocyanates. For esters, aldehydes, and alcohols the peak areas of detected compounds were the highest for the CAR/DVB/PDMS fibre. Only the peak area of sulphides was higher for the CAR/PDMS fibre. Based on this finding, the CAR/DVB/PDMS fibre was selected for subsequent experiments.
To elaborate the optimal temperature for SPME extraction, blended broccoli samples were extracted at 20oC, 40oC, 60oC and 80oC for 30 minutes (Fig. S5C, supplementary materials). The alcohol concentrations increased with increasing temperature from 20oC to 60oC. The peak area increases for aldehydes paralleled that for alcohols, and for sulphides the increment related to temperature was much less prominent than for the remaining two groups of metabolites. It suggests that sampling at 20oC could be sufficient for the extraction of compounds belonging to these groups which would result in minimising thermal changes in the matrix. However, the only group of metabolites, which required a higher extraction temperature was isothiocyanates; they were detected only at 60oC and 80oC. The peak area of isothiocyanates increased by 20% when the extraction temperature was changed from 60oC to 80oC. The increase of extraction temperature can increase the migration of isothiocyanates to the headspace, however, it can also promote the formation of isothiocyanates by thermal decomposition of glucosinolates. The increasing concentration of isothiocyanates with increasing temperature was not likely to be related to a higher myrosinase activity because higher temperatures, such as 80oC, are too high for the enzymatic reaction. This reaction occurred before the extraction process; thus, the high temperature apparently increased the volatility of compounds occurring after the enzymatic process. The isothiocyanates considered in this experiment were isobutyl-, n-pentyl- and hexyl isothiocyanate). The highest total peak area was achieved after extraction at 60oC; therefore, this temperature was applied in the following trials.
The next examined parameter was the extraction time. Different extraction times from 5 minutes to 30 minutes were compared. The highest peak areas of all compounds of interest were achieved after 10 minutes of fibre exposition as presented in Fig. S5D (supplementary materials). Moreover, the standard deviation was also at the lowest level after this short extraction time. This was probably caused by the high volatility of compounds and their rapid transfer from the matrix to the gas phase. It showed the importance of optimising this parameter. An additional 10 minutes of extraction also significantly reduced the extraction time, which made it possible to reduce the total analysis time.