Modulation of Glucosinolate Composition in Brassicaceae Seeds by Germination and Fungal Elicitation.

Glucosinolates (GSLs) are of interest for potential antimicrobial activity of their degradation products and exclusive presence in Brassicaceae. Compositional changes of aliphatic, benzenic, and indolic GSLs of Sinapis alba, Brassica napus, and B. juncea seeds by germination and fungal elicitation were studied. Rhizopus oryzae (nonpathogenic), Fusarium graminearum (nonpathogenic), and F. oxysporum (pathogenic) were employed. Thirty-one GSLs were detected by reversed-phase ultrahigh-performance liquid chromatography photodiode array with in-line electrospray ionization mass spectrometry (RP-UHPLC-PDA-ESI-MSn). Aromatic-acylated derivatives of 3-butenyl GSL, p-hydroxybenzyl GSL, and indol-3-ylmethyl GSL were for the first time tentatively annotated and confirmed to be not artifacts. For S. alba, germination, Rhizopus elicitation, and F. graminearum elicitation increased total GSL content, mainly consisting of p-hydroxybenzyl GSL, by 2-3 fold. For B. napus and B. juncea, total GSL content was unaffected by germination or elicitation. In all treatments, aliphatic GSL content was decreased (≥50%) in B. napus and remained unchanged in B. juncea. Indolic GSLs were induced in all species by germination and nonpathogenic elicitation.

Introduction

Plants from the family of Brassicaceae are of global economic importance. They are consumed throughout the world in many forms, such as leafy vegetables, root vegetables, sprouts, vegetable oil, and condiments. Compounds from Brassicaceae gain more research interest because of their potential antimicrobial activity (both with respect to disease resistance of brassicaceous species and as natural preservatives to enhance the open shelf life of food products),[1−3] which is often associated with isothiocyanates (ITCs), a class of compounds that are degradation products of GSLs.[4,5]

The biosynthesis of GSLs consists of three stages (Figure A): (i) side chain elongation of the amino acids, especially Met, Val/Leu, Ile, and Phe, (ii) formation of the GSL core, and (iii) secondary modification of the side chain, e.g., oxidation and hydroxylation.[6,7] In addition to step (iii), substitution at the thioglucosyl group, e.g., acylation, might occur, which has been identified by Linscheid et al.,[8] Reichelt et al.,[9] and Agerbirk and Olsen.[10] GSLs are classified into various classes based on their side chain: aliphatic GSLs (Figure B) derived mostly from Met, benzenic GSLs (Figure C) from Phe or Tyr, and indolic GSLs (Figure D) from Trp.[11]

Figure 1

Simplified biosynthesis route of GSL classes, i.e., aliphatic, benzenic, and indolic (A), subclasses of Met-derived aliphatic GSLs (B, red box), benzenic GSLs (C, yellow box), and indolic GSLs (D, blue box) in Brassicaceae, especially S. alba, B. napus, and B. juncea. The biosynthesis route was adapted from Bianco et al., Lee et al., Agerbirk and Olsen, Abdel-Farid et al., Wittstock and Halkier, and Yi et al.[11−16] The biosynthesis route of aliphatic GSLs starts from pyruvate (Pyr) and continues to the formation of methionine (Met) for the linear ones or of valine (Val)/leucine (Leu) and isoleucine (Ile) for the branched ones (i.e., branched alkyl GSLs, which have been found in the tribe Cardamineae).[17] As it is still questioned if alanine (Ala) is a precursor of methyl GSL and linear alkyl GSLs,[11,17] Ala is omitted in (A). The biosynthesis route of aromatic GSLs starts from Pyr to the formation of chorismate and continues to the formation of phenylalanine (Phe) or tyrosine (Tyr), or tryptophan (Trp), leading to the formation of benzenic GSLs or indolic GSLs. Dashed arrow in (C) indicates putative biosynthesis step. The GSL core structure given in (B–D) is omitted, except for the thioglucose-acylated GSLs, and only the side chain (−R) is shown. Boldface numbers after the names of GSLs correspond to the number in Table S3.

The content of GSLs in the growing plants can be influenced by applying abiotic and biotic stressors.[20−24] In regards to biotic stressors, fungi are commonly applied on mature Brassicaceae plant tissues, e.g., leaves. Studies have indicated various effects of pathogenic fungi, e.g., Fusarium oxysporum and Alternaria brassicae, on the content of aliphatic, benzenic, and indolic GSLs in infected leaves of various Brassicaceae species and varieties.[4,12,20] In regards to abiotic stressors, phytohormones and salts are often applied on germinating Brassicaceae seeds.[24−29] Most of the studies indicated that total GSL content could be increased, but none reported an improvement in GSL diversity. Research on fungal elicitation of germinating Brassicaceae seeds has been done once in 1996.[30] That study reported the effect of Plasmodiophora brassicae on GSL content per class in roots of four varieties of Chinese cabbage (Brassica campestris ssp. pekinensis) and indicated no clear trend on the compositional changes of GSLs.[30] Studies on the effect of fungal elicitation on the compositional changes of particular major secondary metabolites have been performed on a different plant family, i.e., Leguminosae, and found that fungal elicitation induced the accumulation of isoflavonoids and stilbenoids, which had stronger antimicrobial activity.[31] Using a similar approach, i.e., fungal elicitation, we aimed for an induction of GSL production (in terms of content and diversity) in the germinating Brassicaceae seeds, which then can be used to yield high amounts of ITCs with potential antimicrobial activity.

The induction of GSLs seems to be dependent on the fungus employed and the plant species and variety. Three different fungi were used as stressors in this study, namely, Rhizopus oryzae, F. graminearum, and F. oxysporum, to modulate the composition of GSLs in various Brassicaceae seeds. R. oryzae is food-grade and nonphytopathogenic and often has been used for the elicitation of germinating legume seeds.[31−35]F. graminearum is not pathogenic to most Brassicaceae plants, whereas F. oxysporum is.[36] To the best of our knowledge, our study is the first to report the changes in GSL composition in seeds of Sinapis alba, B. napus, and B. juncea upon germination and fungal elicitation. The changes of GSL composition were monitored by RP-UHPLC-ESI-MSn analysis. It was hypothesized that the pathogenic fungus would induce GSL production extensively, compared to the nonpathogenic fungi.

Materials and Methods

Standard Compounds and Other Chemicals

Authentic standards of 12 different GSLs (with peak numbers in boldface according to Table S3)—benzyl GSL (14), phenethyl GSL (20), p-hydroxybenzyl GSL (5), 4-(methylthio)butyl GSL (15), 5-(methylthio)butyl GSL (21), 3-(methylsulfinyl)propyl GSL (1), 4-(methylsulfinyl)butyl GSL (3), allyl GSL (4), 3-butenyl GSL (8), 4-pentenyl GSL (13), (R)-2-hydroxy-3-butenyl GSL (2), and indol-3-ylmethyl (I3M) GSL (17)—were purchased from Phytolab GmbH & Co (Vestenbergsgreuth, Germany). Isopropanol, acetonitrile (ACN), MeOH, water with 0.1% (v/v) formic acid (FA) (ULC/MS grade), and ACN with 0.1% (v/v) FA (ULC-MS grade) were purchased from Biosolve BV (Valkenswaard, The Netherlands). Hydrogen peroxide (30% v/v) was purchased from Merck (Darmstadt, Germany), and commercial bleach solution (<5% v/v hypochlorite) was purchased from Van Dam Bodegraven B.V. (Bodegraven, The Netherlands). tert-Butanol (99.7%) was obtained from Sigma-Aldrich (St. Louis, MO, U.S.A.). Water used during experiments other than UHPLC-MS analysis was obtained with the use of a Milli-Q A10 Gradient system (18.2 MΩ·cm, 3 ppb TOC) (Merck Millipore, Darmstadt, Germany). Dimethyl sulfoxide (DMSO) was purchased from Ducheda Biochemie (The Netherlands).

Oatmeal agar (OA) was purchased from Becton, Dickinson and Company (New Jersey, U.S.A.). Malt extract agar (MEA) and agar technical were purchased from Oxoid Limited (Hampshire, U.K.). Peptone physiological salt solution (PPS) was ordered from Triticum Microbiologie (Eindhoven, The Netherlands).

Plant Materials

Seeds of Sinapis alba (yellow mustard “Emergo”, 393810), Brassica napus (“Helga”, 392600), and B. juncea var. rugosa rugosa (Chinese mustard/amsoi, 160400) were purchased from Vreeken’s Zaden (Dordrecht, The Netherlands; https://www.vreeken.nl/). B. juncea var. rugosa rugosa is mentioned as B. juncea in the following text.

Fungal Cultures

The fungal strains of Fusarium graminearum CBS 104.09 and F. oxysporum CBS 186.53 were purchased from CBS Fungal Biodiversity Centre (Utrecht, The Netherlands). Rhizopus oryzae LU581 was kindly provided by the Laboratory of Food Microbiology, Wageningen University (Wageningen, The Netherlands).

Surface Sterilization and Germination

For each germination experiment, seeds of all three species (15 g) were sterilized by immersion in a 100× diluted commercial bleach solution (500 ppm of NaOCl) for 15 min. After this, the seeds were rinsed 3 times with sterilized water and soaked for 8 h in excess of sterilized water in the absence of light.

Seeds were germinated at 25 °C in a modified sprouting machine (Sprouter microfarm EQMM; Easygreen, San Diego, CA, U.S.A.) in the absence of light. The machine was modified as described by Aisyah et al.[33] Prior to placing the seeds, the sprouting machine was cleaned according to the cleaning procedure from the manufacturer. Seeds were evenly distributed in one layer on the germination trays (17.8 × 8.9 cm). Before application of the stressor, sterilized water was applied by spraying every 3 h for a period of 15 min (∼17 mL/min). This resulted in a relative humidity (RH) of 90–100%. After application of the stressor, i.e., 48 h after start of the germination, RH was set at 55–85% by replacing spraying for fog distribution over the seedlings for 15 min per 3 h. This fog was generated by a minifogger (Conrad, Oldenzaal, The Netherlands).

Application of Stressor

Stressor, in the form of a fungal spore suspension, was applied to the seedlings 48 h after start of germination. Previously, R. oryzae was grown on a MEA plate for 7 days at 30 °C, whereas F. graminearum and F. oxysporum were grown separately on OA plates for 7 days at 25 °C.

The seedlings were inoculated with the fungal suspension, obtained by scraping off the plate fully covered by the mold with 9 mL of PPS. The suspension with an average count of 1.0 × 106 CFU/mL was evenly distributed over the 2-day-old seedlings (0.2 mL/g seedlings) and gently homogenized.

Seven-day-old seedlings (treated and nontreated) were harvested and freeze-dried. The experiment was repeated independently 3 times.

Sample Extraction

Lyophilized seeds and seedlings were milled into fine powder by using a high-speed rotor mill (Retsch Ultra Centrifugal Mill ZM 200; Haan, Germany) with a 0.5 mm sieve. The sample extraction was performed using a Speed Extractor (E-916; Buchi, Flawil, Switzerland). Ground material (400 mg) was mixed with sand (granulation 0.3–0.9 mm, dried at 750 °C; Buchi) in a 40 mL stainless steel extraction cell. Prior to extraction, samples were defatted using n-hexane. Then, the extraction was carried out with absolute MeOH at 65 °C. Because GSLs are completely soluble in absolute MeOH, extraction with absolute MeOH gives the advantage of rapid solvent evaporation and sample preparation. Results with absolute MeOH were comparable to those obtained with MeOH–H2O (7:3) (data not shown). The in-plant myrosinase was inactive under the extraction conditions used. The extraction was done in 3 cycles consuming 76 mL of solvent. Afterward, the extract was evaporated under reduced pressure (Syncore Polyvap, Buchi), resolubilized in tert-butanol, and freeze-dried. The dried extracts were stored at −20 °C and resolubilized in absolute MeOH to a concentration of 5 mg/mL for RP-UHPLC-MSn analysis. The hexane fractions contained no GSLs; thus, they were not considered for further analysis.

RP-UHPLC-MSn Analysis

Analysis of GSLs was performed on an Accela ultrahigh-performance liquid chromatography (UHPLC) system (Thermo Scientific, San Jose, CA, U.S.A.) equipped with a pump, autosampler, and photodiode array (PDA) detector. An LTQ Velos electrospray ionization (ESI) ion trap mass spectrometer (MS) (Thermo Scientific) was coupled to the LC system.

Sample (1 μL) was injected onto an Acquity UPLC-BEH shield RP18 column (2.1 mm i.d. × 150 mm, 1.7 μm particle size; Waters, Milford, MA, U.S.A.) with an Acquity UPLC BEH shield RP18 VanGuard precolumn (2.1 mm i.d. × 5 mm, 1.7 μm particle size; Waters). Water acidified with 0.1% (v/v) FA + 1% (v/v) ACN (eluent A) and ACN acidified with 0.1% (v/v) FA (eluent B) were used as solvent at a flow rate of 400 μL/min. The temperature of the sample tray was controlled at 15 °C. The column was set at 35 °C. The PDA detector was set to monitor absorption at 200–400 nm. The following elution gradient was used: 0–5.5 min, an isocratic on 0% (v/v) B; 5.5–32.4 min, a linear gradient to 49% B; 32.4–33.5 min, a linear gradient from 49 to 100% B; 33.5–39 min, an isocratic on 100% B; 39–40 min, a linear gradient from 100% to 0% B; 40–45.6 min, an isocratic on 0% B.

Mass spectrometric analysis was performed on an LTQ Velos equipped with a heated ESI-MS probe coupled to RP-UHPLC. The spectra were acquired in an m/z (mass to charge ratio) range of 92–1000 Da in both positive (PI) and negative ionization (NI) modes. Data-dependent MSn analysis was performed on the most intense ion with a normalized collision energy of 35%. The system was tuned with GSB via automatic tuning using Tune Plus (Xcalibur v.2.2, Thermo Scientific). Nitrogen was used as sheath gas, and helium was used as auxiliary gas. The ion-transfer tube (ITT) temperature was set at 300 °C, and the source voltage was 4.5 kV for both ionization modes.

In each set of analyses, calibration curves of p-hydroxybenzyl GSL (for benzenic GSLs), 4-pentenyl GSL (for aliphatic GSLs), and I3M GSL (for indolic GSLs) standards were made in the range of 1–650 μM. Calibration curves (R2 ≥ 0.993) were based on the peak area of the full MS signal of the external standards in NI mode. The 12 GSL standards were analyzed at 10 and 50 μM to support peak annotation and quantification by using MS-based relative response factors (RRF). The concentration of GSLs whose standard compounds were not available in the analysis was quantified by using RRF of a GSL with the most similar structure and molecular weight (Table S1). Peaks 10 and 28 were present in trace amounts, and because their subclass was not annotated, no quantification was done for these peaks. Operation of the LC-MS system and data processing were done by using the software packages Xcalibur 2.2 and LTQ Tuneplus 2.7 (both Thermo Scientific, San Jose, U.S.A.).

Statistical Analysis

To test for significance between treatments within species, the data were statistically evaluated by analysis of variance (ANOVA), followed by Tukey post hoc analysis using IBM SPSS Statistic v.23 software (SPSS, Inc., Chicago, IL, U.S.A.).

Results

Tentative Annotation of GSLs

On the basis of information extracted from the literature, detailed hereafter and in the Supporting Information, 31 GSLs in total were tentatively annotated from S. alba, B. napus, and B. juncea by RP-UHPLC-PDA-MSn analysis. The first criterion to distinguish peaks of GSLs from other types of compounds is the presence of fragment ion at m/z of 259 in NI mode.[37] The second criterion is to distinguish classes of GSLs: the m/z of deprotonated molecular ion [M–H]− of intact aliphatic and benzenic GSLs is at an even number as they contain one nitrogen atom, whereas that of indolic GSLs is at an odd number as they contain two nitrogen atoms. The third criterion is to confirm the classes of GSLs as well as to notice the presence of substituents having a conjugated system: benzenic GSLs with O-substitution on the phenyl ring, indolic GSLs, and aromatic acyl derivatives of any GSLs show specific UVmax. However, for minor trace peaks this criterion is less secure. The three criteria, summarized in Table S2, are useful for fast screening the class of a GSL. The molecular ions [M–H]− (m/z) of GSLs, fragmentation patterns, retention times, and UV absorption spectra allowed the annotation of 31 GSLs from S. alba, B. napus, and B. juncea (Table S3), where the annotation of 12 GSLs was confirmed by the standards listed in the Materials and Methods and that of 19 GSLs was tentative. Because many GSLs share similar fragmentation patterns and modifications of the side chain can create many possible isomers with no substantial differences in polarity (e.g., hydroxylated phenethyl GSL and methoxylated benzyl GSL, Figure C), annotation can be difficult. Consequently, several peaks (e.g., 10, 12, 18, 19, and 28) were tentatively annotated with multiple possible molecular structures and/or formulas (Table S3).

Of 19 GSLs, four (11, 6, 25, and 7) were tentatively annotated as 3-(methylthio)propyl GSL, 5-(methylsulfinyl)pentyl GSL, 10-(methylsulfinyl)decyl GSL, and 2-hydroxy-4-pentenyl GSL, respectively, according to the fragmentation pattern of the analogue within the same subclass whose standards were available in the analysis and the logic of their retention times. The MS spectra of 4-(methylsulfinyl)butyl GSL (3) (the standard) and 5-(methylsulfinyl)pentyl GSL (6) (present in an extract), as representatives, are displayed in Figure . In addition, fragmentation patterns of 6 and 7 were in line with the results of Cataldi et al.,[38] who analyzed the reference rapeseed (B. napus). According to the molecular weight and retention time, 12 and 19 were tentatively annotated as x-hydroxy-4-(methylthio)butyl GSL and x-hydroxy-5-(methylthio)pentyl GSL, respectively, where the hydroxylation might occur at 2- or 3-position.[17,18] Two indolic GSLs (9 and 24) were tentatively annotated as 4-hydroxy-I3M GSL and 1-methoxy-I3M GSL, respectively, according to the fragment ions of those observed in rapeseed in our analysis as well as in the analysis done by Cataldi et al.[38] using a reference rapeseed. Another indolic GSL (22) was tentatively annotated as 4-methoxy-I3M GSL according to the discriminating fragmentation pattern for 4-methoxy-I3M GSL and 1-methoxy-I3M GSL observed by Pfalz et al.[39] in Arabidopsis thaliana and Olsen et al.[17] According to this discriminating fragmentation pattern for the 4- and 1-substituted indolic GSLs (Figure S2), one more indolic GSL (31) was tentatively annotated as 4-salicyloxy (or isomer)-I3M GSL. Four GSLs (26, 27, 29, and 30) were tentatively annotated as GSLs acylated at the thioglucosyl group, by comparison of the fragmentation patterns to the existing reports on various purified acylated GSLs.[9,13] Peak 18 might represent a benzenic GSL with a side chain formula of C8H9O, with potentially different isomeric structures (Table S3). Two GSLs (16 and 23) were tentatively annotated as C5 and C6 alkyl GSLs, respectively, according to the presence of diagnostic fragment ions of GSLs, molecular weight, and retention time.

Figure 2

MS spectra of standard 4-(methylsulfinyl)butyl GSL (4-MSbutyl, 3) (A) and tentatively annotated 5-(methylsulfinyl)pentyl GSL (5-MSpentyl, 6) (B). Fragmentation occurred for the most abundant ion. The fragment ion with a neutral loss of 64 Da (CH3SOH) as the most abundant ion in MS2 spectra is the characteristics of MS2 spectra of (methylsulfinyl)alkyl GSLs. The fragment ions with m/z of 195, 241, 259, 275, and 291 shown in MS3 spectra are diagnostic fragment ions for GSLs.

Two detected peaks were left unclassified (Table S3). Peaks 10 and 28 had a molecular weight 14 Da different from 16 and 23, respectively. However, 10 and 28 demonstrated a neutral loss of 18 Da in MS2 fragmentation, which was not observed for the aliphatic GSLs 16 and 23. Therefore, 10 and 28 were tentatively annotated with potential side chain formulas of C4H9 or C3H5O and C7H15 or C6H11O, respectively (Table S3).

Overall Compositional Changes of GSLs by Germination and Fungal Elicitation

(R)-2-Hydroxy-3-butenyl GSL (2), I3M GSL (17), and 1-methoxy-I3M GSL (24) were present throughout all the studied seeds and (elicited) seedlings (Table S3). GSL profile in the untreated S. alba seeds consisted of 9 GSLs, and this number was increased upon germination up to 15 GSLs (Figure ). This diversity was not improved further by fungal elicitation but even decreased by R. oryzae (Figure ), and this trend applied to B. napus and B. juncea.

Figure 3

NI-MS chromatograms of S. alba seeds (the untreated, Sa_U) (A) and seedlings (the germinated, Sa_G) (B). The color codes refer to those in Figures –7

Figure 4

Total GSL content (μmol/g DW, primary y-axis) and GSL diversity indicated by the number of GSLs (triangles) (secondary y-axis) in the untreated (U), germinated (G), R. oryzae-germinated (Ro-G), F. graminearum-germinated (Fg-G), and F. oxysporum-germinated (Fo-G) seeds of S. alba, B. napus, and B. juncea. Data of total GSL content are the mean values from three biologically independent experiments. Error bars show standard deviation of total GSL content. Different letters at the bars show significant difference of the mean of total GSL content between treatments at p < 0.05.

Figure 5

Total content of aliphatic GSLs and their diversity per subclass in the untreated (U), germinated (G), R. oryzae-germinated (Ro-G), F. graminearum-germinated (Fg-G), and F. oxysporum-germinated (Fo-G) seeds of S. alba (A), B. napus (B), and B. juncea (C). The bars from bottom to top correspond to more downstream in the biosynthesis. Data are the mean values from three biologically independent experiments. Error bars show standard deviation of total content. Different letters at the bars show significant difference of the mean of total content between treatments at p < 0.05. The first compound mentioned in the brackets in the legend was in general the predominant within the subclass. The content of alkyl subclass (16 and 23) was invisible due to its low content (<0.9% of total aliphatic content).

Figure 7

Total content of indolic GSLs and their diversity in the untreated (U), germinated (G), R. oryzae-germinated (Ro-G), F. graminearum-germinated (Fg-G), and F. oxysporum-germinated (Fo-G) seeds of S. alba (A), B. napus (B), and B. juncea (C). The bars from bottom to top correspond to more downstream in the biosynthesis. Data are the mean values from three biologically independent experiments. Error bars show standard deviation of total content. Different letters at the bars show significant difference of the mean of total content between treatments at p < 0.05.

Furthermore, Figure illustrates the total GSL content in the seeds, untreated seedlings, and treated seedlings. S. alba and B. juncea seeds were the richest in GSLs, with 76.5 and 80.5 μmol/g DW, respectively, whereas B. napus seeds contained only 9.6 μmol/g DW. The total GSL content in S. alba was significantly increased by 2.9-fold upon germination, 2.4-fold upon Rhizopus elicitation, and 2.2-fold upon F. graminearum elicitation. In B. napus and B. juncea, neither germination nor fungal elicitation enhanced the total GSL content significantly. These results were not significantly affected after considering a dry weight loss (commonly reported at ∼10%) due to respiration during germination.[40−43]

Compositional Changes of Aliphatic GSLs

The aliphatic GSL content in S. alba seeds was 4.0 μmol/g DW, and this content remained similar in the nonelicited seedling and F. graminearum-elicited seedling but was decreased (>50%) in R. oryzae-elicited seedling (Figure A). Furthermore, the total aliphatic content was increased by 60% in F. oxysporum-elicited seedlings. Differently, the aliphatic GSL content in B. napus seed was 8.3 μmol/g DW, and this content decreased to 3.6 μmol/g DW already upon germination; that content in the seedlings was statistically similar to that in the fungal-elicited ones (Figure B). Figure C shows that the aliphatic GSL content in B. juncea seed was 79.8 μmol/g DW, much higher than that in S. alba and B. napus seeds. Furthermore, this content in B. juncea seed was stable upon any treatment, i.e., germination and fungal elicitation, but decreased (50%) in the F. oxysporum elicitation (Figure C).

With respect to the chemical diversity, Figure A indicates that there were 3 subclasses of aliphatic GSLs, namely, hydroxylated alkenyl, alkenyl, and hydroxylated MTalkyl in S. alba seeds. (R)-2-Hydroxy-3-butenyl GSL (2) represented 80% of total aliphatic GSL content, whereas the rest was composed of 3-butenyl GSL (8), allyl GSL (4), and x-OH-4-(methylthio)butyl GSL (12). Upon germination, the content of hydroxylated alkenyl subclass was stable, the alkenyl subclass was reduced significantly, and the OH-MTalkyl subclass was increased. Upon fungal elicitation the diversity was not improved, but the content of OH-MTalkyl subclass was increased by Fusarium elicitations (Figure A).

Figure B indicates that B. napus untreated seeds consisted of 4 major aliphatic subclasses, namely, hydroxylated alkenyl, alkenyl, MSalkyl, and MTalkyl, from the highest to the lowest content. Upon germination, the four subclasses remained present with the same two top major ones. Interestingly, the acylated alkenyl GSL, particularly 6′-O-sinapoyl (or isomer)-3-butenyl GSL (29), emerged upon germination and Fusarium elicitation. In contrast, MTalkyl GSLs and acylated alkenyl GSLs were absent in Rhizopus-elicited seedlings. It is noteworthy to mention that the acylated GSLs were not artifacts; in our anhydrous methanol extraction at 65 °C, esterification did not occur because the chromatograms did not contain peaks corresponding to methyl esters of sinapic acid or other aromatic acids.

Figure C, with the help of Table S6, indicates that B. juncea seed consisted of 6 aliphatic subclasses, namely, MTalkyl, MSalkyl, alkenyl, hydroxylated alkenyl, acylated alkenyl, and alkyl. 3-Butenyl GSL (8) was the predominant (77% of total aliphatic GSL content). The other 5 subclasses were present at a proportion of <4%. Upon germination, there was a significant increased content of acylated alkenyl subclass, i.e., sinapoyl (or isomer) derivatives of 3-butenyl GSL (from 1.0 to 37.3 μmol/g dW). Both Fusarium-elicited seedlings also contained acylated alkenyl GSLs at a comparable level to that in the nonelicited seedling (Table S6). In contrast, Rhizopus-elicited seedlings contained acylated alkenyl subclass in a very low level (0.4 μmol/g DW).

Compositional Changes of Benzenic GSLs

The benzenic GSL content in S. alba seed was high, 72.4 μmol/g DW, and this content was increased by 2–3-fold in the nonelicited seedlings, Rhizopus-elicited, and F. graminearum-elicited seedlings but remained the same in F. oxysporum-elicited seedling (Figure A1). Differently, the benzenic GSL content in B. napus seed was very low, <0.1 μmol/g DW, and this content increased up to 4.4 μmol/g DW upon F. oxysporum elicitation (Figure B). Figure C shows that benzenic GSL content in B. juncea was as low as that in B. napus. Total benzenic GSL content in B. juncea was increased to 0.5 μmol/g DW by germination but remained the same by fungal elicitation (Figure C).

Figure 6

Total content of benzenic GSLs and their diversity in the untreated (U), germinated (G), R. oryzae-germinated (Ro-G), F. graminearum-germinated (Fg-G), and F. oxysporum-germinated (Fo-G) seeds of S. alba (A1), B. napus (B), and B. juncea (C). Data are the mean values from three biologically independent experiments. Error bars show standard deviation of total content. Different letters at the bars show significant difference of the mean of total content between treatments at p < 0.05.

With respect to the chemical diversity, Figure A1 indicates an abundant amount of p-hydroxybenzyl GSL (5), which contributed to >99% of the total benzenic GSL content in S. alba seed and (elicited) seedlings. The presence of other benzenic GSLs can be seen more clearly in Figure A2. Interestingly, the content of acylated benzenic GSL, i.e., 6′-O-sinapoyl (or isomer)-p-hydroxybenzyl GSL (26), was increased to 1.8 μmol/g DW by germination. However, the content in the fungal-elicited seedlings (0.3–0.6 μmol/g DW) was significantly less than that in the nonelicited seedling.

Figure B, with the help of Table S5, indicates only 2 benzenic GSLs found in B. napus seeds at very low content (<0.1 μmol/g DW), namely, p-hydroxybenzyl GSL (5) and phenethyl GSL (20). The content of GSL 5 was dramatically increased by F. oxysporum elicitation. Two more Phe-derived GSLs, namely, benzyl GSL (14) and hydroxylated or methoxylated Phe-derived GSLs with side chain formula of C8H9O (18), were present in the nonelicited and F. graminearum-elicited seedlings.

Figure C demonstrates that B. juncea seeds contained benzyl GSL (14) and phenethyl GSL (20). Upon germination and F. oxysporum elicitation, the diversity of benzenic GSLs was slightly improved, indicated by the presence of p-hydroxybenzyl GSL (5).

Compositional Changes of Indolic GSLs

Figure indicates that the total content of indolic GSLs in S. alba, B. napus, and B. juncea seeds was generally low, i.e., <1.2 μmol/g DW. The content was increased upon germination, Rhizopus elicitation, and F. graminearum elicitation up to 10.8 μmol/g DW but remained unchanged upon F. oxysporum elicitation. In addition, Figure shows that germination and fungal elicitation resulted in a higher proportion of methoxyl derivatives of I3M GSL (22 and 24). The acylated indolic GSL (31) was only found in S. alba seedlings elicited by R. oryzae (Figure A).

Discussion

Tentative Annotation of GSLs and the Predominant GSLs in the Studied Brassicaceous Seeds

LC-MSn analysis in the NI mode allows GSL peak tentative annotation due to robust ion MSn data that are unique to GSLs as described in the Supporting Information. Of 31 GSLs annotated, 12 were confirmed by comparison with their authentic standards. The predominant GSLs in the three seed species were among the 12 GSLs. p-Hydroxybenzyl GSL (5) (72 μmol/g DW) was the signature GSL in S. alba seed, supporting Popova and Morra (110–210 μmol/g defatted seed meal).[44] (R)-2-Hydroxy-3-butenyl GSL (2), 3-butenyl GSL (8), and 4-hydroxy-I3M GSL (9) were the signature GSLs in B. napus seed (5.0, 2.5, and 1.0 μmol/g DW, respectively), supporting Borgen et al. (3.1, 1.7, and 4.4 μmol/g DW, respectively).[45] 3-Butenyl GSL (8) and allyl GSL (4) were the signature GSLs in B. juncea seed (61.6 and 9.3 μmol/g DW, respectively), supporting Sodhi et al. (66–90 and 20–37 μmol/g DW, respectively).[46] Most of the minor GSLs tentatively annotated in our study were also found or suggested in previous studies. Different GSL contents between studies can be caused by many factors, e.g., origins and varieties of the plants.

Biosynthesis of GSLs Activated by Germination without Fungus

On the basis of total GSL content presented in Figure , there is evidence that germination, without additional elicitor, activates GSL biosynthesis in S. alba, one of the GSL-rich species. Furthermore, on the basis of GSL compositions indicated in Figures –7 and Tables S4–S6, germination activates GSL biosynthesis in all three plant species: indolic GSL in all species, benzenic GSLs in S. alba and B. juncea, and acylated aliphatic GSLs in B. napus andB. juncea. In comparison with Leguminosae seeds, where germination induces mainly deglucosylation of isoflavonoids,[31−33,35] germination of Brassicaceae seeds instead activates biosynthesis of GSLs, leading mainly to a higher accumulation of more downstream GSLs, e.g., acylated aliphatic in Brassica species (Figure B and C), acylated benzenic in S. alba (Figure A2), and methoxylated indolic in all species (Figure ). Given the increase in GSL content and/or diversity upon germination, it is suggested that GSL hydrolysis to ITCs or other turnover is not a quantitatively major process, although this was not further substantiated in this study.

Elicitation by Fungus in Attempt to Boost GSL Content or Diversity beyond That by Germination

Figure indicates clearly that the elicitations by nonpathogenic R. oryzae and F. graminearum, as well as pathogenic F. oxysporum, do not increase GSL content further than what could be obtained by germination alone; instead, GSL content tends to decrease. This could be an indication of GSL degradation. This is related to the possibility that plant cell damage might occur upon fungal elicitation due to the fungal cell expansion on the growing seeds. Consequently, GSLs and myrosinase contact each other, releasing the antimicrobial compounds. Further research is necessary to confirm this. In contrast, in the case of legumes, e.g., lupine beans, elicitation by R. oryzae during germination increased the isoflavonoid content further up to 2-fold.[35]

Compared with germination alone, GSL diversity was not enhanced by fungal elicitation (Figure ); instead, there was a reduction of GSLs, i.e., degradation. Such reduction was observed with sinapoyl (or isomer) derivatives of 3-butenyl GSL (aliphatic class) in B. napus and B. juncea seedlings due to Rhizopus elicitation (Figure B and C, by Ro-G) and with 6′-O-sinapoyl (or isomer)-p-hydroxybenzyl GSL (26) in S. alba due to all three fungal elicitations (Figure A2). Little is known about the degradation of acylated GSLs. In another study,[10] the content of isoferuloylated GSLs decreased upon germination, which was not due to leaching but was probably due to hydrolysis. Our study suggests that the degradation of sinapoyl (or isomer) derivatives is inducible by fungi, but the mechanism by which they are degraded (by in-plant myrosinase and/or fungal enzymes) remains unclear.

Furthermore, fungal elicitation modulated GSL composition in a different manner per GSL class, per plant species, and per fungal species, particularly for aliphatic and benzenic classes (Figures and 6). There was no clear trend to which direction the composition of aliphatic GSLs and benzenic GSLs was shifted. In some cases, more downstream GSLs were induced (e.g., increased content of phenethyl GSL (20) by Ro-G in B. juncea, Figure C), whereas in other cases, more upstream GSLs were induced (e.g., increased content of MTalkyl subclass by Ro-G in B. juncea, Figure C). This inconsistent shift in aliphatic and benzenic biosynthesis suggests that the downstream aliphatic and benzenic GSLs do not necessarily contribute to a better antimicrobial activity than the upstream ones. In contrast, in the case of legumes, the isoflavonoid composition was consistently shifted to more downstream isoflavonoids after germination and fungal elicitation. Isoflavonoids with different skeletons were made along with another modification, i.e., prenylation, known to increase the antimicrobial activity of isoflavonoids.[31−33]

Among GSL degradation products, ITCs are known to be the most potent antimicrobials.[1,47] However, it remains unclear whether ITCs derived from a more downstream GSL are more active than those derived from a more upstream GSL, or whether aliphatic ITCs are more active than benzenic ones.[1,48,49] Further studies to reveal the antimicrobial activities of various ITCs are necessary.

Effect of Pathogenicity of the Fungus on the Content of Indolic GSL

Studies have proven that indolic GSLs in B. oleracea and B. juncea seedlings were inducible by additional phytohormones, e.g., salicylic acid and methyl jasmonate.[27,50] Our study is the first to report that indolic GSLs are inducible by germination as well as by fungal elicitation. The induction of downstream steps in indolic GSL biosynthesis is consistently observed by the presence of the methoxylated indolic subclass as the main subclass induced (Figure ). Methoxylated indolic GSLs, in particular 4-methoxy-I3M GSL (22), have been reported to be vital for defense against pathogens and to facilitate innate immunity.[51,52] We hypothesized that the pathogenic fungus would affect GSL profiles more extensively, compared to the nonpathogenic fungi. We observed a consistently lower content of 4-methoxy-I3M GSL (22), in particular, and total indolic GSLs, in general, in seedlings infected by pathogenic F. oxysporum, compared with the nonpathogenic fungi. Our study potentially suggests that, for Brassicaceae germinating seeds, the content of indolic GSLs could be used as a marker to distinguish between elicitation by pathogenic or nonpathogenic fungi. There is no similar study on this matter. Therefore, further research needs to be done to confirm our findings, for instance, extending the number of Brassicaceae species and varieties and the attacking fungi based on their pathogenicity level.

This is the first report on the compositional changes of GSLs in Brassicaceae seeds, which have their own unique GSL profiles, upon germination and fungal elicitation. The accumulation of GSLs in germinating Brassicaceae seeds could not be successfully enhanced by fungal elicitation. This indicates that the approach that was successfully employed for increasing the accumulation of isoflavonoids in leguminoceous seedlings could not simply be extrapolated to a different plant family to enhance the accumulation of its respective major phytochemicals. Furthermore, our hypothesis that the pathogenic fungus would induce higher and more diverse GSL production could not be accepted. Neither was there a clear trend in the effect of the pathogenicity of the fungus on the total GSL content and diversity. Opposite to our hypothesis, the elicitation with pathogenic fungus (i.e., F. oxysporum) consistently caused lower content of indolic GSLs than that with nonpathogenic fungi (F. graminearum and R. oryzae).