Cytotoxic and genotoxic potential of food-borne nitriles in a liver in vitro model

Isothiocyanates are the most intensively studied breakdown products of glucosinolates from Brassica plants and well recognized for their pleiotropic effects against cancer but also for their genotoxic potential. However, knowledge about the bioactivity of glucosinolate-borne nitriles in foods is very poor. As determined by GC-MS, broccoli glucosinolates mainly degrade to nitriles as breakdown products. The cytotoxicity of nitriles in human HepG2 cells and primary murine hepatocytes was marginal as compared to isothiocyanates. Toxicity of nitriles was not enhanced in CYP2E1-overexpressing HepG2 cells. In contrast, the genotoxic potential of nitriles was found to be comparable to isothiocyanates. DNA damage was persistent over a certain time period and CYP2E1-overexpression further increased the genotoxic potential of the nitriles. Based on actual in vitro data, no indications are given that food-borne nitriles could be relevant for cancer prevention, but could pose a certain genotoxic risk under conditions relevant for food consumption.

has been done on their effect on phase II detoxification enzymes and cytotoxicity as parameters for primary and secondary/tertiary cancer prevention, respectively. A previous study compared the ability of 4-(methylsulfinyl) butyl ITC (sulforaphane) and its structural analog 5-(methylsulfinyl)pentylnitrile (4-MSOB-CN, sulforaphane nitrile) to induce the activity of mammalian phase II detoxification enzymes in Fischer 344 rats. The potential of sulforaphane nitrile to induce activity of quinone reductase was less than sulforaphane, reaching the same induction as sulforaphane in a thousand-fold concentration 12 . In contrast, administration of the nitrile crambene (3-hydroxy-4-pentenenitrile) to Fischer 344 rats [50 mg/kg BW/day] yielded an induction of hepatic quinone reductase similar to sulforaphane 13 . A recent metabolism study of 3-butenenitrile (allyl-CN) showed that the cytochrome P450 isoform CYP2E1 is involved in the allyl-CN induced lethality of mice resulting from cyanide liberation 14 . Because of their reactivity, ITC are basically able to act also as genotoxic agents, form DNA adducts, and induce DNA mutations. Some studies even report about carcinogenic effects 15 . Whether these adverse effects could also be relevant for nitriles formed from GSL is not known.
The present study aimed at initially characterizing the formation of different food-borne nitriles by enzymatic hydrolysis of GSL in selected fresh Brassica vegetables. In order to investigate structure dependent effects of nitriles in vitro, three groups (unsaturated aliphatic nitriles, aromatic nitriles, and methylthioalkylnitriles) were selected, that belong to the most common and most important groups of glucosinolate-borne nitriles. Secondly, this study evaluated the cytotoxic and genotoxic potential of food-borne nitriles in comparison to ITC using human liver HepG2 cells as well as primary murine hepatocytes. To study whether CYP2E1 can promote hepatotoxicity, CYP2E1-overexpressing HepG2 cells were used.

Results
Occurrence of glucosinolate-derived nitriles in Brassica vegetables. In order to investigate the relevance of plant-derived nitriles from fresh vegetables, the GSL content before and the nitrile content after autolysis of Brussels sprouts, broccoli, cauliflower, savoy cabbage, white cabbage, and red cabbage were studied. The results are presented in Table 1. A GC-MS chromatogram of nitriles from brokkoli is presented in the supplementary information. Individual GSL contents as well as the corresponding nitriles are given in μ mol/g fresh weight (FW). Additionally, the percentage of the nitrile concentration in relation to the total content of all degradation products derived from an individual GSL are given. All B. oleracea vegetables formed nitriles. Especially, broccoli and cauliflower were producers of significant contents of nitriles, with 57% and 48% of all detected breakdown products being nitriles. The predominant nitriles were 4-(methylsulfinyl)butylnitrile (3-MSOP-CN) ranging from 0.01 (in cauliflower) to 0.43 μ mol/g FW (in savoy cabbage), and 5-(methylsulfinyl)pentylnitrile (4-MSOB-CN) ranging from 0 μ mol/g (in cauliflower) to 0.32 μ mol/g FW in broccoli, that was exceptionally rich in this nitrile. Further, broccoli and Brussels sprouts had also very high nitrile ratios (65% of the degradation products of 3-MSOP GLS were 3-MSOP-CN in broccoli) with regard to total GSL except the unsaturated aliphatic ones (Table 1). However, Brussels sprouts are also a main producer of epithionitriles 16 . Cytotoxicity of food-borne nitriles in human liver (HepG2) cells. As there is only limited information about secondary and tertiary cancer chemopreventive effects, cytotoxicity of unsaturated aliphatic, aromatic, and methylthioalkylnitriles ( Fig. 1) against human liver cancer (HepG2) cells was studied. Keeping the toxicity of the related ITC in mind 17,18 , first investigations of the cytotoxicity of nitriles were carried out using concentrations ranging from 0.1 to 100 μ M. However, these concentrations did not impact cell viability (data not shown). Thus, a concentration range between 0.3 to 30 mM was tested. Due to a limited solubility of nitriles in purified water, no higher concentrations than 30 mM were achieved. Treatment of HepG2 cells with unsaturated aliphatic nitriles (allyl-CN, 3-but-CN and 4-pent-CN) for 72 h did not result in significant changes in cell viability. In contrast, aromatic nitriles and methylthioalkylnitriles induced a loss of viability at concentrations exceeding 10 mM (Fig. 2). However, a significant reduction of cell viability was detected only at the highest concentration tested. The cytotoxic potential of the two aromatic nitriles was similar and lead to IC 50 values of 19.95 mM (benzyl-CN) and 18.21 mM (2-phenylethyl-CN), respectively (Table 2). Comparative investigations with benzyl-ITC resulted in a 1000-fold stronger cytotoxic potency with an IC 50 of 15.75 μ M. Using methylthioalkylnitriles, a structure dependent increase in cytotoxicity could be observed with increasing chain length, with the lowest IC 50 at 8.15 mM for 6-MTH-CN. The IC 50 for 3-MTP-CN, 4-MTB-CN, and 5-MTP-CN are given in Table 2. Figure 1 shows the structural differences of these nitriles.
Relevance of cytochrome-P450 expression for nitrile cytotoxicity. Recently, a study showed that the cytochrome-P450 isoform CYP2E1 is involved in the allyl-CN induced lethality of mice 14 . In contrast to normal human hepatocytes, the HepG2 cell line expresses only very low amounts of CYP2E1. This was verified by the present study using qPCR. Consequently and in order to assess whether the presence of CYP2E1 could further elevate the toxicity of the food-borne nitriles, the compounds were tested again using CYP2E1-overexpressing HepG2 cells. In comparison to wildtype HepG2 cells, the CYP2E1 expression in HepG2-CYP2E1 cells was 300 times higher. Results from two unsaturated aliphatic nitriles (allyl-CN, 3-but-CN), the aromatic nitrile benzyl-CN, and the methylthioalkylnitrile 4-MTB-CN are shown in Fig. 3. Only 10 mM 4-MTB-CN showed a significant difference in cytotoxicity between CYP2E1-overexpressing and vector control cells. To the best of our knowledge, there are no other proteins described in the literature that have an influence on the cytotoxicity of nitriles other than CYP2E1.
Cytotoxicity of food-borne nitriles in normal murine hepatocytes. To analyze the effect of nitriles on primary hepatocytes, allyl-CN, 3-butyl-CN, benzyl-CN, and 4-MTB-CN were studied in hepatocytes freshly isolated from mice. Allyl-CN, 3-but-CN and benzyl-CN significantly decreased the viability of hepatocytes in a concentration-dependent manner (Fig. 4) as analyzed by WST-1 assay after 72 h of exposure. This loss in cell Scientific RepoRts | 6:37631 | DOI: 10.1038/srep37631 viability was higher for allyl-CN and 3-but-CN than in HepG2 cells ( Fig. 2A). For 4-MTB-CN, a viability loss was detected only at 30 mM, comparable with the effect in HepG2 cells (Fig. 2C).

Genotoxicity of food-borne nitriles in HepG2 cells in comparison to isothiocyanates and the relevance of CYP2E1.
Risk-benefit evaluation of bioactive compounds formed in Brassica vegetables is not only based on its cancer preventive potency, but also on further aspects such as genotoxicity 15 . ITC are comparatively electrophilic compounds that have been shown to act genotoxic, at least in vitro 19,20 . Representatively, the genotoxicity was investigated in terms of DNA break induction for 4-MTB-CN and benzyl-CN in comparison to benzyl-ITC using the Comet assay (Fig. 5). 4-MTB-CN and benzyl-CN induced a concentration-dependent increase in DNA strand breaks as determined by the Olive Tail Moment and percent tail DNA ( Fig. 5A and B). A significant increase in DNA strand breaks was evident for 4-MTB-CN and benzyl-CN at a concentration of ≥ 3 μ M. The DNA damage induced by 3 μ M benzyl-CN was then comparable with the effect of benzyl ITC at the same concentration ( Fig. 5A and B). After 24 h of recovery, the DNA damage still persisted ( Fig. 5C and D). Analog to the cytotoxicity tests, next the effect of CYP2E1 expression on the genotoxic potency of nitriles was analyzed (Fig. 6). CYP2E1-overexpression caused a significant increment of DNA strand breaks at concentrations of ≥ 3 μ M benzyl-CN as compared to vector control cells. This indicates a relevance of CYP2E1 for the genotoxic potential of this nitrile.

Discussion
Many Brassica vegetables contain alkenyl-GSL that can be hydrolyzed to ITCs, nitriles, and epithionitriles. However, studies on the occurrence of food-borne nitriles and the knowledge about their biological effects are still not comprehensively. In the present study, the predominant nitriles in vegetables were 3-MSOP-CN and 4-MSOB-CN, the latter ranging from 0 μ mol/g (in cauliflower) to 0.32 μ mol/g FW in broccoli that was exceptionally rich in this nitrile. This is in line with data reported in the literature. In the study of Matusheski et al. (2004), 4-MSOB-CN concentrations ranged from 0.35 to 1.50 μ mol/g fresh weight (in broccoli). Moreover, the preferential formation of 3-MSOB-CN and 4-MSOB-CN compared to the corresponding ITC has been already observed by Daxenbichler et al. 21 for freshly autolyzed cabbage.
In the actual study, Brussels sprouts had also very high nitrile ratios with regard to total GSL, except the unsaturated aliphatic ones. The latter released, probably due to a high ESP activity, the corresponding epithionitriles and therefore Brussels sprouts were in total a producer of epithionitriles which is in agreement to others 21 . However, the highest total nitrile concentration with 0.58 μ mol/g FW was found in savoy cabbage, although it was, likewise to red cabbage, also a well-known producer of ITC. This was due to a high content of 3-MSOP-GSL and its hydrolysis products comprised to 37% of the nitrile. Therefore, consumption of 200 g savoy cabbage would mean an ingestion of 86 μ mol of this nitrile. Moreover, it has been shown that the aliphatic nitrile allyl-CN  resulting from allyl-GSL occurred, in all of the B. oleracea vegetables investigated, except in broccoli. Consistently, in all analyzed B. oleracea vegetables also 3-MTP-CN was formed. The occurrence of this nitrile has been already reported for cabbage and cauliflower 21,22 . However, during the last decades, ITC were made responsible for any physiological effects of Brassica consumption, e.g., also for cancer prevention. Therefore, it is mandatorily necessary to also clarify possible effects of food-borne nitriles. As shown by Matusheski and Jeffery 12 , the ability of the sulforaphane nitrile to induce quinone reductase was about 1000-fold less than that of sulforaphane itself. Sulforaphane induced quinone reductase at a concentration of 2.5 μ M, reaching a maximum induction of 3.1-fold, whereas the corresponding nitrile reached a similar induction only at a concentration of 2 mM 12 . In a recently published study, strong interference of benzyl-ITC at ≥ 3 μ M with the cellular pro-inflammatory response including prostaglandin PGE2 release was shown in activated human T lymphocytes. In contrast, benzyl-CN did not influence PGE2 release even at a concentration of 1 mM 23 . These results confirm the observation that nitriles affect their target less than their corresponding ITC  Indications are given that a minor structural difference such as the loss of a methyl group, could lead to different structurally-dependent effects being involved in cytotoxicity mechanism. Ahmed & Farooqui 24 reported for rats that the toxicity of aliphatic nitriles not only diversifies with cyanide liberation but is also highly dependent on their chemical structure. In that study, saturated nitriles at amounts of 0.7 mmol produced symptoms of poisoning similar to the symptoms produced by inorganic cyanide (0.15 mmol). In contrast, unsaturated nitriles with amounts of 1.7 mmol showed only moderate symptoms 24 . In the present study, methylthioalkylnitriles with increasing chain length lead to a structure-dependent increase in cytotoxicity, with 6-MTH-CN being the most toxic substance (Table 2). It has to be taken into account that concentrations inducing cell toxicity were still very high. In a previous study, the IC 50 of 4-MTB-ITC was determined using the neutral red retention assay with 23.18 μ M in HepG2 cells 18 . This ITC, again, is a 1000-fold more potent in inducing cell death as compared to the cytotoxicity observed with the nitrile 4-MTB-CN in the present study. In accordance to others 24 , data of the present study also imply that no relevant cytotoxicity occurs when an aliphatic moiety in the nitrile is present. However, presence of sulfur or an aromatic ring in the chemical structure of the chain led to an increased cytotoxicity in HepG2 cells. This effect was still very weak as compared to ITC, underlining the hypotheses already described in the literature 12,13 . Keck et al. 13 reported that treatment of mouse hepatoma cells (Hepa1c1c7) with the brassicacaeous nitrile crambene (2-OH-3-but-CN) in a concentration of 5 mM for 72 h, diminished cell proliferation by 76%. Of those, cell viability was reduced by 12%. Thus, tested concentrations in this study are comparatively high and not expected to occur in vivo after consumption of processed or non-processed Brassica foods.
Nitriles seem only to be activated by CYP2E1 and other CYPs. Thus, it was suggested that the toxic effects result from cyanide liberation following the activation of certain enzymes 14 . Although cytotoxicity of nitriles was not affected by CYP2E1-overexpression in HepG2 cells in the present study, this was the case for genotoxicity. Here, a genotoxic effect was already significant at a concentration of ≥ 3 μ M. Moreover, CYP2E1 overexpression increased genotoxicity. Whereas ITC are well known genotoxic agents in vitro, this was not known for nitriles. For example Lamy et al. (2009) investigated the genotoxic potential of 3-MTP-ITC, 4-MTB-ITC, and 5-(methylthio)pentyl ITC (5-MTP-ITC) in HepG2 cells using the Comet assay as well as the micronucleus test. That study concluded that the ITC induced DNA damage and mutagenicity at concentrations exceeding 1 μ M. Based on the present data nitriles seem to be as potent as ITC to induce DNA damage and this damage seems to be persistent. The intensity of DNA strand breaks of 3 μ M benzyl-CN was comparable with the effect of benzyl-ITC in the same concentration. In vitro experiments with benzyl-ITC showed characteristic concentration-dependent genotoxic effects under different conditions (dependent on the cell line or endpoint) 25 . Comparatively weaker effects were observed in in vivo experiments. The genotoxicity of benzyl-ITC was reduced by bovine serum albumin and human body fluids, which suggests that benzyl-ITC is inactivated under in vivo conditions 25 . If nitriles are also detoxified in vivo, needs to be clarified in future studies.
Under the conditions applied in the present study, B. oleracea vegetables released high amounts of nitriles upon hydrolysis of GSL. Especially broccoli and cauliflower were intense producers of nitriles. The highest total nitrile concentration was detected in savoy cabbage, though comprising mainly of 3-MSOP-CN. With regard to cell culture experiments, tested nitriles were cytotoxic towards cancer cells only when being applied in 1000-fold higher concentrations, compared to the ITC. Thus, an impact of nitriles for a secondary and tertiary cancer preventive potential by Brassica vegetables seems neglectable. Moreover, with regard to common habits of food intake, an indication for a preventive potential of GSL-derived nitriles can still not be given with the present data. However, the tested nitriles showed genotoxic potential in HepG2 cells, comparable in effect to their corresponding ITC. Thus, the preference of nitrile instead of ITC formation under certain food processing conditions should be taken into account. Moreover, the weak cytotoxic activity of nitriles while being quite potent genotoxic agents should be considered carefully, as well. This genotoxic activity of nitriles, as shown in this study, can be considered as a risk factor in cancer prevention. Thus, this could be one relevant factor being involved in the contradictory observations made with cancer prevention studies using Brassica vegetables. As only three structure-types of nitriles were tested in the present study, further research with nitriles differing in their chemical structure is necessary. Based on the in vitro data presented, there are no indications that nitriles affect health without a preceding activation by CYP enzymes. However, with respect to genotoxicity data, it cannot be ruled out that nitriles pose a risk under conditions relevant for food consumption. Chemical synthesis of methylthioalkylnitriles. Due to the poor commercial availability of methylthioalkylnitriles, the four tested components have been synthesized according to Moon et al. 26 : 4-(methylthio)butylnitrile (3-MTP-CN). Sodium thiomethoxide (0.23 g, 3.3 mmol; 95%, Sigma-Aldrich GmbH, Steinheim, Germany) was dissolved in 5 mL methanol and gradually added to a solution of 4-bromobutyronitrile (0.5 g, 3.4 mmol; 97%) in 5 mL methanol. The mixture reacted 7 h under reflux conditions. Subsequent to evaporating the methanol in vacuo, the resulting product was dissolved in 13 mL water and extracted with diethyl ether (3 × 25 mL). The combined organic extracts were dried over sodium sulfate and the solvent was removed. 5-(methylthio)pentylnitrile (4-MTB-CN), 6-(methylthio)hexylnitrile (5-MTP-CN), 7-(methylthio)heptylnitrile (6-MTH-CN). The nitriles were synthesized according to the same method as described above. 0.5 g 5-bromovaleronitrile (3.1 mmol) to synthesize 5-(methylthio)pentylnitrile, 0.6 g 6-bromohexanenitrile (3.4 mmol) for 6-(methylthio)hexylnitrile, and 0.6 g 7-bromoheptanenitrile (3.2 mmol) for synthesis of 7-(methylthio)heptylnitrile were used. The amount of sodium thiomethoxide was kept unaltered: 0.23 g (3.3 mmol). For determining the amount of the different compounds, fully developed heads (cabbages, broccoli, and cauliflower) or fully developed leaf buds (Brussels sprouts) were harvested. For each Brassica vegetable, five biological replicates were taken using a representative aliquot for the determination of GSL and nitriles. Vegetables were free of injuries or pests and analyzed without delay after harvest. Brassica head forming and inflorescence vegetables were cut into 5 × 5 cm pieces (cabbages, broccoli, cauliflower), the leaf buds of Brussels sprouts were cut in half and mixed. Half of each sample was immediately frozen at − 50 °C and lyophilized for GSL determination. From the other half, an aliquot was crushed with a homogenizer (Edmund Bühler H04, Tübingen, Germany) after Scientific RepoRts | 6:37631 | DOI: 10.1038/srep37631 addition of an aliquot of water (one part of the plant/one part of water). From the homogenized sample an aliquot (1 g containing 500 mg of water) was immediately subjected to the analysis of nitriles.

Cell culture. HepG2 cells were purchased from the German Collection of Microorganisms and Cell
Cultures (DSMZ, Braunschweig, Germany). HepG2 cells which constitutively express human CYP2E1 (E47 cells, HepG2-CYP2E1) and control HepG2 cells transfected with empty vector (C34 cells, HepG2-vector), originally established by Wu and Cederbaum 27 , were kindly provided by C. Hellerbrand (University Medical Center Regensburg, Germany). The cells were cultured in low glucose Dulbecco's modified eagle medium supplemented with 15% fetal calf serum and 1% Pen/Strep in an atmosphere with 5% CO 2 at 37 °C and 95% air humidity.
Determination of the drug effect in primary murine hepatocytes. Murine hepatocytes were obtained from 8-16 week-old, male C57BL6 mice (Charles River Laboratory, Sulzfeld, Germany) as described elsewhere 18 . The institutional and national guidelines for the care and use of animals were followed, and all experimental procedures were approved by the institutional Animal Care and Use Committee at the University of Freiburg according to § 8.I of the Animal Welfare Act. 2-2.2 × 105 cells were plated in 24 well plates. The cells were left undisturbed for 4 h in a 5% CO 2 atmosphere at 37 °C and then the medium was changed to serum free Williams' Medium E supplemented with 100 nM dexamethasone. Within 24 h the cells were exposed to the test compounds and controls for 72 h.

Determination of glucosinolates from Brassica vegetables. GSL concentration in vegetable samples
was determined as desulfo-GSL using the DIN EN ISO 9167-1 based method as previously described in ref. 28. Briefly 15 mg of lyophilized plant material were extracted methanolically, GLS were desulfated and analyzed by UHPLC-DAD using a Poroshell 120 EC-C18 column, 100 mm × 2.1, particle size 2.7 μ m (Agilent Technologies and a gradient comprising of water and acetonitrile as reported previously 28 . Desulfo-GSL were identified by comparing retention times and UV absorption spectra with those of authentic standards. Quantification was done by using an external calibration curve with allyl-GSL and a wavelength of 229 nm. Determination of cytotoxicity. Cytotoxicity was determined with the WST-1 assay (Roche Diagnostics Deutschland GmbH, Mannheim, Germany). The assay is based on the ability of viable cells to cleave the tetrazolium salt WST-1 to formazan by their mitochondrial succinate-tetrazolium-reductase system. Cells were cultured at an initial density of 5000 cells/well in 96-well plates for 48 h. Then, cells were treated with the nitriles (0.3 to 30 mM, solved in distilled water) for 72 h. As positive control, 0.01% Triton X-100 was used. Following compound exposure and incubation, the WST 1 assay was performed according to the manufacturer's instructions.

Determination of nitriles in
Single cell gel electrophoresis assay (Comet Assay). The single cell gel electrophoresis assay also known as Comet assay was carried out as described earlier with slight modifications 20 . The percent tail DNA and the Olive Tail Moment (OTM) were calculated as indicators of DNA damage. For each sample, 102 systematically screened cells were evaluated. Briefly, HepG2 cells were seeded onto well plates. After 48 h of growth, the cells were exposed to the nitriles or ITC for 24 h. As positive control 50 μ M B(α )P and as solvent control distilled water or 0.1% DMSO was used. To analyze the repair capacity of HepG2 cells, the test substances were removed after 24 h and incubated for another 24 h before the assay was terminated.
Real-time polymerase chain reaction. Total RNA was isolated with the RNeasy mini Isolation kit from QIAGEN GmbH followed by a purification step using the RNase-free DNase kit from QIAGEN GmbH according to manufacturer's instructions. 5 μ g RNA of each cell line was dried by a vacuum concentrator (Eppendorf AG, Hamburg, Germany). Isolated RNA was re-suspended in 10 μ L of RNAse-free water. Each sample was treated twice with 2 μ L RNAse-free DNAse 1 unit/μ L (QIAGEN GmbH) for 10 min at 37 °C to eliminate remaining DNA. The RNA prepared was reverse-transcribed as described elsewhere 31  All experiments were performed using four replicates and were repeated at least two times, independently. Significance of difference between the treatment groups in the Comet assays was analyzed using the ordinary one-way ANOVA 37 and WST-1 data was analyzed by using the one way ANOVA followed by Bonferroni correction. The comparison of the two groups was analyzed using an unpaired t-test. The differences for both tests with p ≤ 0.05 (*) were considered statistically significant.