Toxicological Evaluation of Piceatannol, Pterostilbene, and ε-Viniferin for Their Potential Use in the Food Industry: A Review

The application of stilbenes in the food industry is being considered because of their biological activities. Piceatannol, pterostilbene and ε-viniferin have awakened the industry’s interest. However, before they can be commercialized, we must first guarantee their safety for consumers. The present work reviews the toxicological studies performed with these stilbenes. A wide variety of studies has demonstrated their cytotoxic effects in both cancer and non-cancerous cell lines. In contrast, although DNA damage was detected by some authors, in vitro genotoxic studies on the effects of piceatannol, pterostilbene, and ε-viniferin remain scarce. None of the three reviewed substances have been evaluated using the in vitro tests required by the European Food Safety Authority (EFSA) as the first step in genotoxicity testing. We did not find any study on the toxic effects of these stilbenes in vivo. Thus, more studies are needed to confirm their safe use before they can be authorized as additive in the food industry.


Introduction
During the last decades, the interest in polyphenolic phytochemicals has increased markedly due to their beneficial properties [1]. Natural polyphenols are abundant in fruits, vegetables, whole grains, and foods and beverages derived from them such as chocolate, wine, olive oil, or tea; thus making it the most important phytochemical present in the human diet [2]. These compounds are highly diversified and comprise several subgroups of phenolic compounds ranging from simple substances, including phenolic acids and stilbenes, to complex polymerized molecules, such as tannins [3].
Natural stilbenes are secondary metabolites produced by plants to protect themselves against stressful conditions such as ultraviolet irradiation, excessive heat and fungal or bacterial infections [2]. Structurally, stilbenes are characterized by the presence of a 1, 2-diphenylethylene nucleus [4] and they can be found in E, or trans, and Z, or cis configurations, the trans form being the one that exhibits more potent pharmacological activities [5,6]. Moreover, these compounds exist as monomers, such as resveratrol, piceatannol, or pterostilbene, and oligomers, like ε-viniferin [1] (Figure 1).
There are more than 400 natural stilbenes reported, however, they are only distributed in a small and heterogeneous group of plants such as wine grape (Vitis vinifera), peanut (Arachis hypogaea), and some tree species (Pinus and Picea genera) because stilbene synthase, the key enzyme involved in stilbene biosynthesis, is not ubiquitously expressed [4]. In general, the highest amount of stilbenes is found in grapes and wine derivatives. However, data related with the available concentrations of these compounds from different sources is very scarce because it depends on the variety of grapes, agricultural and environmental factors (soil, temperature, pathogen attack) and the complexity of the qualitative and quantitative analysis of different stilbenes [7,8]. Moreover, residues produced during variability as a function of the non-cancerous cell line model selected for the test. The toxicity of this stilbene seems to be related to the ortho-dihydroxyl groups on the phenyl ring, also known as catechol. This is in agreement with other authors who stated that the hydroxylation of resveratrol in positions 3' and 4' resulted in increased cytotoxicity [47]. Thus, some authors have reported that the toxic effects of piceatannol are even more potent than those exhibited by trans-resveratrol, pterostilbene, or trans-stilbene-oxide [23,[47][48][49][50][51].
In contrast, very few authors reported tan absence of cytotoxic effects after exposure to piceatannol in different leukemic cell lines at concentrations up to 50 µM after 24 h and 48 h, and up to 100 µM after 48 h of exposure [37,38,48]. Moreover, high concentrations of piceatannol (400 µM) showed a non-cytotoxic effect in murine melanoma cell lines [42].
In order to complete these results, morphological assays were performed by these authors ( Table 2). The results showed that this compound induced apoptosis in a dosedependent manner causing cell shrinkage, chromatin and nuclear condensation, and apoptotic bodies. Low concentrations (1 µM) of the compound can induce spherical apoptotic beads after 48 h of exposure in SK-Mel-28 cancer cells [40]. In contrast, it is interesting to point out that no study has been performed to evaluate the effects of piceatannol in the morphology of non-cancerous cells.
The results of the in vitro cytotoxicity studies carried out with pterostilbene are shown in (Table 3). A comparison between all cytotoxic studies is difficult since the exposure conditions, cell lines, and endpoints differed. In general, most of the authors indicated that this stilbene shows cytotoxic effects in several cell models at different conditions in a range of 25-100 µM. The lowest IC 50 value reported was 1.81 µM in SOSP-9607 cells after 24 h of exposure measured by the MTT assay [49].
On the other hand, although the IC 50 values for non-cancerous cell lines could not always been calculated, a reduction in cell viability was observed after exposure to pterostilbene. The percentage of cell viability of Chang human liver cells was reduced to 75% after exposure to 100 µM of this stilbene after 24 h [50]. Moreover, a very important decrease in cell proliferation was observed in CRL-158 human placenta cells exposed to pterostilbene at concentrations of 40 and 80 µM resulting in reductions of 61.8% and 72.2% as compared to the control [29].
Pterostilbene is expected to be a potent cytotoxic agent since the introduction of one or more methoxy groups into the stilbene structure was previously observed to increase the cytotoxicity of stilbene derivatives [43]. This agrees with the results obtained by several authors comparing the effect of this stilbene with other structurally modified stilbenes, observing that pterostilbene exhibits more potent effects than resveratrol, piceatannol, trans-3,5,4'-trimethoxystilbene, and 3,5,4'-triacetylstilbene [45,64,68,84,87].
Moreover, the cytotoxicity study of pterostilbene has been completed with several morphological assays ( Table 4). The methods used for this purpose were fluorescence microscopy using acridine orange (AO) and ethidium bromide (EB), staining with 4, 6-diamidino-2-phenylindole (DAPI) or Hoechst 33342, and electron microscopy. Low concentrations of pterostilbene caused morphological changes indicating the induction of apoptosis in different cells. The SOSP-9607 cell line treated for 24 h with 1 µM of pterostilbene showed loss of confluence [49]. Moreover, MCF-7 cells exposed to 5 µM for 24 h suffered shrinkage, membrane and cytoplasmic blebbings and chromatin condensation [72]. Moreover, in the case of pterostilbene, no morphological assays were performed on non-cancer cell lines.           Cytoplasm and cell surface had granular appearances in exposed cells. Autophagic vacuoles and autolysosomes were also found. Induction of acidic vesicular organelles was also reported.  Apoptotic morphological changes, cell shrinkage and cytoplasmic blebbing. H520 cells were more sensitive than NCIH-226 cells.

Pure Stilbene Experimental Model Assays Performed Exposure Conditions Main Results Reference
Trans-ε-viniferin and cis-ε-viniferin         The cytotoxic studies performed with ε-viniferin are reported in Table 5. In general, concentrations ranging from 10-200 µM of ε-viniferin caused a significant decrease in the cell viability of cancer cells in a time-and concentration-dependent manner. Low IC 50 values for trans-ε-viniferin were found in HL-60, HepG2, and AGS carcinoma cell lines with values of 5.6 µM ± 1.4, 7.7 µM ± 0.2, and 9.3 µM ± 0.3, respectively [93]. Moreover, ε-viniferin cytotoxicity in non-cancerous cells has also been demonstrated [30,32,46]. Chowdhury et al. (2005) [32] stated that the 50% cytotoxic concentrations of (-)-ε-viniferin in human oral cell lines HGF, HPC, and HPLF were 111 µM, 146 µM, and 94 µM, respectively, which is of interest since ε-viniferin concentrations of 100-200 µM were used in most of the studies performed. Moreover, only 49.9 µM of this compound was required to inhibit the growth by half in MRC-5 normal human lung cells [46]. Higher concentrations were needed to induce toxicity in the non-transformed human hepatocyte cell line HH4, and the IC 50 values obtained after 24 and 48 h of exposure were 192.7 µM and 177.9 µM, respectively. This compound's lack of cytotoxicity has also been demonstrated in various cancer and non-cancer cell lines (SW480, L1210, K562, HCT116, PC12, HepG2, and Chang cells) [37,[99][100][101]. It is interesting to indicate that, although different exposure times have been evaluated (24-96 h), the absence of toxic effects in some cases may be due the low concentrations studied (10, 30, and 50 µM) [37,99,100].
Furthermore, since ε-viniferin is a resveratrol dimer, it can possess a trans or cis configuration. Moreover, this stilbene is a chiral molecule that can cause dextrorotation (−) and levorotation (+). Most of the authors did not specify which ε-viniferin configuration was evaluated. Among those who reported the configuration, trans-ε-viniferin was the most studied was because it is more stable than the cis configuration. The effects of both isomers have been evaluated by Kim et al., (2002) [94]. Concentrations up to 100 µM of both cis and trans isomers induced similar cytotoxic effects in C6, HepG2, HeLa, MCF-7, and HT-29 cancer cell lines after 70 h of exposure. Moreover, the IC50 values obtained in all cell lines were comparable for both configurations [94]. Furthermore, (−)-ε-viniferin was also selected by several authors, but only Chang et al. (2017) [102] evaluated the cytotoxic effects of (+)-ε-viniferin, hindering the comparison between both configurations.
The morphological changes produced by ε-viniferin have been reported by four authors, as far as we know. The main results of these studies are described in Table 6. After exposure to 100 µM for 24 h, and 95 µM and 130 µM for 48 h different cancer cells (HL-60 and C6) suffered chromatin condensation, nuclear fragmentation and contraction [96,104]. Thus, it seems that a prolonged exposure to this compound does not result in more damage. Moreover, low concentrations of ε-viniferin (30 µM and 60 µM) for 48 and 72 h did not produce apoptotic changes in SW480 and HT144 cancer cell lines [99,103], evidencing that concentrations higher than 60 µM are needed to induce ultrastructural damage. Finally, it should be emphasized that nuclear staining with Hoechst was the only technique performed in these assays and there were no studies evaluating the effect of ε-viniferin in non-cancerous cells.

Genotoxicity in In Vitro Studies Performed with Stilbenes
In general, very few in vitro studies have been performed to investigate the potential genotoxic effects and the DNA damage produced by piceatannol, pterostilbene, or ε-viniferin. In fact, there is no research whose main objective has focused on this aspect. Specifically, only 11, 10, and 3 studies of piceatannol, pterostilbene and ε-viniferin, respectively, are related to this topic ( Table 7).
The Guidance for submission for food additive evaluations of the EFSA Panel on Food Additives and Nutrient Sources added to Food [15,116] reported that the mutagenic and genotoxic potential of new additives must be assessed in view of the adverse consequences of genetic damage to human health. To address genotoxicity studies, EFSA guidelines indicate two mandatory tests for all food additives, the Ames test and the in vitro mammalian cell micronucleus test. These tests meet the basic requirements to cover the three genetic endpoints with the minimum number of tests.   The stilbene exhibited an antimutagenic effect in a dose-dependent manner against the mutations produced by N-methyl-N-nitro-N-nitrosoguanidine. [94] ε-viniferin contained in grapevine-shoot extract A431 cells Standard and modified alkaline Comet assay ≥25 µg/mL for 1 h An increase in DNA strand breaks was reported. [115] ε-viniferin C6 cells TUNEL assay 95 and 130 µM for 12, 24, and 48 h ε-viniferin produced DNA damage in a dose-dependent manner. Among all the studies conducted with piceatannol, only Makena and Chung (2007) [110] performed one of the two tests required by the EFSA for the evaluation of its genotoxic potential. These authors carried out the Ames test using only one Salmonella typhimurium strain (TA102), out of the 5 strains recommended by the EFSA. They showed a non-mutagenic effect at 50 µg/plate of piceatannol in the presence and absence of metabolic activation (rat liver S9 mix). However, the main objective of their work was not to evaluate the potential genotoxicity of piceatannol, but to demonstrate the antimutagenic effect of this compound against the mutations induced by benzidine at 50, 100, and 200 µg/plate in the TA102 strain. In addition to this work, there are also two reports that use the comet assay to evaluate the DNA damage produced by piceatannol in different cell lines. The comet assay is an efficient tool to measure single and double-strand DNA breaks at the cellular level [85]. Thus, Azmi et al. (2005) [109] stated that piceatannol produced more damage than resveratrol in the DNA of human peripheral lymphocytes at 10, 20, and 50 µM of piceatannol in the presence of Cu (II); however, no data for piceatannol without Cu (II) was reported. On the other hand, the other study only focused on demonstrating the protective effect of this compound. Ovesná et al. (2006) [35] showed a decrease in the DNA damage produced by H 2 O 2 in L1210, K562, and HL-60 cell lines at 1, 2.5, and 5 µmol/L. Moreover, other techniques such as flow cytometry, western blot analysis and electrophoresis have indicated that piceatannol produces DNA damage, electrophoresis being the most widely used assay [32,34,53,60,112,113]. In general, different studies have demonstrated that piceatannol produces fragmentation in a dose-dependent manner in some cell lines such as HL-60, HSC-2 [32], U937 [53], A549, and HepG2 [112] by electrophoresis. To date, no in vitro micronucleus assays have been performed with piceatannol as required by the EFSA to ensure its safety as far as we know.
In relation to genotoxicity and DNA damage studies performed with pterostilbene, different techniques such as the micronucleus test, comet assay, electrophoresis, western blot analysis, and the TUNEL assay have been performed. Rossi et al. (2013) [114] stated that pterostilbene does not produce micronuclei at concentrations of 20, 40, and 80 µM in CHO-K1 cells after 3 h of exposure. Furthermore, they confirmed that this stilbene reduced basal DNA damage present in untreated cells under these same conditions by the comet assay. Moreover, the latter authors observed that 80 µM of pterostilbene can reduce the oxidative damage produced by H 2 O 2 as measured by the comet assay but it did not show a protective effect against the induction of micronuclei produced by H 2 O 2 . Furthermore, antimutagenic effects of pterostilbene against 4-nitroquinoline-N-oxide have been detected by the comet assay at 50 µM [67]. Similar to piceatannol reports, most of the studies performed with pterostilbene used electrophoresis. Different authors have evidenced that pterostilbene can produce DNA fragmentation in different cell lines such as HeLa [74,82], MCF-7 [62,72], PC3 [62], and MOLT4 [69] at different concentrations (from 10 to 200 µM) and exposure periods (from 12 to 48 h). Despite being required by the EFSA, no Ames test studies have been performed with this substance thus far.
Among the three stilbenes studied in this review, ε-viniferin has been the least studied in regard to its genotoxic and DNA-damaging potential. Kim et al. (2002) [94] performed the Ames test in order to evaluate the antimutagenic potential of ε-viniferin. They used the TA100 strain of Salmonella typhimurium, exhibiting its antimutagenic potential at a concentration of 35.2 g/plate. However, no information about the mutagenic potential of the substance was reported for this assay. In addition, more recent studies have demonstrated that ε-viniferin produces DNA damage in the A431 cell line by the comet assay [115] and the C6 cell line by the TUNEL assay [104].
As the results showed, none of the three reviewed substances have been assessed by both in vitro tests (Ames test and micronucleus assay) required by the EFSA as the first step in genotoxicity testing. Moreover, most of these studies have been carried out on cancer cell lines and their main objective was not to study the genotoxic potential of these stilbenes as required by the EFSA for all food additives to ensure consumer safety. In this sense, the DNA damage has been investigated as a possible mechanism of cytotoxicity against cancer cells.
Taking into account these results, we consider it necessary and scientifically relevant to evaluate the performance of the in vitro genotoxicity assays and the DNA damage caused by these stilbenes prior to their use in the food industry.

Toxicological In Vivo Studies Performed with Stilbenes
Studies focused on assessing the toxicity of substances using in vivo models are necessary to guarantee the safety of their use. In this sense, in vivo toxicity studies of piceatannol, pterostilbene, and ε-viniferin in rodents with potential application in the food industry (novel foods, food additives, etc.) are very scarce, and none have fulfilled the assessment required by the EFSA [15,116]. These studies compromise genotoxicity and other toxicity studies such as subchronic and carcinogenicity studies, etc. [103]. It has only been in recent years that studies have been performed to assess the protective effect of these substances against stress and disease in rodents [117][118][119][120].
With respect to piceatannol, as far as we know, only two authors have evaluated its potential toxic effect. Kiliç (2019) [118] showed that albino mice administered a dose of 4 mg/kg/day IP for 7 days did not show significant differences in biochemical parameters such as superoxide dismutase, catalase, and malonyldialdehyde as compared to the control group. There was no observable nuclear signal of rabbit monoclonal antibody against proliferating cell nuclear antigen or hepatic DNA damage in the treated group. With respect to the results of the histological analysis, apoptotic hepatocytes were rarely observed in animals exposed to piceatannol. Moreover, Shi and Fu (2019) [120] showed that 10 mg/kg/day of piceatannol administered orally via gastric gavage did not induce testicular toxicity. Additionally, beneficial effects such as a marked improvement in mRNA-and protein-expression levels of Nrf2 and its regulated genes and proteins were observed in rats.
The first study that investigated the safety profile of pterostilbene was conducted by Ruiz et al. (2009) [121]. They demonstrated that mice exposed to pterostilbene during 28 days at a dose up to 3000 mg/kg/day caused no mortality during the experimental period. Histopathologic examination and evaluation of biochemical parameters also revealed no alterations regarding organ weight or clinical signs. However, the red blood cell number and hematocrit increased after polyphenol administration as compared to the control group (Ruiz et al., 2009). Later, Riche et al. (2013) [122] assessed the toxicity of pterostilbene in mice after IV administration of 30 mg/kg/day for 23 days. Even at this high dose, pterostilbene was found to be pharmacologically safe as its administration was accompanied by no systemic or organ related toxicity. Moreover, these authors evaluated the long-term safety of pterostilbene administration in a randomized double-blind placebo-controlled trial in humans [122]. They reported that daily doses from 100 mg to 250 mg in adults with hyperlipidemia did not produce a significant adverse drug reaction on hepatic, renal, or glucose markers, with pterostilbene being well-tolerated twice daily. The data available in animal and human models suggests that this compound does not have significant toxic effects. However, the existing information is not adequate to justify the positive effects of this compound in humans after prolonged administration beyond the recommended dietary dose [119]. To our knowledge, no in vivo studies about the safety profile of ε-viniferin were described in the scientific literature. In this sense, it is imperative to perform clinical animal research and human trials to address the safety of ε-viniferin after acute and chronic administration prior to its industrial use.
Taking into account all these facts, further research should include study designs aimed to investigate the safety of these stilbenes in in vivo models. More studies are needed which focus on genotoxicity, subchronic, and chronic toxic effects, etc. to portray the comprehensive safety aspects and to reinforce its human relevancy and market prospects.

Conclusions
Considering the increasing interest in stilbenes as additives in the food industry, toxicological assays are needed to assure their safety. The present review describes the available data on the cytotoxic, mutagenic, and genotoxic aspects of piceatannol, pterostilbene, and ε-viniferin. Their cytotoxic effects depend on the cell lines used, assays performed, and exposure conditions. In general, most of the authors stated that these compounds exhibit toxic effects not only in cancer cells but in non-cancer cell lines. Moreover, the DNA damage induced by these compounds has been demonstrated by several methods as a possible mechanism of cytotoxicity. However, the in vitro genotoxic potential of piceatannol, pterostilbene, and ε-viniferin has been poorly studied and no studies following EFSA guidelines were performed. The largest gap in the toxicity assessment of these compounds is the lack of in vivo studies, since most of the authors have evaluated their beneficial properties but have not evaluated their in vivo toxicity. Thus, in order to guarantee the safe use of piceatannol, pterostilbene, and ε-viniferin, more studies are needed such as toxicokinetic, genotoxicity, subchronic, chronic, and carcinogenicity assays, etc. to fulfill the EFSA's recommendations.