Molecular basis of the chemoprotective effects of isothiocyanates

Understanding the chemoprotective mechanisms of ITC is important not only because these compounds block the formation of a wide variety of carcinogen-induced tumours in rodents but also because ITC and their glucosinolate precursors are widely available in human dietary plants and are consumed in considerable quantities⁽Reference Zhang and Talalay⁸⁾. Glucosinolates are relatively biologically inert, but can degrade to a range of bioactive compounds, such as ITC and indoles, on hydrolysis by the plant-based enzyme myrosinase. However, in the human diet, the myrosinase in cruciferous vegetables is often heat-inactivated during cooking, and in this situation glucosinolates can also be hydrolysed less efficiently by the colonic microflora⁽Reference Fahey, Zalcmann and Talalay⁹⁾. Verhoeven et al. reviewed the evidence for Brassica consumption and cancer risk, and reported that 67% of all studies showed an inverse association between total Brassica vegetable intake and risk of cancer at various sites⁽Reference Verhoeven, Goldbohm and van Poppel¹⁰⁾ and a wide range of studies in human subjects, animals and in vitro have confirmed this finding⁽⁵⁾. For instance, cruciferous vegetables have been found to reduce morphological markers of colon cancer risk in dimethylhydrazine-treated rats⁽Reference Arikawa and Gallaher¹¹⁾. Also, a prospective study carried out in Japan to investigate associations between fruit and vegetable consumption and risk of oesophageal squamous cell carcinoma, showed that only cruciferous vegetables were associated with a significantly decrease in risk (hazard ratio per 100 g/d 0·44 95% CI 0·23, 0·82)⁽Reference Yamaji, Inoue and Sasazuki¹²⁾. This reduced cancer risk has been attributed to the ability of ITC to influence the process of carcinogenesis partly by modulation of phase I and II enzymes, induction of apoptosis and cell-cycle arrest⁽Reference Zhang^13, Reference Johnson¹⁴⁾. Inhibition of phase I enzymes is thought to be a preventive measure against chemically induced carcinogenesis. Pro-carcinogens are usually converted into highly reactive intermediates that form adducts with DNA, RNA and protein. These transformations are catalysed by cytochrome P450 enzymes. DNA adducts that persist unrepaired can generate mutations in critical genes such as oncogenes and tumour suppressor genes⁽Reference Hecht^15, Reference Juge, Mithen and Traka¹⁶⁾. A dose-dependent inhibition of cytochrome P450, family 1, subfamily A, polypeptide 1 and cytochrome P450, family 2, subfamily B, polypeptide 1/2 by sulforaphane (SFN) was observed in rat hepatocytes, and also the expression of cytochrome P450, family 3, subfamily A, polypeptide 4 (the major cytochrome P450 enzyme in human liver) was markedly decreased at both mRNA and activity levels⁽Reference Maheo, Morel and Langouet¹⁷⁾.

Induction of antioxidant and detoxifying enzymes by isothiocyanates

Much of the evidence suggests that a decrease in the expression of antioxidant enzymes together with an increase in the production of free radical species might render cells susceptible to permanent damage and initiate the sequence of events leading to cancer (Fig. 1). On the other hand, elevation of phase II enzymes and other antioxidant systems in specific tissues exposed directly to bioactive food components, such as the colon, may confer cytoprotection against the toxicity of electrophiles and reactive oxygen species⁽Reference Talalay, Fahey and Holtzclaw¹⁸⁾. These groups of functionally diverse phase II enzymes include glutathione transferases, UDP-glucuronosyltransferases and NAD(P)H:quinone oxidoreductase 1. The modulation of phase II gene expression and enzyme activity by ITC has been assessed in a number of cell lines of different origin, but the most commonly used are derived from liver hepatoma, such as human HepG2 and mouse Hepa1c1c7.

Fig. 1.

Role of oxidative stress and antioxidant systems in a simplified scheme of carcinogenesis. Reactive oxygen and nitrogen species (RONS) are constantly generated in cells as unwanted by-products of aerobic metabolism and this together with the inflammatory microenvironment of the tissue acts as a predisposing factor to multistage carcinogenesis⁽Reference Kumar Kundu, Na, Surh and Surh¹⁹⁾. Although a low physiologic level of RONS is scavenged efficiently by the cellular antioxidant defence system, an imbalance between the generation of RONS and cellular antioxidant capacity generates a state of oxidative stress that contributes to carcinogenesis.

Stimulation of these enzymes is one of the most important components of cellular defence mechanisms and the ability of ITC to promote their activation is now understood to be enabled by the Keap1-NF-E2-related factor-2–antioxidant responsive element pathway. However, the list of detoxifying or antioxidant enzymes stimulated by ITC is not only limited to the phase II group, as the NF-E2-related factor 2–antioxidant responsive element signalling pathway also targets other redox-active proteins and glutathione-related selenoenzymes such as thioredoxin reductase-1 (TrxR1) and gastrointestinal glutathione peroxidase (GPx2), respectively, that will be examined in the following section. Studies have revealed that SFN and its glutathione conjugate were found to increase significantly in a time-dependent manner both UDP glucuronosyltransferase 1 family, polypeptide A1 and glutathione S-transferase α1 mRNA levels in HepG2 (P<0·005) and HT-29 (P<0·05) cells⁽Reference Basten, Bao and Williamson²⁰⁾. In another study, when mouse Hepa1c1c7 cells were incubated with 2·5 μm SFN, an increase in NAD(P)H:quinone oxidoreductase 1 activity was observed, reaching a maximum induction of three-fold over control⁽Reference Matusheski and Jeffery²¹⁾. Also, perfusion of broccoli extracts to human jejunum (equivalent to about 1·2 g dry weight broccoli) resulted in an induction of glutathione S-transferase α1 and UDP glucuronosyltransferase 1 family, polypeptide A1 in exfoliated enterocytes; the changes in gene expression were also confirmed in Caco-2 cells, where SFN was responsible for the induction of glutathione S-transferase α1 (three-fold)⁽Reference Petri, Tannergren and Holst²²⁾.

Cell-based models have been used extensively for screening and detection of novel cancer chemopreventive agents from food compounds. However, the induced level and type of phase II enzyme varies between different cell lines as reported in a study that used seven widely adopted cell lines, including Hepa1c1c7, HepG2, MCF-7, MDA-MB-231, LNCaP, HeLa and HT-29, where the effects of 25 μm SFN on the enzymatic activity of glutathione transferase, NAD(P)H:quinone oxidoreductase 1, aldo-keto reductase and glutathione reductase were evaluated⁽Reference Jiang, Chen and Yang²³⁾.

This tissue-specific response has been confirmed in vivo after analysing the ability of six plant-derived ITC (allyl-ITC, iberverin, erucin, SFN, iberin and cheirolin) to increase tissue levels of NAD(P)H:quinone oxidoreductase 1 and glutathione transferase in a variety of rat tissues at doses of 40 μmol/kg per d. The results revealed different levels of enzyme induction in the duodenum, forestomach and urinary bladder out of fifteen tissues analysed, and such responses differed depending on the ITC employed⁽Reference Munday and Munday²⁴⁾. These differences observed at the cell and tissue level may be explained by the degree at which different ITC accumulate in cells⁽Reference Zhang and Talalay²⁵⁾.

Selenium and cancer prevention heterogeneity: metabolism and antioxidant functions

Although dietary Se intake was shown to be inversely associated with cancer mortality as early as the 1960s⁽Reference Shamberger and Frost²⁶⁾, it was not until 1996 that an intensive effort was launched to try to understand the mechanism of action of Se as a cancer-preventive agent after Clark et al.⁽Reference Clark, Combs and Turnbull²⁷⁾ reported results from The Nutritional Prevention of Cancer study, where they showed that supplementation with selenised yeast decreased cancer incidence by almost 50%. However, further trial results have been contradictory and heterogeneous among organs⁽Reference Rayman²⁸⁾ and as a consequence the American Institute for Cancer Research and the World Cancer Research Fund have concluded in its latest report that there is limited evidence that food containing Se protects against colorectal and stomach cancer. Nevertheless, for other sites, such as prostate, such foods probably do decrease risk of cancer⁽¹⁾. Another disappointing outcome arose from the Selenium and Vitamin E Cancer Prevention Trial, which demonstrated that Se (200 μg/d from l-selenomethionine (SeMet)), vitamin E, or Se+vitamin E did not prevent prostate cancer in healthy population groups of men in the study and also had no effect on secondary endpoints, which included lung cancer and colorectal cancer⁽Reference Lippman, Klein and Goodman²⁹⁾.

Nonetheless, there is extensive evidence that monomethylated forms of Se such as methylselenol are critical metabolites for the chemopreventive effect of Se⁽Reference Ip, Thompson and Zhu³⁰⁾. The major forms of Se occurring in food are the organic forms SeMet (found in plants and animal sources) and selenocysteine (found in animal sources). SeMet can also be metabolised via the multi-step trans-sulfuration pathway to selenocysteine, in turn degraded to hydrogen selenide for subsequent methylation by methyltransferases that give rise to methylselenol (Fig. 2)⁽Reference Ip, Dong and Ganther^31, Reference Fairweather-Tait, Bao and Broadley³²⁾. In this respect, Se-methylselenocysteine (SeMSC; present in plants of the Brassica family) and γ-glutamyl-Se-methylseloncysteine (present in plants of the Allium family), which are known to be converted to methylselenol (Fig. 2)⁽Reference Dong, Lisk and Block³³⁾, have been found to account for the anti-carcinogenic effect of Se-enriched broccoli and garlic, respectively, and have proved to be more effective in reducing colon and mammary tumorigenesis in rodents than selenate, selenite or Se-enriched yeast, which contains mostly SeMet⁽Reference Ip, Birringer and Block^34, Reference Finley, Davis and Feng³⁵⁾.

Fig. 2.

Pathways of Se metabolism. Se salts undergo reductive metabolism with glutathione reductase and NADPH or can be directly reduced by thioredoxin reductase (TrxR1) to hydrogen selenide (H₂Se). Selenomethionine and selenocysteine can generate methylselenol through H₂Se. The latter is the source for incorporation into selenocysteine-containing proteins by co-translational incorporation of selenocysteine by seryl-transfer RNA at UGA codons. Diet may also enrich the methylselenol pool by-passing the H₂Se pool, by using direct precursors such as Se-methylselenocysteine (SeMSC) or γ-glutamil-Se-methylseloncysteine, which is converted to SeMSC first and then acted upon by a β-lyase to also give methylselenol. Methylation of H₂Se constitutes an excretion pathway for Se⁽Reference Combs, Lu, Hatfield, Berry and Gladyshev^7, Reference Brigelius-Flohé, Banning and Surh³⁶⁾.

This protective effect attributed to Se could be explained by its multiple cellular functions including cell cycle regulation, immune surveillance, apoptosis, cancer cell migration and angiogenesis⁽Reference Jackson and Combs³⁷⁾. Moreover, some of its associated chemopreventive properties may relate to the crucial fact that Se is a component of the amino acids selenocysteine and SeMet, which are incorporated into twenty-five genes encoding over thirty mammalian selenoproteins⁽Reference Kryukov, Castellano and Novoselov³⁸⁾. Selenoproteins that might be relevant to cancer risk include glutathione peroxidase-1 (GPx1), GPx2, Sep15, SelP and TrxR1⁽Reference Bellinger, Raman and Reeves³⁹⁾.

Considering that different forms of Se supplementation, either as inorganic Se (sodium selenite) or as organic forms such as SeMSC may have different efficacies in the expression of different selenoproteins and that ITC have been found to promote the induction of TrxR1 and GPx2 due to the presence of an antioxidant responsive element in their gene promoter, we have evaluated the effects of the ITC SFN and iberin, which contains one alkyl group less than SFN, together with different forms of Se (selenite and SeMSC) on the expression of these selenoproteins in Caco-2 cells to establish whether a combination of ITC+Se offer additive or synergistic effects on their regulation. Our results have shown that ITC in combination with either form of Se induced more protein expression of TrxR1 and GPx2 than either compound did individually⁽Reference Barrera⁴⁰⁾. The importance of these findings relates to the fact that in the pre-initiation and/or early stages of colon carcinogenesis the disruption of cellular defence mechanisms (consisting of a battery of detoxifying or antioxidant enzymes) would make cells more susceptible to DNA damage by both unwanted by-products of normal cellular metabolism (Fig. 1), or other environmental sources of reactive oxygen species, which in rapidly dividing cells, such as in the colonic epithelium, may escape repair mechanisms resulting in somatic mutations⁽Reference Stone, Krishnan, Papas, Fuchs, Podda and Packer⁴¹⁾. Therefore, an increase in the cellular antioxidant defence mechanisms may prevent the deleterious effects of free radicals that would otherwise affect important biomolecules and render colon cells susceptible to the accumulation of genetic alterations that lead to cancer (Fig. 1). It has been previously suggested that the up-regulation of these selenoproteins by ITC and Se results from an independent mechanism. While the former works transcriptionally through the antioxidant responsive element located in the promoter region of the genes, the latter acts by post-transcriptional mechanisms that involve the provision of an adequate supply of selenocysteine for incorporation into TrxR1 or Gpx2 that delays its degradation⁽Reference Zhang, Svehlikova and Bao^42–Reference Brigelius-Flohe and Banning⁴⁴⁾. Similar results in relation to the up-regulation of TrxR1 by co-addition of different forms of ITC and Se have been observed in other studies that have employed different cell lines⁽Reference Zhang, Svehlikova and Bao^42, Reference Campbell, Howie and Arthur⁴⁵⁾.

Nevertheless, data emerging from in vivo and in vitro studies suggest that TrxR1 and GPx2 are up-regulated in cancer. These observations provide evidence that TrxR1 and GPx2 are critical for self-sufficiency in growth of malignant cells, in which these selenoproteins act predominantly as pro-cancer proteins⁽Reference Yoo, Xu and Carlson^46, Reference Lin, Furukawa and Tsunoda⁴⁷⁾. In addition to its role in the activation of antioxidant and detoxifying enzymes, NF-E2-related factor 2 has been suggested to have a positive role in cancer tumorigenesis and chemoresistance⁽Reference Rushworth, Bowles and MacEwan⁴⁸⁾. Thus, ITC+Se-mediated increase in selenoprotein expression, such as TrxR1 and GPx2, may be more relevant in normal cells as a cancer-preventive measure to inhibit malignant transformation prior to the initiation of carcinogenesis. However, a long-term exposure of dietary ITC and/or Se might potentially lead to other modifications to the human genome that will ultimately impact upon gene expression and cancer susceptibility. With these possibilities in mind we have investigated whether modifications to epigenetic marks may be a potential mechanism whereby these dietary constituents inhibit the development of neoplastic colonic lesions.

Dietary modulation of epigenetic modifications and cancer

Epigenetics refers to heritable changes in the pattern of gene expression without any alteration in the nucleotide sequence⁽Reference Esteller and Esteller⁴⁹⁾. In human subjects and other mammals the most frequent epigenetic modification of DNA bases involves cytosine, which is modified reversibly by adding a methyl group (CH₃) to its C₅ position. This modification occurs only on cytosines that precede a guanosine in the DNA sequence, referred to as the CpG dinucleotide⁽Reference Herman and Baylin⁵⁰⁾. Short regions of 0·5–4 kb in length, known as CpG islands, are rich in CpG content. These islands are typically found in or near promoter regions of genes where transcription is initiated. In normal somatic cells, the vast majority of CpG dinucleotides in the genome are methylated, whereas CpG islands often remain unmethylated, allowing gene expression to occur. In cancer cells, this pattern of CpG methylation becomes disrupted, with high level of methylation within promoter regions of genes causing abnormal gene silencing, in addition to global hypomethylation of genomic DNA, which promotes chromosomal instability, translocation and gene disruption through the reactivation of endoparasitic sequences⁽Reference Esteller⁵¹⁾. Furthermore, the state of histone acetylation is also important in regulating chromatin structure and gene transcription. Acetylation of histones is required to maintain chromatin in an open and transcriptionally active state. On the other hand, histone deacetylases (HDAC) act to keep histone residues deacetylated, promoting transcriptional silencing.

There is currently considerable interest in the development of effective and non-toxic inhibitors of DNA methyl transferase (DNMT) not only for therapy but also for chemoprevention. In this regard, some chemical components of edible fruits and vegetables are promising chemotherapeutic agents that have been widely used in an attempt to reverse abnormal DNA hypermethylation patterns in cancer, and restore the expression of silenced genes.

Epigenetic modulation of gene expression by dietary isothiocyanates

Despite the fact that most work on the cellular effects of ITC relates to their influence on detoxifying enzymes, recent data support the effect of these dietary compounds on the reactivation of epigenetically silenced genes in cancer cells, particularly through the inhibition of HDAC activity⁽Reference Dashwood and Ho⁵²⁾. The first in vitro study reporting these changes was conducted using SFN on prostate and colon cancer cell lines, where the inhibition of HDAC activity was accompanied by global increases in histone H3 and H4 acetylation on the promoter regions of p21 and bax genes, facilitating cell-cycle arrest and apoptosis in the context of cancer chemoprevention⁽Reference Myzak, Karplus and Chung⁵³⁾. Later, the same group confirmed HDAC inhibition by SFN in vivo, using the adenomatous polyposis coli multiple intestinal neoplasia mouse model that had ingested 443 mg SFN/kg (about 6 μmol SFN/d) for 70 d, and observed the re-expression of p21 and bax genes that triggered cell-cycle arrest and apoptosis in transformed cells and microadenomas, thereby suppressing polyp formation compared with controls⁽Reference Myzak, Dashwood and Orner⁵⁴⁾. Although the growing interest in the epigenetic regulation mediated by ITC has focused mainly on its HDAC inhibitory activity, their potential chemopreventive mechanisms involving DNA methylation mechanisms remain relatively unknown.

Studies carried out in a human myeloma cell line have found that phenylhexyl ITC induced histone H3 hyperacetylation and demethylation of the aberrantly methylated p16 promoter in a concentration-dependent manner, suggesting that phenylhexyl ITC has dual epigenetic modulating effects on both DNA methylation and chromatin⁽Reference Lu, Lin and Feng⁵⁵⁾. Comparable results were obtained in prostate cancer cell lines, where phenethyl ITC inhibited the activity and level of histone deacetylases and promoted glutathione S-transferase pi 1 promoter demethylation (dual action)⁽Reference Wang, Beklemisheva and Liu⁵⁶⁾.

More recently, the involvement of ITC on DNMT has been confirmed in a study in which SFN was found to inhibit DNMT in breast cancer cells. Meeran et al. showed that SFN treatment dose- and time-dependently inhibited human telomerase reverse transcriptase, the catalytic regulatory subunit of telomerase, in both MCF-7 and MDA-MB-231 human breast cancer cells and that it had insignificant effects on normal control cells. Furthermore, DNMT protein expression (particularly DNMT1 and DNMT3A), was also reduced in SFN-treated breast cancer cells. Additionally, site-specific CpG demethylation was observed primarily in the first exon of the TERT gene. This facilitated binding of CCCTC-binding factor, which is associated with TERT repression, leading to cellular apoptosis of the breast cancer cells⁽Reference Meeran, Patel and Tollefsbol⁵⁷⁾.

Role of selenium in DNA methylation changes

Numerous studies have reported an effect of ITC on the methylation status of genes involved in the cancer process in different cell lines and animal models. While many of these studies have focused particularly on SFN others have also focused on Se. This mineral, apart from being an important player in Se-anti-carcinogenesis by way of its intermediary Se-metabolites, or as an essential component of antioxidant enzymes that are active in the removal of reactive oxygen and nitrogen species, has been found to affect DNA methylation by interfering with DNMT activity⁽Reference Davis and Uthus⁵⁸⁾. However, previous animal and in vitro studies investigating the impact of Se on DNA methylation have been inconsistent. For instance, rats fed Se-deficient diets had global DNA hypomethylation in the colon and liver⁽Reference Davis, Uthus and Finley^59, Reference Davis and Uthus⁶⁰⁾. In contrast, in vitro studies have shown that exposure of human colon carcinoma HCT116 cells to phenylenebis-(methylene)-selenocyanate (another form of Se), sodium selenite and benzyl selenocyanate for 24 h caused DNMT inhibition⁽Reference Fiala, Staretz and Pandya⁶¹⁾. Also, reports from other investigators have indicated that dietary Se can inhibit DNMT1 activity in vitro from rat liver and Friend erythroleukaemic cells⁽Reference Cox^62, Reference Cox and Goorha⁶³⁾. Furthermore, treatment of LNCaP prostate cancer cells with this trace mineral was found to cause an induction of GSTP1, APC and CSR1 promoter demethylation (through a decrease in the protein levels of DNMT1), in addition to a reduction in HDAC activity, leading to increased acetylation together with decreased methylation levels of lysine 9 on histone H3, which promoted gene expression⁽Reference Xiang, Zhao and Song⁶⁴⁾.

In order to clarify discrepancies between in vitro and in vivo studies reported previously Pilsner et al. studied the association between plasma Se concentration and genomic methylation of leucocyte DNA in a cross-sectional study of 287 Bangladesh adults to test the hypothesis that Se deficiency is linked with genomic DNA hypomethylation. The study results were consistent with earlier in vitro findings (i.e. Se is inversely associated with genomic DNA methylation), suggesting that high levels of Se are associated with a decrease in DNMT expression and/or activity⁽Reference Pilsner, Hall and Liu⁶⁵⁾.

However, the efficacy of different forms of Se, particularly SeMSC, in relation to cancer prevention through DNA methylation mechanisms has not been investigated in sufficient depth. To further expand this knowledge, we examined the effect of SeMSC or selenite (ranging from 0·2 to 5 μm) either individually or in combination with the ITC SFN or iberin (ranging from 6 to 8 μm) for up to 12 d on factors modulating gene-specific (p16 ^INK4A, ESR1), global (long interspersed nuclear element-1) methylation (a surrogate marker of genome-wide methylation), and DNMT expression, in two colorectal cancer cell lines namely Caco-2 and HCT116. However, none of the compounds assessed influenced the methylation status of the genes studied, nor did we find changes in the 5-methylcytosine content of the genome⁽Reference Barrera⁴⁰⁾.

Deregulation of methylation patterns is a common characteristic in tumour cells observed in almost all types of cancer. Several previous studies suggest that diet-derived factors offer potential for the prevention and therapy of a wide variety of cancers by altering various epigenetic modifications. However, the literature and our results suggest that, while some tissues may respond effectively to particular food compounds by impacting on levels of gene-specific methylation, others do not respond.

Also, even within the same gene, exon-specific DNA methylation patterns have been found to be affected differently by dietary compounds. For instance, folate has been found to modify the level of methylation within exons 6 and 7 of the colonic p53 but not that of exon 8⁽Reference Kim, Pogribny and Salomon⁶⁶⁾. Although for some genes it has been suggested that methylation occurs in an ‘all-or-none’ manner⁽Reference Belshaw, Elliott and Williams⁶⁷⁾, it is also apparent that particular cytosines within a CpG island can have a distinctly greater likelihood of being methylated⁽Reference Mund, Beier and Bewerunge^68, Reference Feltus, Lee and Costello⁶⁹⁾ than others. Thus, despite significant progress in understanding cancer chemoprevention through dietary agents, much remains to be elucidated about the effects of ITC and Se on epigenetic alterations and antioxidant enzyme expression at different stages of colorectal carcinogenesis, and about their importance as mechanisms by which diets may modulate gene expression and attenuate (or in some cases exacerbate?) cancer progression.