Polyphenols and DNA Damage: A Mixed Blessing

Polyphenols are a very broad group of chemicals, widely distributed in plant foods, and endowed with antioxidant activity by virtue of their numerous phenol groups. They are widely studied as putative cancer-protective agents, potentially contributing to the cancer preventive properties of fruits and vegetables. We review recent publications relating to human trials, animal experiments and cell culture, grouping them according to whether polyphenols are investigated in whole foods and drinks, in plant extracts, or as individual compounds. A variety of assays are in use to study genetic damage endpoints. Human trials, of which there are rather few, tend to show decreases in endogenous DNA damage and protection against DNA damage induced ex vivo in blood cells. Most animal experiments have investigated the effects of polyphenols (often at high doses) in combination with known DNA-damaging agents, and generally they show protection. High concentrations can themselves induce DNA damage, as demonstrated in numerous cell culture experiments; low concentrations, on the other hand, tend to decrease DNA damage.


Introduction
For many years now it has been recognised that fruits and vegetables play an important role in preventing or alleviating the effects of various chronic diseases, notably cardiovascular disease and various cancers. The mechanism(s) of this protection is still not clear. A common explanation is the so-called antioxidant hypothesis; oxidative stress is a factor in many diseases; fruits and vegetables contain various phytochemicals with antioxidant properties, and so these are likely to be the agents of protection. This is clearly a simplistic hypothesis; phytochemicals have been shown to have a wide array of influences on the physiological processes of human cells, and reducing them to sources of antioxidant activity is misguided and misleading. A meta-analysis of clinical trials indicates that antioxidant phytochemicals taken as supplements have no beneficial effect on mortality and may even increase it [1]. In natural plant foods, of course, phytochemicals of different kinds are present, acting in concert, often in all likelihood synergistically, and so studies of whole foods or extracts are particularly valuable. The reductionist approach (looking at individual components) is still popular, however, as evidenced by the large number of studies of individual phytochemicals, and by the growing catalogue of plant species that have been extracted and tested for potential health-promoting effects using a range of molecular markers. DNA damage is one of the most commonly employed such markers, in the reasonable belief that a decrease in DNA damage-as the initiating event of carcinogenesis-must signify a decrease in cancer risk.
Currently, the most popular assay for DNA damage at the cellular level is single cell gel electrophoresis, or the comet assay [2]. It is based on the ability of a strand break (SB) to relax

Methods
In this review, we have concentrated on papers published from 2010 to the present.
We used PubMed with the followings terms in the title or abstract: polyphenols/polyphenol/flavonoids/flavonoid combined with DNA damage/DNA protection/DNA repair. We found a total of 386 papers. We have concentrated on papers where the effect of polyphenols, in the form of real food, plant extract or pure compound, is tested in cell culture, animals and humans. We have excluded papers where only gene expression was studied, papers specifically focused on other diseases than cancer, and papers, for example, with deficient experimental design. Papers in which the main interest is in the induction of apoptosis were also excluded.
The reports are summarised in tables according to whether they deal with whole foods (or drinks) (Table 1), with extracts of plants (Table 2), or with single phytochemicals (Table 3). Studies are further classified as 'in humans', 'in vivo' (animal studies), or 'in vitro' (experiments with cultured cells). Extracts and phytochemicals are, where possible, grouped according to functional, chemical or botanical relationships (such as 'tea and coffee related compounds', or 'flavonoids', or 'Lamiaceae'). We have generally excluded in vitro experiments with plants or compounds appearing in just one or two publications, unless they fall into one of these groups.

Whole Foods and Drinks
Relatively few investigations of effects of whole foods on genetic damage endpoints have been published. A variety of fruit-derived drinks as well as tea (though this could be considered an extract), and dark chocolate, were tested in human supplementation trials. A decrease in urinary 8-OH-G was seen in overweight or obese adults supplemented with orange juice [10] but levels of plasma 8-OH-G in triathletes were too low to see any effect of Aronia-citrus juice [11]. De-alcoholised wine given daily for one month was without effect on DNA SBs or Fpg-sites in peripheral blood mononuclear (PBMN) cells of post-menopausal women [13]. However, a daily blueberry drink taken for 6 weeks protected PBMN cells from H 2 O 2 -induced damage, but had no effect on SBs or DNA repair capacity [14]. Malhomme de la Roche et al. [15] found that ingestion of green tea protected PBMN cells challenged ex vivo with UV(A)/VIS (ultraviolet(A)/visible) radiation, but only in some subjects, described as responders. Alleva et al. [16] gave a honey supplement to humans exposed to pesticides, and found, after two weeks' supplementation, lower levels of EndoIII-and Fpg-sensitive sites in lymphocytes as well as an enhanced capacity for DNA repair. Dark chocolate induced a transient protection against H 2 O 2 -induced DNA damage in PBMN cells ex vivo [12].
Most of the animal studies have looked at the possible protection afforded by polyphenol-rich foods or drinks against DNA damage induced by treating the animals (rats or mice) with known carcinogens such as doxorubicin (Dox), n-nitrosodiethylamine, or sodium arsenite. Protection was claimed with Chrysobalanus icaco fruit [17], Piquia pulp [19], Açai pulp [20], and tea [18,22]; but cloudy apple juice actually increased SBs and had no effect on nitrosamine-induced damage [21]. Treatment of hyperlipidemic rats with spinach increased the resistance of blood cells ex vivo to H 2 O 2 -induced damage [23].
Experiments with cultured cells and whole foods/drinks are understandably rarely performed. Incubation of PBMN cells with green tea decreased DNA damage at low concentrations but increased it at the highest concentration tested (representing 71 mM catechins) [24]. Various honeys afforded slight protection of HepG2 cells against SBs produced by treatment with certain organic carcinogens [26]. A Chinese herbal preparation caused SBs in mouse lymphoma cells and rat fibroblasts, but at extreme concentrations (1-13 mg/mL) [25].

Tea-Related Extracts
One human trial and several animal experiments have been reported with tea-related extracts. Post-menopausal women with osteoporosis were supplemented with green tea polyphenols for 6 months; the level of urinary 8-OH-G decreased [27]. Xu et al. [28] found a decrease in 8-OH-G in rats given a very high dose of green tea polyphenols. Protective effects of green tea extracts against genetic damage were reported by Garcia-Rodriguez et al. [30] in mice treated with Cr(IV); and by Pu et al. [31] in rats treated with acrylonitrile. Katiyar et al. [29] found that green tea polyphenols promoted the repair of UV-induced DNA lesions in mice proficient in nucleotide excision repair (NER), but not in NER-mice. Two studies with cultured cells have found increases in DNA SBs induced by green tea extract; Prasad et al. [35] in melanoma cell lines (though at rather high concentrations), and Durgo et al. [36] in a human laryngeal carcinoma cell line.

Lamiaceae Family Plants
The Lamiaceae family includes many plants used as culinary herbs, and so they have been grouped together here. All publications in our search deal with effects in cell culture.
Calo et al. [39] tested an extract of Thymus vulgaris (and thymol in parallel) on keratinocytes irradiated with UV(A) or UV(B); they found a decrease in SBs, though no effect on MN or γ-H2AX foci. A similar protective effect was reported by Cornaghi et al. [38] in a human skin model exposed to UV(B). A citrus and rosemary extract (but at high concentrations) decreased the frequency of MN induced by X-rays in human lymphocytes, and decreased UV(B)-induced SBs in keratinocytes [37]. This last group also tested lemon balm extract on UV(B)-irradiated keratinocytes and found a decrease in SBs (at a high concentration) and in γ-H2AX foci at a more moderate concentration [40]. Thirugnanasampandan et al. [42] studied three Lamiaceae species; HepG2 cells were incubated for 4 h with an extract before treating with CdCl 2 . Dose-dependent decreases in SBs were seen with all three (though even the lowest concentration tested was high). An extract of Ocimum sanctum (a form of basil) was tested by Venuprasad et al. [41] on human neuroblastoma cells; it protected against H 2 O 2 -induced SBs (at a high concentration).

Honey-Related Extracts
In parallel experiments to their human honey trial, Alleva et al. [16] showed that pre-treatment of cells with honey extract protected against pesticide-induced DNA damage and inhibition of DNA repair. Propolis extract (at high concentration) decreased the frequency of γ-ray-induced SBs in fibroblasts [50], and yet-at a much lower concentration-it caused oxidative damage (SBs measured with Fpg and EndoIII together in the comet assay) in a human cancer cell line, which was suppressed by antioxidants or catalase and so was imputed to the production of H 2 O 2 [51].

Fruits and Berries
All papers on extracts of fruits and berries reviewed here describe cell culture experiments and with one exception they have made use of high to extremely high extract concentrations. The extract of one Australian fruit (among several studied) caused an increase in MN [48]. Other reports are of protection against oxidation damage caused by H 2 O 2 [43,44,46]; or tert-butyl-hydroperoxide (t-BOOH) [45,49]. The exception to usage of high doses is a report by Bellion et al. [47] with apple polyphenol extracts; they found that 24 h pre-incubation of Caco2 cells decreased the DNA damage induced by menadione (low concentrations actually giving the greatest protection).

Miscellaneous Plant Extracts
Animal experiments with various plant extracts have shown protection against SB production in liver cells of pyrogallol-treated rats (at very high doses of extract) [34]; accelerated rejoining of γ-ray-induced DNA SBs [33]; and a decrease in pyrimidine dimers in the skin of UV(B)-irradiated mice [32].
One human trial with epigallocatechin gallate in prostate cancer patients showed no significant effect on 8-OH-G in leukocytes [52]. Animal studies with single polyphenols have generally involved treating mice or rats with a known DNA-damaging agent and looking for protection against DNA breaks, MN and chrom abs. Generally, protection is seen [54,57] though in some cases at rather high doses [56,63]. Pretreatment of rats with epicatechin reduced the level of DNA breaks induced in bone marrow cells by the topoisomerase poison etoposide [55]. High concentrations have also been used in in vitro experiments with cultured cells, and have given increases in SBs and γ-H2AX foci [71] and in γ-H2AX and 8-OH-G [58]. Kumar et al. [79] found that a combination of ellagic acid with curcumin (25 µM each) caused SBs while the separate compounds had no significant effect. A decrease in (background) SBs with epigallocatechin gallate or epicatechin gallate was reported by Durgo et al. [36] at 48 but not 72 h. At more reasonable concentrations, the results are mixed: decreases in UV(B)-induced SBs [70] and cyclobutane pyrimidine dimers [72]; decreases in H 2 O 2 -induced SBs and MN [74]; a decrease in SBs induced by cumene hydroperoxide [75]; but SBs and Fpg-sites increased in tetradecanoyl phorbol acetate (TPA)-stimulated neutrophils [78] and an increase in SBs with ellagic acid in prostate cancer cells was reported by Vanella et al. [73] at concentrations of 9 µM in one of the cell lines but higher concentrations in two other lines.

Curcumin
Curcumin was examined alongside epicatechin by Papiez [55]; at high concentration (up to 0.2 g/kg/day), it decreased DNA damage in the bone marrow of rats treated with etoposide. In cultured cells, curcumin at rather high concentrations caused SBs [85,86] and γ-H2AX foci [83]. The production of SBs in combination with ellagic acid was noted above [79]. Lewinska et al. [81] reported a pro-oxidant effect of curcumin at concentrations of 10 µM (and below), indicated by an increase in 8-oxo-G in smooth muscle cells. Sebastia et al. [80] compared effects of curcumin on human lymphocytes, both stimulated by TPA and unstimulated, and γ-irradiated. In non-cycling cells, the phytochemical was radioprotective (decreasing the level of premature chromosome condensation), whereas in cycling cells it acted as a radiosensitiser, increasing the frequency of chrom abs.

Resveratrol
At high concentration, in human lymphocytes, resveratrol decreased the frequency of chromosome aberrations caused by aflatoxin [87]. At a lower concentration, it protected human epithelial cells against SBs and MN induced by sodium arsenite [88], and rat astrocytes against SBs caused by ethanol [91]. However, a low dose enhanced the frequency of γ-H2AX foci after ionising irradiation of prostate epithelial cells [90]. A moderately high concentration applied to colon cancer cells caused γ-H2AX foci, apparently as a result of topoisomerase II poisoning [89]. SBs as well as Fpg-sites were increased in non-cycling cells but decreased in TPA-stimulated, cycling cells [78]. In contrast, Sebastia et al. [80] found that, as with curcumin, effects of resveratrol on irradiated lymphocytes differed depending on whether the cells were non-cycling (showing a decrease in premature chromosome condensation), or cycling (in which it had the opposite effect, acting as a radiosensitiser, increasing chromosome aberrations).

Flavonoids
Kozics et al. [97] performed a useful comparative study of 10 flavonoids, concluding that their effectiveness at protecting against B(a)P-induced SBs and MN depended on their chemical structure. Tested over a relatively low concentration range, fisetin, quercetin, galangin, kaempferol and luteolin (in order of decreasing effectiveness) were more effective than chrysin,7-hydroxyflavone, 7,8-dihydroxyflavone or baicalein, while rutin was without effect.
Quercitrin, the rhamnose glycoside of quercetin, protected mouse epidermal cells against UV(B)-induced SBs [102]. Rutin at low concentrations showed the same protective effect as quercetin on HepG2 cells treated with AFB1, MMS or Dox [100]; at much higher concentrations, it caused SBs, but still protected against MN induced by B(a)P [93].

Discussion and Conclusions
Many of the papers that we have reviewed report experiments with high or very high concentrations of phytochemicals. When investigating the role of phytochemicals in normal human nutrition, the aim should always be to study concentrations close to those likely to be present in humans as a result of dietary intake. As a rule of thumb, we have assumed this concentration to be in the low micromolar range. Many papers quote concentrations in µg/mL. To convert these concentrations to micromolar, again as a rule of thumb, we have assumed a molecular weight of 500; then 1 µg/mL = 2 µM. We would regard a concentration of over 20 µM or 10 µg/mL as high, and over 50 µM or 25 µg/mL as very high. Clearly, in functional foods or phytochemical supplements, the concentration is likely to be higher than in natural foods, and experiments showing genotoxicity of phytochemicals at high doses should at least serve as a warning to designers of functional foods.
It is always instructive to carry out experiments over a range of concentrations. Often, in the case of micronutrients in general, the dose-response curve is U-shaped, i.e., a beneficial effect at low concentrations changes to a detrimental effect at higher concentrations, and this tendency is clear in many of the reports described here.
Of course, if genotoxicity is specifically directed to cancer cells while healthy cells are unaffected, it is regarded as beneficial, and it is evidently the aim of some of the papers that we have reviewed to identify plant extracts or particular polyphenols that have such targeted action and so might have potential value as therapeutic agents. The differential response of cycling vs non-cycling cells to certain polyphenols might be exploited therapeutically in targeting dividing cancer cells.
With such a wide-ranging set of phytochemicals, not to mention the variety of test systems, experimental designs and assays applied in their study, it is difficult to generalise. However, high concentrations are likely to show DNA-damaging effects, while also in many cases protecting cells against damaging effects of other agents, apparently acting as pro-oxidants when present alone, but as anti-oxidants in combination. This is not a novel observation: many years ago, Duthie et al. reported DNA-damaging effects of quercetin at 50 µM [103] alongside an ability to protect cells against H 2 O 2 -induced DNA damage at concentrations of 10-50 µM [104]. Low concentrations are generally protective, in some cases even decreasing the already low background level of cellular DNA damage.
To summarise, results reported in the recent literature, on the whole, lend support to the hypothesis that dietary polyphenols protect the body against the effects of reactive oxygen species on DNA integrity, but do so reliably only when present at low concentrations. We recommend that greater attention be paid to the concentrations used, particularly in in vitro experiments, if the results are to be extrapolated to issues of human health. An important consideration when extrapolating is that plant foods contain a variety of micronutrients which might be expected to act in concert, whereas most experiments are carried out with single compounds. In this respect, there are clear advantages in using plant extracts or whole foods, though this approach does present practical difficulties. We also recommend that, since oxidative damage to DNA, and its prevention, are of major concern, the modified comet assay incorporating Fpg or EndoIII should be employed, since it provides increased sensitivity and specificity.