Evidence for a Role of Oxidative Stress in the Carcinogenicity of Ochratoxin A

The in vitro and in vivo evidence compatible with a role for oxidative stress in OTA carcinogenicity has been collected and described. Several potential oxido-reduction mechanisms have been identified in the past. More recently, the possibility of a reduction of cellular antioxidant defense has been raised as an indirect source of oxidative stress. Consequences resulting from the production of oxidative stress are observed at different levels. First, OTA exposure has been associated with increased levels of oxidative DNA, lipid, and protein damage. Second, various biological processes known to be mobilized under oxidative stress were shown to be altered by OTA. These effects have been observed in both in vitro and in vivo test systems. In vivo, active doses were often within doses documented to induce renal tumors in rats. In conclusion, the evidence for the induction of an oxidative stress response resulting from OTA exposure can be considered strong. Because the contribution of the oxidative stress response in the development of cancers is well established, a role in OTA carcinogenicity is plausible. Altogether, the data reviewed above support the application of a threshold-based approach to establish safe level of dietary human exposure to OTA.


Introduction
Ochratoxin A (OTA) is a mycotoxin produced by several food-borne species of Aspergillus and Penicillium fungi. OTA occurs in various food materials and therefore humans are continuously exposed to relatively small amounts of it. Because of its wide occurrence and consequent exposure, together with a potent carcinogenic potential in animal models, OTA has attracted significant public health attention over the last few years. Several national and international food safety organizations and expert groups have conducted a thorough review of the situation as well as risk assessments in order to provide an insight on the health significance of OTA in food.
Human epidemiology has been inconclusive: a number of studies have suggested a correlation between exposure to OTA and Balkan Endemic Nephropathy (BEN) and mortality from urinary tract tumors [1,2]. Epidemiological data were recently reviewed by several expert groups [3][4][5][6]. All concluded that causality between intake of OTA and human nephropathy could not be established. Therefore, the IARC statement that there is inadequate evidence for carcinogenicity in humans (group 2B) [7] appears still valid. Recently, other nephrotoxic agents have been put forward as the primary cause of BEN [6,8,9].
In absence of adequate human data, risk assessments have relied on animal data. Kidney has been considered as the key target organ of OTA toxicity. In all animal species studied, OTA was found to produce renal toxicity, while in rodents renal carcinogenicity was clearly established. Recently, OTA renal and hepatic carcinogenicity was also observed in chicks [10]. Using a LOAEL of 8 mcg/kg bw/day based on early markers of renal toxicity in pig (the most sensitive animal species) and applying an uncertainty factor of 450, EFSA [4] allocated a Tolerable Weekly Intake (TWI) of 120 ng/kg bw. Analysis of dietary exposure throughout Europe revealed that the current average OTA exposure (50-60 ng/kg bw/week) is well within the TWI [4]. The joint FAO/WHO Expert Committee on Food Additives (JECFA) first evaluated OTA at their 37th meeting [11]. Based on the LOAEL in pig, and applying an uncertainty factor of 500, JECFA allocated a Provisional Tolerable Weekly Intake (PTWI) of 112 ng/kg bw [11]. This value was rounded to 100 ng/kg bw/week and confirmed in several subsequent meetings [12,13].
In 2008, JECFA applied a benchmark dose (BMD) modeling approach using carcinogenicity data [14]. The BMD is the dose estimated to cause a predefined increase (e.g., 10% for the BMD10) in tumor incidence over background. The BMDL is the lower limit of a one-sided 95% confidence interval of the BMD. The occurrence of combined adenomas and carcinomas in the kidneys of male rats was considered by JECFA to be the most appropriate data for modeling. Values of 18-33 mcg/kg bw/d and 15-25 mcg/kg bw/day were calculated for, respectively, the BMD10 and BMDL10. Because the BMD approach did not provide a lower point of departure than the LOAEL in pig, JECFA decided to reconfirm the PTWI of 100 ng/kg bw/day [14].
A recent health risk assessment, performed by Health Canada [15] recommends to regulate OTA as a nonthreshold carcinogen, because of the uncertainties regarding the mode of action. The authors defined a negligible cancer risk intake (NCRI, risk level 1 : 100 000) using the tumorigenic dose at which 5% of rats are likely to develop tumors (TD 05 , derived through modeling) as point of departure. Importantly, there is considerable convergence between the NCRI established in this assessment and the TDI derived by EFSA.
In the risk assessment of carcinogenic substances, consideration of the mode of action (MOA) is essential, determining the method to be applied in order to define levels of exposure below which a low safety concern is expected.
The key events analysis framework of the MOA has not yet been formally applied to OTA. However, the approach used by most expert groups (EFSA, JECFA, ILSI) to establish the safe level of exposure of OTA (based on uncertainty factors) implies the consideration of key events compatible with a threshold effect. For these groups, amongst the mechanisms of action highlighted as possible, oxidative stress has been presented as one of the most probable [5,14].
1.1. Scope of the Paper. Over the last decades, studies aimed at elucidating the modes of action implicated in OTA toxicity and carcinogenicity have been published [16]. There has been considerable debate for many years over the genotoxicity of OTA and its actual role in carcinogenicity [3-5, 11-15, 17, 18].
Although genotoxicity is likely to play a role in OTA carcinogenicity [3,4], the actual molecular mechanism involved, either through covalent adduct formation, through other indirect modifications, or both is still unclear. The potential of OTA to form covalent DNA adducts has been subjected to debate due to conflicting data in the literature. Using 32 P-postlabelling analysis, large numbers of putative OTA-derived DNA adducts have been reported to be present in a wide range of tissues from OTA-treated rats, mice as well as pigs [2,17,[19][20][21][22]. However, so far, these adducts have never been observed by any other highly specific techniques such as radioactivity measurements using 3 H-labelled OTA ( 3 H-OTA) [23], accelerator mass spectrometry (AMS) [24], or liquid chromatography-tandem mass spectrometry (LC-MS/MS) [25].
The present paper is not intended to provide a thorough review of the complex and controversial scientific literature on DNA-adduct formation. However, it is important to keep in mind that DNA adducts are increasingly considered as markers of exposures and not only of effects [26,27] and that DNA-covalent binding does not necessarily determine the shape of the dose-response at low level of exposure [28,29]. According to Mantle and coworkers [21,30], experimental dose-response data for OTA's renal carcinogenesis makes a compelling case for OTA being a thresholded carcinogen in male rat. In this context, it appears important to also consider other potential modes of action, which could potentially contribute to OTA carcinogenicity. In the near future, the application of the mode of action framework [28] will likely help to understand the individual contribution of all mechanisms described up to now for OTA.
The focus of the present short paper was to collect and highlight the evidence associated with a role for oxidative stress as a plausible mechanism to consider for OTA. A list of the studies used to illustrate the main messages of the present paper is provided in Table 1. Although not exhaustive, the list shows that over the last two decades, numerous investigators have documented the generation of oxidative stress as a result of OTA treatment in both in vitro and in vivo model systems.

Sources of OTA-Mediated Oxygen-Species Generation
Production of reactive oxygen species (ROS) leading to oxidative stress and macromolecular damage is known to contribute to the pathogenesis of age-related as well as chronic diseases including cancer [31][32][33][34][35]. A number of studies are available documenting that OTA is associated with the production of reactive oxygen species and resulting oxidative stress through various direct and indirect mechanisms.

Oxido-Reduction Mechanisms.
Several oxido-reduction mechanisms elicited by OTA have been proposed. In a reconstituted system consisting of phospholipid vesicles, the flavoprotein NADPH-cytochrome P450 reductase and Fe 3+ , OTA was found to chelate ferric ions (Fe 3+ ), facilitating their reduction to ferrous ions (Fe 2+ ), which in the presence of oxygen, provided the active species initiating lipid peroxidation [36]. Results indicated that the hydroxyl radical was not involved in the process. A role for cytochrome P450 in this reaction was also suggested [36]. In contrast, others found that OTA induced oxidative damage through the generation of hydroxyl radicals. This reaction conducted with microsomes, in presence of NADPH and O 2 did not require exogenous Fe [20]. Structure-activity studies have also suggested that the toxicity of OTA may be attributable to its isocoumarin moiety and that the lactone carbonyl group may be involved in its toxicity. Using a Bacillus brevis model, Hoehler's et al. showed that OTA behaved as a cell pro-oxidant through mobilization of Fe 2+ and Ca 2+  Journal of Toxicology   Journal of Toxicology  pathway leading to uncoupling oxidative phosphorylation and increased production of hydroxyl radical via the Fenton reaction [37]. However, in other studies using OTA and structural analogs, a direct correlation between toxicity and iron chelating capacity was only partially supported [38]. The generation of an OTA hydroquinone/quinone couple from the oxidation of OTA (phenol oxidation) by electrochemical, photochemical, and chemical processes was reported [39,40]. The quinone is thought to undergo reductions to form hydroquinone, postulated to be responsible for the formation of the glutathione conjugate of OTA. Such events are likely to result in redox cycling and in the generation of reactive oxygen species [20,40,41]. The formation of OTA-derived quinones has been observed in cell cultures in vitro [41] as well as in vivo [42,43].

Reduction of Antioxidant Defenses.
OTA was found to reduce the expression of several genes regulated by nuclear factor-erythroid 2 p45-relatetd factor (Nrf2) [44][45][46][47]. This was observed at the RNA and protein levels, both in vitro and in vivo test systems. Nrf2 is involved in both the basal expression as well as in the induction of genes encoding detoxification, cytoprotective, and antioxidant enzymes [48,49]. A reduction of the expression of these genes is likely to result in decrease in antioxidant defenses leading to oxidative stress and macromolecular damage. This was confirmed with OTA. Through in vitro and in vivo studies, a correlation was observed between the OTA-dependent reduction of the Nrf2 pathway and an increased production of oxidative damage [46]. In this context, it is interesting to note that OTA was found to increase the expression of inducible nitric oxide synthase (iNOs) [50][51][52], an enzyme responsible for the production of nitric oxide (NO). An association between iNOs expression and the development of cancers was suggested in humans and animals in vivo [53]. In excess, NO may behave as a toxic radical producing nitrosative stress. NO is known to react with oxygen anion radical superoxide to form the pro-oxidant peroxynitrite. Under physiological conditions, peroxynitrite rapidly decomposes to generate a nitro radical intermediate leading to protein and DNA nitration. OTA was shown to stimulate protein and possibly DNA nitration [52,53] indicating that OTA exposure may be considered as a source of both oxygen and also reactive nitrogen radicals/species (RNS).

OTA-Mediated Oxidative Damage
3.1. DNA-Damage. ROS, such as hydroxyl radicals and nitric oxide, are capable of forming oxidized DNA bases that directly produce diverse types of DNA damage [54][55][56]. The oxidized DNA bases appear to be capable of inducing mutations that are commonly observed in neoplasia [33]. As illustrated in Table 1, diverse biomarkers have been analyzed showing that OTA induces DNA damage. In cell cultures, the OTA-dependent production of ROS was correlated with an increased formation of 8-oxoguanine [57][58][59]. Moreover, OTA was shown to induce DNA strand breaks as assessed by comet assay in liver, kidney, and spleen of F344 rats given 0, 0.25, 0.50, 1.0, and 2.0 mg/kg bw/day by gavage for 2 weeks [60]. In liver and kidney, the extent of DNA damage analyzed by comet assay was further enhanced in the presence of formamidopyrimidine glycosylase (Fpg), an enzyme involved in excision repair of oxidized DNA bases [57,58,61]. Another study [62] detected DNA-strand breaks (using the comet assay) in the kidney of female Wistar rats treated intraperitoneally with OTA (0.5 mg/kg bw/day for 7, 14, and 21 days; n = 5 per group). The severity of the DNA lesions in the kidney increased according to the OTA dose and was at maximum after 21 days of treatment. Other authors have observed oxidative DNA-damage in various tissues of animals treated with a wide range of OTA doses and treatment durations [46,60,63,64]. [54,[65][66][67]. Carbonylation is an important protein modification associated with alterations of protein (enzymes) function, protein misfolding, protein fate, and proteolysis. An increase in protein carbonyl content of tissues has been associated with a number of pathological disorders. Due to their abundance in mammalian cells, cytoskeletal proteins like actin are common targets for a variety of ROS and low-molecular weight reactive carbonyl species [67]. Inconsistent data have been reported on the potential impact of OTA on protein carbonylation [52,59,63,68,69]. No increase in protein oxidation was observed in liver and kidney of F344 rats treated with OTA at 0.3 mg/kg bw per day for 4 weeks [69]. The same finding was reported in liver of Wistar rats treated with OTA at 289 μg/kg bw for 90 days [63]. In contrast, in another in vivo study measuring protein oxidation in Wistar rats treated with OTA (0.5 mg/kg bw/day) for 24 h, 7, 14 and 21 days, a significant increase in protein carbonyls was found after 14 and 21 days of treatment in, respectively, the kidney and the liver [68]. Oxidative protein modification was observed in vitro [52].

Lipid Damage.
Lipid peroxidation is among the most extensively investigated processes induced by free radicals. Of these, the by-products, 4-hydroxy-2-nonenal (HNE), the tautomer malondialdehyde (MDA), acrolein and crotonaldehyde have been widely studied. The ability of these reactive electrophiles to modify DNA bases, yielding promutagenic lesions, is considered to contribute to the mutagenic and carcinogenic effects associated with oxidative stress-induced lipid peroxidation. HNE and MDA have increasingly been implicated in carcinogenesis [33,[70][71][72]. OTA has been reported to increase MDA formation. Initially, Rahimtula's group [73] observed that OTA was able to stimulate lipid peroxidation when added to liver or kidney microsomes or when administered to rats in vivo. Stimulation of lipid peroxidation by OTA-iron complex facilitating the reduction of iron was further reported [36]. Moreover, additional studies [74] indicated that OTA induced lipid peroxidation accompanied by leakage of calcium from calcium-loaded microsomes. Increased formation of MDA was observed in animal models treated orally with different doses of OTA [75][76][77][78][79]. HNE-protein adducts were measured in cell cultures treated with OTA [52].

OTA-Mediated Biological Response
It is widely acknowledged that reactive oxygen and nitrosative species can trigger biological responses such as stimulation or inhibition of signal transduction and gene expression. Such biological responses are considered to contribute to the expression of the carcinogenic potential of the reactive chemicals. A number of in vitro and in vivo studies are consistent with a role of oxidative and probably nitrosative stress as messengers involved in the adverse biological effects of OTA (Table 1).

Cell Signaling
ROS induces release of calcium from intracellular stores, resulting in the activation of kinases, such as protein kinase C (PKC). ROS species play also a critical role in the selective mobilization of other cell signaling responses. Cell signaling phosphoproteins of mitogen-activated protein (MAP) kinases including ERK, c-Jun N-terminal kinases (JNK), and p38 kinases are involved in proliferation, differentiation, and apoptosis. Activation of these molecules has been observed in response to changes in the cellular redox balance and are considered as vectors of ROS biological effects [33]. In vitro experiments provided evidence of an effect of OTA on intracellular calcium (Ca 2+ ) homeostasis [74,[80][81][82]. Ca 2+ homeostasis modulation by OTA was also observed in vivo [83]. An increased rate of microsomal Ca 2+ uptake was observed after OTA administration in vivo and in vitro [84]. Gene profiling analysis suggested a modulation of genes involved in calcium homeostasis by OTA [47,85]. These data indicate that Ca 2+ dependent signal transduction pathways may be affected by OTA treatment. Interferences were observed with other cell signaling pathways. OTA was shown to stimulate phosphorylation of ERK1/2, SAP/JNK, and p38 using in vitro models [82,86,87]. Using an in vivo model ERK1/2 a specific signaling response was also observed with mobilization of the atypical protein kinase C (PKCζ) and the insulin-like growth factor-1 (IGF-1) system [88]. Gene profiling data supported a potential role of the (IGF)-PI3K-PKB pathway in OTA-mediated renal toxicity in male rats [89]. [90][91][92]. Transcription factors contain a conserved redox sensitive cysteine residue; the oxidation of this residue inactivates the DNA-binding domain of the factor. Several studies have observed an inactivation of transcription factors resulting from increased concentrations of ROS [93][94][95]. For example, it was shown that the complex AP-1 is a basic leucine zipper protein, which is highly sensitive to changes in redox environment due to a cysteine residue in the DNA-binding domain [96]. Gene expression profile studies performed using in vivo models have shown that OTA impairs the antioxidant defense system regulated by Nrf2 [47]. This effect was supported by further studies [44][45][46]. Using in vitro models together with electrophoretic mobility shift assay, an inhibition of Nrf2 and AP-1 activity was shown as a result of OTA treatment [46]. Interestingly, Nrf2 binding site represents a bZip domain that interacts with the Antioxidant Response Element (ARE) DNA-binding site triggering the transcription of Nrf2 regulated genes. The ARE motif (5 -A/GTGAC/TnnnGCA/G-3 ) shares structural similarity with AP-1 binding site (5 -TGACTCA) involving both cysteine rich residues, the target of ROS oxidation [97,98]. These data strongly suggest a correlation between the generation of ROS by OTA and the subsequent inactivation of Nrf2 as previously described for AP-1. This molecular mechanism appears as a likely molecular component explaining the reduction of the defense response observed under OTA treatment.

Redox Modulation of Transcription Factors. Numerous reports have characterized interactions of ROS and RNS with activity of transcription factors
In vivo experiments provided direct evidence that Snitrosylation can interfere with transcription [91,92]. Nitric oxide (NO) induces the nitrosylation of cysteine residues (thiol groups) within or near the DNA-binding domain and/or insertion of the zinc finger, which is a DNA-binding motif, resulting in the inhibition of the DNA-binding activity of transcription factors [99][100][101]. This is illustrated by the example of the suppression of P450 gene expression by NO. NO-donors were found to suppress CYP2D6 promoter activity through inhibition of the transcription factor called hepatocyte nuclear factor 4 (HNF4) [94,95]. Interestingly, in a gene expression study performed by our group, OTA was associated with a significant reduction of the expression of genes regulated by HNF4 suggesting an indirect evidence of the role oxidative stress and the transcription factors regulation by OTA [47]. (GJIC). A strong correlation between the ability of a compound to block cell-to-cell communication in cultured cells and its ability to induce rodent tumors through nongenotoxic mechanisms has been documented [33,[128][129][130][131]. Disruption of gap junction intercellular communication was specifically reported in human renal cancer cell lines [132]. For example, Connexin 32 (Cx32) was discovered to be generally downregulated in human renal cell carcinoma (RCC) cell lines and in cancerous regions of the kidney [133]. ROS such as H 2 O 2 , an established tumor promoter, is known to modulate cell-to-cell communication. Likewise, certain chemicals inducing ROS were shown to inhibit intercellular communication in a variety of cells in culture systems [33]. Data on the potential effects of OTA on GJIC are inconsistent. Kidney epithelial cells treated with OTA resulted in modulation of gap junction-mediated intercellular communication, through a reduced expression of the gap-junction protein CX43 [134]. In addition, OTA strongly reduced the expression of other gap-junction proteins, CX26, CX32, and CX43, in liver of rats treated with OTA [63]. In another study, even though a strong reduction of CX43 was found in renal cells in vitro, OTA inhibited GJIC only in liver but not in kidney [135].

Counteracting OTA-Mediated Reduction of Nrf2 Activity.
As mentioned earlier, OTA was found to reduce the expression of antioxidant enzymes through inhibition of Nrf2 activity [46]. This reduction in antioxidant gene expression was correlated with increased oxidative damage of protein and DNA [46]. To further confirm the actual role of the reduction of defense mechanisms in the induction of cellular oxidative damage, inducers of Nrf2 were applied in cell cultures in vitro. All reported OTA effects at the levels of Nrf2 activity, Nrf2-regulated gene expression, and DNA damage were prevented in cell pretreated with Nrf2 inducers [46], strongly indicating a causal relationship between Nrf2 effects and oxidative damage.

Application of Radical Scavengers.
Several studies have been performed to try to counteract the adverse effects of oxygen radicals generated under OTA-treatment. A number of molecules with various antioxidant properties were tested, including inula crithmoides, cyanidin 3-O-b-D-glucoside, catechins, melatonin, superoxide dismutase, catalase, and Nacetyl-L-cysteine (NAC), using in vivo or in vitro models. Protection against OTA-induced DNA damage, lipid peroxidation as well as cytotoxicity was observed [75,79,102,105,106,116,118,127] further confirming the link between OTA exposure and oxidative macromolecular damage. However, up to now, the application of chemicals known to possess antioxidant properties failed to prevent the development of OTA-induced tumors in animal models. Authors reported Transcription factors inactivation Cytoskeleton (actin) Figure 1: Scheme to illustrate the oxidative stress-mediated mode of action proposed for OTA. Increased production of ROS, RNS, and RCS is likely to originate either from direct redox reactions involving OTA or through the inhibition of cellular defenses such as through the inhibition of transcription factors as Nrf2 which regulates enzymes with antioxidant properties. The generation of radicals will induce macromolecular oxidative damage such as oxidized DNA bases which may be converted into mutation resulting into generation of transformed cells. In addition, radicals will trigger biological responses which may impair intercellular communication and induce cell proliferation as well as reduction in cellular defense in oxidative stress. This last effect is likely to amplify the oxidative stress-mediated effects of OTA. Altogether, these molecular mechanisms will result in cancer development.
protection against nephrotoxicity but not carcinogenicity induced by ochratoxin A, implicating two separate pathways [122,123].

Application of Peroxidase Inhibitors.
Indomethacin and aspirin were found to prevent OTA genotoxicity in the urinary bladder and kidney of mice [136]. These data suggested the possible co-oxidation of OTA by enzymes involved in arachidonic acid biotransformation, especially prostaglandin-H-synthase (PGHS) and/or lipoxygenase [124]. Such reactions thought to produce activated OTA metabolites are also known to generate ROS which then may induce oxidation [124].

Conclusion
The carcinogenic mycotoxin OTA has been reviewed by a number of expert groups [3-5, 14, 137]. These expert groups identified the production of oxidative stress as an important event in the mode of action of OTA-induced nephrocarcinogenicity. In the present paper, the actual evidence available supporting such hypothesis was collected and reviewed.
It has been clearly shown that OTA generates oxidative stress predominantly in kidney through potentially direct (redox cycling) and indirect (reduction of cellular antioxidant defenses) mechanisms. Interestingly, these two mechanisms may interact with each other. The reduction of defense may amplify the impact of the direct production of radicals. Consequences resulting from the production of oxidative stress were observed at different levels. High kidney susceptibility to oxidative stress conditions may explain the target-specific toxicity of OTA. Oxidative stress has been incriminated in a number of kidney pathological pathways [138][139][140].
As depicted in Figure 1, first, OTA exposure was associated with increased levels of oxidative DNA, lipid, and protein damage. Second, various biological pathways known to be mobilized under oxidative stress were shown to be altered by OTA. Importantly, these effects were observed in both in vitro and in vivo test systems. Active in vivo doses were within doses known to induce tumors in kidney. These mechanisms are likely to be relevant for humans.
In conclusion, the evidence for the induction of an oxidative stress response resulting from OTA exposure can be considered strong. Because the contribution of the oxidative stress response in the development of cancers is well established, a role in OTA carcinogenicity is plausible. Altogether, the data reviewed above support the application of a threshold-based approach to establish safe level of dietary human exposure to OTA.