Mechanistic insights into the cytotoxicity and genotoxicity induced by glycidamide in human mammary cells

Abstract

Acrylamide (AA) is a well-known industrial chemical classified as a probable human carcinogen. Benign and malignant tumours at different sites, including the mammary gland, have been reported in rodents exposed to AA. This xenobiotic is also formed in many carbohydrate-rich foods prepared at high temperatures. For this reason, AA is an issue of concern in terms of human cancer risk. The epoxide glycidamide (GA) is thought to be the ultimate genotoxic AA metabolite. Despite extensive experimental and epidemiological data focused on AA-induced breast cancer, there is still lack of information on the deleterious effects induced by GA in mammary cells. The work reported here addresses the characterisation and modulation of cytotoxicity, generation of reactive oxygen species, formation of micronuclei (MN) and quantification of specific GA–DNA adducts in human MCF10A epithelial cells exposed to GA. The results show that GA significantly induces MN, impairs cell proliferation kinetics and decreases cell viability at high concentrations by mechanisms not involving oxidative stress. KU55933, an inhibitor of ataxia telangiectasia mutated kinase, enhanced the cytotoxicity of GA (P < 0.05), supporting a role of this enzyme in regulating the repair of GA-induced DNA lesions. Moreover, even at low GA levels, N7-GA-Gua adducts were generated in a linear dose–response manner in MCF10A cells. These results confirm that human mammary cells are susceptible to GA toxicity and reinforce the need for additional studies to clarify the potential correlation between dietary AA exposure and breast cancer risk in human populations.

Introduction

Acrylamide (AA) is a well-known toxic chemical classified as probably carcinogenic to humans (Group 2A) by the International Agency for Research on Cancer (1). More recent evaluations by other major international bodies, such as the US Environmental Protection Agency (2) and the World Health Organization/Food and Agriculture Organization (3), have confirmed that AA is a likely human health hazard. AA was originally regarded as an industrial or occupational genotoxicant; however, Tareke et al. (4) reported the presence of AA in human foodstuffs heated during cooking or manufacturing. Subsequently, it was shown that AA can be generated by a Maillard reaction between asparagine and reducing sugars in food processed upon heating (5,6). Further studies demonstrated that AA can be found in considerable amounts in regularly consumed foods and beverages, such as processed cereals, French fries, potato chips and coffee, with an average daily intake estimated to be about 1.0 µg/kg body weight (bw) and up to 4 µg/kg bw for high consumers (7). Cigarette smoke is an additional source of exposure and can contribute ca. 3.1 µg AA/kg bw/day (8).

The carcinogenicity of AA has been demonstrated at multiple organ sites in a number of rodent models (9–12). Several recent epidemiological studies, focused on dietary exposure to AA and cancer outcome, have reported conflicting or inconclusive results (reviewed in ref. 13). Some of these studies specifically examined the relationship between AA intake and breast cancer, since the data from rodent bioassays consistently indicate that AA exposure causes benign and malignant mammary tumours in female rats and mice (9,10,12). While some of the epidemiological studies failed to identify associations between dietary exposure to AA and breast cancer incidence (14–16), others found weak positive correlations with endocrine-related breast cancers, after adjusting for smoking status (17,18).

The mechanism of AA carcinogenesis has been a matter of debate. A hormonal mode of action was hypothesised from the observation of AA-induced tumours in the mammary gland and other rat tissues regulated by the endocrine system (e.g. thyroid, peri-testicular mesothelium), but mechanistic evidence supporting this assumption is limited or lacking (2,11). On the other hand, a substantial body of evidence is consistent with a genotoxic mechanism for AA-induced rodent carcinogenesis. AA can be conjugated with reduced glutathione (GSH), resulting in the urinary excretion of a mercapturic acid conjugate, or undergo metabolic oxidation, mediated by cytochrome P450 (CYP) 2E1, to the genotoxic epoxide glycidamide (GA). GA can be then detoxified via conjugation with GSH or hydrolysis by epoxide hydrolase (19). However, in contrast to AA, GA is highly reactive toward DNA, giving rise to covalent adducts. A number of DNA adducts have been fully characterised, including the depur inating adducts N7-(2-carbamoyl-2-hydroxyethyl)guanine (N7-GA-Gua; major) and N3-(2-carbamoyl-2-hydroxyethyl)adenine (N3-GA-Ade) and a minor 2′-deoxynucleoside derivative, N1-(2-carboxy-2-hydroxyethyl)-2′-deoxyadenosine (N1-GA-dA) (20,21). Recently, N1-GA-dA has been reported in DNA from cells treated in vitro with GA, although it was not detected in vivo (22).

GA is typically more genotoxic and cytotoxic than AA in vitro and in vivo (23,24) and increasing evidence suggests that GA plays a pivotal role in AA carcinogenesis. For instance, AA administration to CYP2E1-deficient mice resulted in significantly reduced levels of male germ cell mutagenicity, micronuclei (MN) and GA–DNA adducts as compared to those observed in wild-type mice (25–27). Moreover, comparative studies of DNA adduct formation, mutation induction and tumourigenicity in mice treated neonatally with AA or GA suggested that AA carcinogenicity is dependent upon metabolism to GA (28,29).

Given that covalent adduction of DNA by GA is a relevant genotoxic event, additional studies focusing on the mechanisms of GA-induced detrimental effects are fully warranted. These studies are particularly relevant if performed in cell systems corresponding to target organs of concern, as those identified in AA rodent carcinogenesis studies. The work presented herein addresses this issue using human mammary MCF10A epithelial cells as the test system. We describe DNA adduction (N7-GA-Gua and N3-GA-Ade) and MN profiles in MCF10A cells treated with GA over a large range of concentrations. These genotoxicity endpoints target differential levels of DNA damage that are integrated with cytotoxicity data. In addition, we address the role of reactive oxygen species (ROS)/oxidative stress, GSH and ataxia telangiectasia mutated (ATM) kinase in the viability of MCF10A cells exposed to GA. This mechanistic information, focusing on additional key aspects of genetic toxicology, can shed further light into the possible mode of action of GA in mammary cells.

Materials and methods

Chemicals

Dulbecco’s modified Eagle’s medium/Ham’s Nutrient Mixture F-12 (DMEM/F12), RPMI-1640 with l-glutamine, horse serum, fetal calf serum, penicillin–streptomycin solution (10 000U penicillin and 10mg streptomycin per millilitre in 0.9% NaCl), phosphate-buffered saline (PBS; 10mM, pH 7.4), trypsin from bovine pancreas, insulin from bovine pancreas, hydrocortisone, cholera toxin, human epidermal growth factor (EGF), thiazolyl blue tetrazolium bromide (MTT), tert-butylhydroperoxide (TBHP), GSH, l-buthionine sulphoximine (BSO), superoxide dismutase–polyethylene glycol (SOD–PEG) from bovine erythrocytes, catalase–polyethylene glycol (CAT–PEG) from bovine liver, ribonuclease A from bovine pancreas and cytochalasin B from Drechslera dematioidea were purchased from Sigma–Aldrich (St Louis, MO, USA). GA (CAS Registry Number 5694-00-8, >98.5% pure, containing ~1% AA) was obtained from Toronto Research Chemicals (North York, Ontario, Canada). Ethanol, methanol, potassium chloride, sodium chloride, dimethylsulfoxide (DMSO), acetic acid and Giemsa dye were obtained from Merck (Darmstadt, Germany). KU55933 was obtained from Tocris Bioscience (Bristol, UK). Dihydrorhodamine 123 (DHR123) and dihydroethidium (DHE) were acquired from Molecular Probes (Eugene, OR, USA). For these probes, 10mM stock solutions were prepared in DMSO, aliquoted and stored under nitrogen at −20°C. Mn(III) 5,10,15,20-tetrakis(N-ethylpyridinium-2-yl)porphyrin (MnTE-2-PyP⁵⁺, the charge is omitted from the text for clarity) was synthesised according to previously reported procedures (30).

Cell culture

The MCF10A human mammary epithelial cell line (ATCC CRL-10317) was purchased from the American Type Culture Collection. MCF10A cells were cultured in DMEM/F12 medium containing 5% horse serum, 100U/ml penicillin, 0.1mg/ml streptomycin, 10 µg/ml insulin, 0.5 µg/ml hydrocortisone, 20ng/ml EGF and 100ng/ml cholera toxin. Cells were kept at 37°C under a humidified atmosphere containing 5% CO₂ in air.

MTT reduction assay

Cell viability was evaluated using the MTT reduction assay, according to a procedure previously described (31). Cells were seeded in 200 µl of culture medium at a density of 4×10³ to 4.5×10³ cells per well in 96-well plates. Cultures were then incubated and exposed to the test compounds according to the following conditions:

• (a) For characterisation of the cytotoxicity profile of GA, cells were incubated for 48h and then exposed to different concentrations of GA (dissolved in PBS), ranging from 0.1 to 4mM, for 24h.

• (b) To study the toxicity of GA in cells with reduced levels of endogenous GSH, 24-h cultures were incubated with 100 µM BSO for a further 24h. The medium was removed, the cells were washed with culture medium and then exposed to 1mM GA for 24h. For the other redox modulators, a co-incubation protocol was used. Forty-eight hours after inoculation, cultures were simultaneously exposed to GA (1–3mM) and either GSH (1mM), CAT–PEG (50U/ml), SOD–PEG (50U/ml) or MnTE-2-PyP (1 and 5 µM) for 24h.

• (c) To study the impact of ATM inhibition on GA cytotoxicity, cells were grown for 46h and then incubated with the inhibitor KU55933 (aliquoted from a stock solution in DMSO to obtain a 10 μM concentration in the incubation mixtures). The final concentration of DMSO in the culture medium did not exceed 0.5% (v/v). Two hours after adding the inhibitor, GA (1 or 2mM) was added and cultures were incubated for a further 24h. Besides the standard protocol, in which the MTT assay was performed immediately after the 24-h treatment with GA, an additional protocol was conducted. In this supplementary procedure, the initial cell number was 1.0×10³ per well, and the MTT assay was performed 48h after the end of the 24-h treatment with GA.

After exposure of cultures to the test compounds as described above, the MTT assay was conducted according to previously reported procedures (31–33). In all experiments, the absorbance obtained with non-treated control cultures was considered to indicate 100% cell viability. Eight replicate cultures were included in each independent experiment. The IC₅₀ value for GA in MCF10A cells was calculated by non-linear regression analysis using Graph Pad Prism Software (version 5.01, La Jolla, CA, USA).

Assessment of ROS

The levels of intracellular ROS in GA-treated MCF10A cells were assessed using the fluorescence probes DHR123 and DHE, according to the procedure described by Gonçalves et al. (31). Briefly, ~8×10³ cells/well were cultured for 48h in 96-well plates (black-wall/clear-bottom). The culture medium was then replaced and cells were exposed to GA (up to 4mM final concentration), in the presence of DHR123 (10 µM) or DHE (10 µM), for 140min. This short incubation period was selected to avoid pronounced cell death. The assays were then conducted as previously reported (31). TBHP (1mM), a recognised oxidant (32,33), was used as a positive control. At least two independent experiments were performed, each comprising six replicate cultures for each experimental point.

Cytokinesis-blocked micronucleus assay

GA treatment and cell harvesting.

Approximately 2.5×10⁵ cells were cultured for 48h in 25-cm² culture flasks. The cells were then incubated with different concentrations of GA for 24h. After the treatment, the cells were washed with fresh culture medium, and cytochalasin B was added at a final concentration of 4.5 µg/ml. The cells were grown for a further 24h to allow for the recovery of binucleated (BN) MCF10A cells. The cells were then harvested by trypsinisation, washed and submitted to mild hypotonic treatment as described previously (34). After centrifugation, the cells were placed on dry slides and smears were made. After air drying, the slides were fixed with cold methanol (30min) and stained with Giemsa (4% v/v in 10mM phosphate buffer, pH 6.8) for 8min. Three independent experiments were performed.

MN frequency.

For each experimental point, 1000 BN cells were scored using a ×1000 magnification with a light microscope according to described criteria (35). The frequency of micronucleated cells in 1000 BN cells (‰ MNBN) and the total number of MN present in 1000 BN cells (MN/1000BN) were used as genotoxicity indices. These indices give complementary information. ‰ MNBN represents the frequency of DNA damaged cells (micronucleated), regardless of the number of MN present in each damaged BN cell, whereas MN/1000BN reflects the average number of total MN per BN cell.

Cell proliferation indices.

The decrease in cell proliferation induced by GA was assessed by two standard methods: the percentage of binucleated cells (%BN) and the nuclear division index (NDI) (35,36). To determine these indices, 500 cells were classified according to the number of nuclei using a ×400 magnification with a light microscope.

GA–DNA adduct analysis

GA treatment and cell harvesting.

MCF10A cells were seeded into 75-cm² culture flasks (6×10⁵ cells/flask). After 48h, the culture medium was removed and replaced with fresh medium. The cells were then grown for 24h in the presence of GA (1–1000 µM). The cells were harvested by trypsinisation and washed with PBS, and the cell suspension was immediately stored at −20°C.

DNA extraction.

The DNA extraction was performed as previously described in detail (23). DNA was extracted from the stored cell suspensions using the QIAamp DNA mini Kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions, with slight adaptations to prevent depurination of the DNA adducts (37). Namely, cell suspensions (250 µl) were lysed for 1h at 37°C with 20 µl of Proteinase K, provided by Qiagen, and 20 µl of ribonuclease A (10 µg/µl). The lysates were then submitted to purification steps and, at the end of chromatographic process, DNA samples were eluted in water (200 µl) and stored at −80°C for subsequent DNA quantification and DNA adduct determination.

DNA adduct quantification.

The DNA was quantified spectrophotometrically on a Nanodrop 1000 apparatus (Thermo Scientific, Waltham, MA, USA). The levels of N7-GA-Gua and N3-GA-Ade were assessed by high performance liquid chromatography (HPLC) coupled with electrospray ionisation tandem mass spectrometry (ESI-MS/MS) essentially as described previously (23,37). Briefly, aliquots of the DNA solutions, containing synthetic ¹⁵N-labelled N7-GA-Gua and ¹⁵N-labelled N3-GA-Ade as internal standards, were thermally hydrolyzed at 100°C for 15min, cooled to room temperature and then filtered through a prewashed Amicon Microcon 3kDa molecular weight cut-off spin filter (Millipore Co., Billerica, MA, USA). The adducts were separated on a 2×150mm C18 analytical column (Luna C18(2), Phenomenex, Torrance, CA, USA) with 2% acetonitrile in water and quantified by ESI-MS/MS in the multiple reaction monitoring mode, using a Quattro Ultima triple quadrupole mass spectrometer (Waters, Milford, MA, USA) equipped with an electrospray source. The analyses were conducted with 3–61 µg DNA. Under these conditions, the limit of detection (LOD) was ~8 adducts in 10⁸ nucleotides.

Statistical analyses

The Kolmogorov–Smirnov test was used to assess the normality of continuous variables. The homogeneity of the variances was evaluated using Levene’s test. The mean value differences of the results from cultures submitted to different treatments were evaluated by Student’s t-test, except for the experiments with KU55933. In this case, the Mann–Whitney test was used. All statistical analyses were performed using SPSS for windows version 17 (SPSS Inc., Chicago, IL, USA) and the level of significance was chosen as P < 0.05.

Results

Cytotoxicity of GA in MCF10A cells

Treatment of MCF10A cells with GA resulted in a linear dose-dependent decrease in cell viability as measured by the MTT reduction assay (Figure 1). Significant differences in cell viability were observed for GA concentrations of 0.5mM (P < 0.05) and higher (P < 0.001), when compared with control cells. Very low viability values (<7%) were observed at the highest concentrations of GA tested (3 and 4mM). The IC₅₀ calculated for GA in this 24-h treatment protocol was 1.56±0.18mM.

Fig. 1.

Evaluation of GA cytotoxicity (0.1–4mM; 24-h incubation) in MCF10A cells using the MTT reduction assay. The results are expressed as mean ± SD (n ≥ 3).

Effect of redox modulators on GA cytotoxicity

In order to understand the influence of the GSH status on the cytotoxicity of GA, two different approaches were carried out. First, BSO was used to inhibit the synthesis of GSH. Additionally, the effect of exogenous GSH was evaluated. The results of the experiments using a pre-incubation with BSO are shown in Figure 2A. The viability of MCF10A cells treated with 1mM GA markedly decreased from about 62% in the absence of BSO to 11% upon pre-incubation with 100 µM BSO (P < 0.001). As also displayed in Figure 2A, pre-incubation with BSO alone (100 µM) caused only a slight decrease in cell viability (~3%).

Fig. 2.

Cell viability of MCF10A cells exposed to GA (24-h incubation) in the presence of different redox modulators. The values represent mean ± SD and are expressed as percentages relative to non-treated control cells. (A) BSO (100 µM, n = 3); (B) GSH (1.0mM, n = 7); (C) CAT–PEG (50U/ml, n = 2); (D) SOD–PEG (50U/ml, n = 2); (E) MnTE-2-PyP (1 and 5 µM, n = 3). *P < 0.05 and ***P < 0.001 when compared with MCF10A cells treated only with GA.

Figure 2B shows the results of simultaneous exposure to GA and GSH. The co-treatment with GSH (1mM) consistently caused an increasing trend in the viability of GA-treated cells at all the GA concentrations tested (1, 2 and 3mM). This protective effect was significant for the highest concentration of GA (3mM), in which the presence of exogenous GSH led to a 12% increase in absolute cell survival (P < 0.05). A control experiment showed that 1mM GSH per se led to a MCF10A cell viability similar to that of the control cultures, with a decrease in cell viability of only ~ 6% (P < 0.05).

Besides GSH, other cellular redox pathways were studied using three different redox modulators: two antioxidant enzymes in a pegylated form (SOD–PEG and CAT–PEG) and the synthetic polyfunctional antioxidant MnTE-2-PyP. These modulators were used at non-toxic concentrations for MCF10A cells. In GA-treated cells, the presence of CAT–PEG (50U/ml, Figure 2C), SOD–PEG (50U/ml, Figure 2D) and MnTE-2-PyP (1 and 5 µM, Figure 2E) did not change significantly the decreases in cell viability induced by 1mM GA.

Assessment of ROS generation

The levels of intracellular ROS in GA-treated MCF10A cells were assessed using the fluorescence probes DHR and DHE. As depicted in Figure 3, no increases in fluorescence were observed upon treatment of MCF10A cells with GA at concentrations up to 4mM, either using DHR (Figure 3A) or DHE (Figure 3B) as probes. In contrast, the positive control TBHP (1mM) led to marked increases in fluorescence intensity with both DHR (P < 0.01) and DHE (P < 0.05).

Fig. 3.

Intracellular levels of ROS in GA-treated MCF10A cells, as evaluated by the DHR (A) and DHE (B) probes. The values (mean ± SD; n = 2–4) represent relative fluorescence units (RFU) expressed as percentages of controls. TBHP (1mM) was used as a positive control. *P < 0.05 and **P < 0.01 when compared with controls.

Quantification of specific GA–DNA adducts

MCF10A cells were exposed to a wide range of GA concentrations, from 1 µM to 1mM, in order to evaluate the formation of specific GA–DNA adducts. The depurinating adducts, N7-GA-Gua (major) and N3-GA-Ade (minor), were assessed by HPLC-ESI-MS/MS as described previously (21,23,37). The data are presented in Figure 4A and B for N7-GA-Gua and N3-GA-Ade, respectively. The results are expressed as mean ± SD from two independent experiments for all data points except 500 µM GA (only one experiment was considered valid at this concentration).

Fig. 4.

GA–DNA adduct levels, as assessed by HPLC-ESI-MS/MS of thermally hydrolyzed DNA from human mammary MCF10A cells exposed to GA (1–1000 µM). (A) Levels of the predominant N7-GA-Gua adduct. (B) Levels of the minor adduct, N3-GA-Ade. The levels of both adducts were <LOD (8 adducts/10⁸ nucleotides) for non-treated controls. The levels of N3-GA-Ade were <LOD for GA ≤100 µM. The values are expressed as mean ± SD (n = 2, except for 500 µM GA, where n = 1).

A linear dose–response relationship is evident for N7-GA-Gua up to 1mM GA (R² = 0.9979). The data obtained for this adduct at the lower GA concentrations are depicted in the inset of Figure 4A. The adduct was quantified in both experiments at GA concentrations as low as 1 µM. Untreated negative controls had N7-GA-Gua levels below the LOD (8 adducts/10⁸ nucleotides) in both experiments.

N3-GA-Ade could only be quantified at GA concentrations ≥250 µM (Figure 4B) due to the fact that it is found at ~1% of the level of N7-GA-Gua (21,23,28). At these very high concentrations, a linear dose–response relationship was also found (R² = 0.9999). For lower GA concentrations (up to 100 µM), the levels of N3-GA-Ade were <LOD in both experiments.

Induction of MN in GA-exposed MCF10A cells

The results from the cytokinesis-blocked micronucleus (CBMN) assay performed with MCF10A cells exposed to GA (up to 2mM) are depicted in Figure 5. The impairment of cell proliferation by GA was first evaluated using the %BN cells (Figure 5A) and NDI indices (Figure 5B). A linear concentration-dependent decrease, with roughly similar profiles for both indices, was observed from 0.25 to 2mM GA (Figure 5A and B); most of the cells were either mononucleated or BN. The highest GA concentration tested (2mM) profoundly affected the cell proliferation indices. In fact, the %BN cells was so low at this GA concentration (<10%, P < 0.001) that a correct assessment of MN was precluded (Figure 5A).

Fig. 5.

CBMN assay in MCF10A cells exposed to GA. (A) %BN cells; (B) NDI; (C) ‰ MNBN. The values are expressed as mean ± SD (n = 3). *P < 0.05, **P < 0.01 and ***P < 0.001 when compared with controls.

The frequency of micronucleated BN cells (expressed as ‰ MNBN) was about 18‰ for controls (Figure 5C), reflecting the extent of basal DNA lesions that occur spontaneously during cell culture. In the presence of GA, the MN frequencies showed an increasing dose–response trend, although the formation of GA-induced MN was significant only at the highest GA concentration tested (1mM), for which the ‰ MNBN significantly increased by more than 2-fold (P < 0.01) when compared to untreated controls.

GA also increased the total number of MN per BN cell, expressed as MN/1000BN. The profile obtained with this index (data not shown) was similar to the one presented for ‰ MNBN in Figure 5C since most of the micronucleated cells contained one MN per cell; the exception was treatment with 1mM GA, which resulted in some micronucleated cells with two or more MN. The MN/1000BN levels were 46.0±7.5‰ for 1mM GA, compared to 21.3±4.7‰ for untreated control cells (P < 0.01).

Effect of the ATM inhibitor KU55933 on GA cytotoxicity

The ATM activity was inhibited using 10 μM KU55933 (Figure 6). Initially, the standard protocol used for the redox modulators was employed, i.e. the MTT assay was performed at the end of a 24-h treatment with GA. The results are depicted in Figure 6A and show that KU55933 slightly potentiated the cytotoxicity of GA (1 and 2mM). The effect was significant for 2mM GA (P < 0.05). Under these conditions, KU55933 was not cytotoxic to the MCF10A cells: cell viability values were within ~98% of controls.

Fig. 6.

Viability of MCF10A cells exposed to GA in the presence of the ATM inhibitor KU55933 (10 µM). Two protocols were followed: (A) the MTT assay was performed at the end of a 24-h treatment with GA (n = 4–5) and (B) the MTT assay was performed 48h after the end of a 24-h treatment with GA (n = 4). The values represent mean ± SD and are expressed as percentages relative to non-treated control cells (*P < 0.05 when compared with MCF10A cells treated only with GA).

This finding was confirmed using an additional experimental protocol, in which the MTT assay was performed 48h after the end of the 24-h treatment with GA and KU55933. Under these conditions, KU55933 significantly increased the cytotoxicity of 1mM GA (Figure 6B) causing a decline in cell viability from 70±8.2 to 49±7.7% (P < 0.05). Using this protocol, cultures treated only with KU55933 had a cell viability of about 93%.

Discussion

Dietary exposure to AA represents a public health concern in terms of cancer risk. This toxicant is easily absorbed and distributed throughout the body (7). A study performed in female rats showed that at 2 and 4h after the oral administration of equimolar doses of AA or GA, the levels of GA in breast tissue were similar regardless of the compound administered, although at relatively lower levels compared to those measured in serum (38). Mammary epithelial tissue is considered an important target in the context of dietary AA exposure, based mainly on the results of long-term rodent carcinogenesis studies (9,10,12).

Although epidemiological evidence to date is not conclusive, some studies have recently reported weak associations between AA intake and the incidence of endocrine-related breast cancer (17,18). Due to these epidemiological data and the experimental evidence from rodent models, including GA accumulation in the rat mammary gland (38), it is crucial to understand the mechanisms of GA toxicity in breast tissue. However, so far very few studies have addressed this issue in mammary cells. To fill this gap, the present work investigated GA-induced deleterious effects in MCF10A cells, a spontaneously immortalised and growth factor-dependent human mammary epithelial cell line (39) that retains many of the functions of normal human breast cells and is commonly used as a reference for non-tumour breast cells (31). These cells express multiple CYPs (40), including low levels of CYP2E1 (41).

The data obtained in the present work show that GA is significantly cytotoxic only at concentrations above 500 µM. The IC₅₀ value is relatively high, showing that MCF10A cells tolerate large amounts of GA. Overall, these results suggest that mammary cells remain metabolically viable at very high GA concentrations. This may be relevant from the standpoint of carcinogenesis, in the sense that GA may form non-lethal mutations and foster a subsequent tumourigenesis process.

GSH, a fundamental defence line for the mammalian cell, is involved in several key functions, including the conjugation of deleterious electrophilic compounds, being considered a major intracellular antioxidant (42). In the context of AA exposure, both AA and GA can be effectively conjugated with GSH (7). Toxicokinetic studies have demonstrated the presence of mercapturic acid derivatives of AA and GA in the urine of experimental animals and human populations exposed to AA (24). While there is some information focusing on the role of GSH conjugation (43) and oxidative stress in AA-treated cells (44), few corresponding data are available for GA in cell-based assays. In the current study, MCF10A cells were pre-incubated with 100 µM BSO (43) for 24h, a period that roughly corresponds to the length of the cell cycle. While the pre-treatment of cells with BSO effectively reduces the level of endogenous GSH (45), co-treatment with GSH and a xenobiotic agent essentially results in an extracellular effect, since GSH does not easily enter the cells (46). The results from BSO depletion show that endogenous GSH is crucial for MCF10A cells to cope with GA, while additional GSH had only a small although significant effect in terms of increase in cell viability.

As mentioned above, GSH is also crucial to decrease chemically induced oxidative stress. Some reports suggest that AA is associated with oxidative stress in different mammalian cells, although this phenomenon was described at very high AA concentrations (44). Conversely, there is little information on the induction of ROS and the role of antioxidants in GA-exposed cells. In order to clarify this issue, the role of antioxidants with distinct modes of action was studied in MCF10A cells treated with GA. SOD and CAT are key enzymes in the cellular antioxidant defence network. SOD catalyzes the dismutation of O₂^·− into O₂ and H₂O₂, while CAT catalyzes the decomposition of H₂O₂ to O₂ and H₂O (47). To facilitate intracellular delivery of these enzymes, SOD and CAT were used in the form of PEG conjugates. A different approach to modulate cell redox pathways was the use of the synthetic polyfunctional antioxidant MnTE-2-PyP. Besides being a well-established SOD mimic, this manganese(III) porphyrin can modulate the cellular redox status by scavenging other ROS and reactive nitrogen species, including ONOO⁻, CO₃^·− and HOCl, by influencing the GSH status and by modulating redox-sensitive transcription factors. MnTE-2-PyP is thus a highly promising drug for the treatment of a plethora of oxidative stress-related diseases and is a very useful tool for mechanistic redox studies (48). The results obtained with these three complementary redox modulators are consistent and suggest that oxidative stress is not involved in GA-induced cytotoxicity in MCF10A cells. This finding was further confirmed by evaluating the generation of ROS in MCF10A cells upon GA treatment. To detect intracellular ROS, two complementary oxidant-sensitive probes, DHR and DHE, were used. DHR easily crosses cell membranes and undergoes oxidation to the fluorescent rhodamine. This oxidation may be mediated by different ROS, reflecting the overall intracellular ROS levels (31,49). However, DHR is considered unreactive towards superoxide (50). We thus used the fluorogenic probe DHE to detect the intracellular levels of superoxide radical anion. DHE is oxidised intracellularly to the fluorescent product hydroxyethidium and, although this oxidation is not absolutely specific, it is primarily superoxide dependent (33,51). The results obtained in these assays demonstrate that GA, up to an extremely high concentration in the millimolar range (4mM), does not induce ROS. Taken together, the data presented in this work consistently suggest that oxidative stress is not the mode of action of GA in human mammary cells.

A further step in this work was to assess the importance of genetic damage induced in MCF10A cells by different GA concentrations. We initially evaluated the formation of specific GA–DNA adducts in GA-treated human mammary cells. The results revealed a linear dose-dependent formation of the predominant adduct, N7-GA-Gua, over the entire range of concentrations studied (1–1000 µM). Notably, this DNA adduct was quantified at GA concentrations as low as 1 µM, highlighting the likely relevance of DNA adduction in a context of dietary exposure to AA.

In view of the GA–DNA adduct data, the next step was to assess if this damage was associated with a cytogenetic genotoxic response. The CBMN assay is a methodology primarily focused on the evaluation of cytogenetic damage from the formation of MN in BN cells (35). This technique detects both clastogens and aneugens. The use of cytochalasin B greatly improved MN detection and is currently an established and widely used assay. This assay also provides further information, including the assessment of cell proliferation kinetics. In this context, and according to the proliferation indices, %BN cells and NDI, GA affected cell division/proliferation, although this impairment was more evident for GA concentrations >500 µM. This could be explained by a blockage in the cell cycle progression in order to allow the repair of GA-induced lesions.

In assessing the cytogenetic burden caused by GA in mammary cells, it is clear that this AA metabolite is genotoxic to MCF10A cells, but only at high concentrations. In fact, a more than 2-fold increase in the MN indices was only significant at 1000 µM GA. GA has previously been found to be clastogenic at a similar concentration level in other mammalian cells (e.g. V79 Chinese hamster cells; 23), but few data are available for cells from human origin. By integrating GA adducts, MN and cell viability data, it becomes clear that while the primary DNA damage (i.e. DNA adducts) is evident at low levels of exposure, its consequences, either in terms of MN formation or in decreased cell proliferation kinetics, are only obvious at much higher, non-dietary relevant GA concentrations. DNA repair mechanisms may be contributing to this differential outcome.

Previously, we have reported a strong correlation between GA–DNA adduct levels and the induction of sister-chromatid exchanges (SCEs). GA was also found to be a very effective inducer of SCEs (23). These cytogenetic biomarkers have been described to be a consequence of the homologous recombination (HR) repair of double-strand breaks (DSBs). The formation of DSBs may also be responsible for the clastogenic effects of GA observed using the CBMN assay. The accumulation of DSBs eventually ends in a cell-death-driven process. In view of this, and with the purpose of understanding DNA repair mechanisms triggered in GA-exposed cells, we used a pharmacological inhibition approach focusing on ATM kinase, a key enzyme involved in DSB repair. ATM is a crucial signalling enzyme that regulates HR of DSBs (52,53), although it is also important for cell cycle progression control and other cell functions (54). In this study, KU55933, a recognised ATM inhibitor (55), significantly increased the cytotoxicity of GA in MCF10A cells, suggesting an important role of this kinase and HR in the repair of GA-inflicted lesions. The HR pathway has been reported as a possible repair pathway in a GA exposure context (23,56). Using a different methodology, our results reinforce this perception and emphasise the need to use additional approaches to evaluate mechanisms of DNA repair. Other targets, including base excision repair, nucleotide excision repair and DSB repair, should also be evaluated using not only DNA repair inhibitors but also complementary strategies, namely studies of the expression of DNA repair proteins or investigation of correlations between DNA repair polymorphisms and the extent of GA-induced DNA damage.

In summary, the results from this study clearly show that, even at low level GA generates the N7-GA-Gua adduct in a linear dose–response manner in human mammary epithelial cells. Moreover, GA induces MN, impairs cell proliferation kinetics and decreases cell viability at higher GA concentrations by mechanisms not involving oxidative stress. Taken together, these findings confirm that human mammary cells are susceptible to GA toxicity and reinforce the need for further experimental studies, as well as for additional well-designed epidemiological investigations, in order to clarify the potential correlation between dietary AA exposure and breast cancer risk in human populations.

Funding

This work was supported by Fundação para a Ciência e a Tecnologia (FCT), Portugal (Projects PTDC/SAU-OSM/105572/2008, PEst-OE/SAU/UI4013/2011 and PEst-OE/QUI/UI0100/2013, Programme Ciência 2008, and PhD grants SFRH/BD/70293/2010 and SFRH/BD/22612/2005).

Acknowledgement

The opinions expressed in this paper do not necessarily represent those of the US Food and Drug Administration.

Conflict of interest statement: The authors declare that they have no conflict of interest.