**4. Discussion**

This is the first study to examine the effect of a clinically relevant combination of the chemotherapeutics Dox and 4-Cyc on cytotoxicity and ROS production by human breast and ovarian cell lines in vitro. Our finding that γToc reduced chemotherapeutic-generated ROS production by transformed ovarian granulosa COV434 and epithelial adenocarcinoma OVCAR cells, but not by breast cancer cells, indicates the potential to develop antioxidant γToc as an adjunct treatment to reduce the adverse e ffects of chemotherapeutic-stimulated ROS on proliferating ovarian granulosa cells.

DAPI and CV staining showed that there were similar numbers of viable cells in cell culture medium and culture medium containing 0.8% DMSO, but there were approximately 33% fewer viable COV434 cells in the culture medium than in the other three cell lines. The cell doubling times have been reported as COV434 24 h [48] to 36 h [49], MCF7 30 h [50], T47D 39 h [50] and OVCAR 48 h [51]. If doubling times were the explanation for the di fference in viable cell numbers there should have been fewer OVCAR than COV434 cells, because OVCAR are reported as having the slowest doubling times. Although the ECACC recommends growing COV434 cells in high glucose DMEM, Tsai-Turton et al. (2007, [52]) used high glucose DMEM/F12, and we repeated their method because primary-derived human granulosa cells are commonly cultured in the same medium [53–55]. In the present study, however, the pH indicator phenol red was omitted to avoid interference with ROS quantification. Phenol red is a weak estrogen and COV434 cells respond to estrogen by proliferating. We speculate that our estrogen-depleted control culture medium resulted in lower levels of COV434 proliferation. The addition of 0.8% DMSO to this culture medium did not change the numbers of DAPI-stained COV434 nuclei with normal morphology, nor the numbers of viable cells quantified in our crystal violet assay, but significantly increased ROS and the percentages of COV434 condensed nuclei. It is likely that, if COV434 cells were cultured in 0.8% DMSO for longer periods of time (than 24 h), the numbers of viable cells would decrease to reflect the increase in ROS and condensed cell nuclei.

Breast cancer patients are commonly administered an infusion of Dox intravenously (60 mg/m2) then an infusion of cyclophosphamide (600 mg/m2) [56,57], and di fferent types of breast cancer have additional agents added to their treatment regimens (e.g., addition of paclitaxel to four cycles of Dox and cyclophosphamide [58,59], but breast cancer is not treated with either agen<sup>t</sup> alone. In early animal studies, the combination of Dox and cyclophosphamide was therapeutically potentiating against four di fferent murine mammary tumour lines compared to Dox as a single agen<sup>t</sup> [60] and this was attributed to di fferent cytotoxic mechanisms of action. However, we have been unable to locate any in vitro or human in vivo studies that compared Dox + cyclophosphamide to either Dox or cyclophosphamide alone.

The chemotherapeutics (without tocopherols) stimulated ROS in all four cell lines, reduced cell viability in COV434 and T47D cells, and increased the percentage of condensed cell nuclei in COV434, OVCAR and T47D cells, hence chemotherapeutic-stimulated ROS production was associated with cell damage when measured using DAPI or CV staining. 4-Cyc induced significant ROS generation in all cell lines within 1 h, and this increased 3- to 10-fold after 24 h exposure. Our data support previous reports that increases in ROS mediate 4-Cyc-induced apoptosis in di fferent types of cells [17,18]. Dox on the other hand, did not generate ROS as quickly as 4-Cyc, and in three of the cell lines Dox generated fewer ROS than 4-Cyc after 24 h exposure. ROS production in H9c2 cardiac muscle cells exposed for 1 h to 10 μM Dox was four times higher than in the medium control [61]. In the present study, the lower 1.21 μM concentration of Dox that killed 25% of MCF-7 cells was compared with Tan et al. (2010) [61], and this suggests that higher amounts of ROS are produced with increasing concentrations of Dox. COV434 cells exposed to 50 μM 4-Cyc for 2 h, or 1 μM for 6 h, significantly increased the production of ROS [52]. Although the concentration of DCFDA used in this study was 100 times higher than the one used in our experiments, and higher DCFDA concentrations can be toxic and contribute to ROS generation [38], the amount of ROS generated by COV434 cells after 24 h exposure to 21 μM 4-Cyc in our study was in broad agreemen<sup>t</sup> with the previous study [52].

The four cell lines displayed di fferent sensitivities to the cytotoxic and ROS-inducing activities of the test agents. Several factors a ffect cell line responses in vitro [62]. One factor is the cell doubling time, because some chemotherapeutics are phase-specific agents, which means that only cells that are passing through the relevant cell cycle phase when the drug is present are killed [63–65]. Because cells that are in a di fferent cell cycle phase are not targeted by the phase-specific agent, a single of dose of the drug may only kill a fixed fraction of cells and multiple doses may be needed to eradicate

the tumour [66]. 'Fractional kill' predicts a strong correlation between proliferation rate and drug sensitivity. Neither Dox or cyclophosphamide are considered cell cycle phase-specific drugs, but they have been known to preferentially target more metabolically active cells [23], and Fan et al. [67] found that Dox inhibited the growth of HepG2 cells by induction of G2/M cell cycle arrest. It's therefore possible that cell lines with longer doubling times might require a longer exposure time to Dox and 4-Cyc in vitro for cytotoxic effects. However, MCF7 cells have the shortest reported doubling time (30 h, [50]) but were the least sensitive to the cytotoxic effects of the combined chemotherapeutics. Previously, MCF7 and ovarian granulosa KGN cells were exposed to 5 μM Dox or 4-Cyc for 24 h then cultured for a further 48 h or exposed for 72 h continuously in culture [68]. There was no difference in cytotoxicity between these two exposure regimens; >80% of both cell lines died within the first 24 h. In the present study, then, the differing proliferation rates of the four cell lines do not appear to explain the data.

Another factor that affects in vitro responses to test agents is the origin and phenotype of the cell line. The MCF-7 and T47D cell lines were isolated from a pleural effusion of patients with breast carcinoma [69,70]. MCF-7 cells maintain several of the functional characteristics of differentiated mammary epithelium, including the expression of estrogen receptors [71], which means that control proliferation rates may have been reduced in our phenol red-free system. Similarly, the T47D line expresses receptors for estradiol, progesterone and other steroid hormones [70]. The COV434 cell line was derived from a solid primary human ovarian granulosa cell carcinoma but is a good in vitro model for normal healthy granulosa cells because the cell line maintains many of the functional characteristics required for follicle growth and development [72]. These include the expression of estrogen receptors and a steroidogenic pathway that enables COV434 cells to synthesise steroid hormones such as estrogen and progesterone. Steroidogenesis generates ROS, and human steroidogenic ovarian cells have relatively high levels of intracellular antioxidants which include Vitamin E in humans [73,74]. The OVCAR-3 cell line also synthesizes steroid hormones and was obtained from a patient who was administered a combination of Dox, cyclophosphamide and cisplatin to treat an epithelial adenocarcinoma of the ovary. Eight months later, her ascites fluid-containing ovarian adenocarcinoma cells were injected into nude mice, and the resulting tumours were disaggregated and used to generate the OVCAR-3 cell line. These cells are resistant to clinically relevant concentrations of doxorubicin, although the effects of 4-Cyc have not been reported [57]. In the present study, the MCF7-derived EC25 value of Dox reduced viable OVCAR cells to 75 ± 5% of control, which suggests that OVCAR have a resistance to Dox comparable to MCF7 cells, whereas the same concentration of Dox reduced COV434 cell viability to 57 ± 2% of control. The combination of the EC25 Dox with the EC25 4-Cyc was additive towards OVCAR cells (57 ± 2% of control) but did not increase cytotoxicity towards MCF7 (76 ± 14%) or T47D cells. Although the combined chemotherapeutics were not more cytotoxic to the two breast cancer cell lines than either agen<sup>t</sup> alone, there were marked increases in chemotherapeutic-stimulated ROS, to 1413 ± 230% (T47D) and 1085 ± 31% (MCF7) of control. If the chemotherapeutics also stimulate a 10-fold increase in ROS in vivo, perhaps it is accompanied by a more impressive increase in cytotoxicity than occurred in our in vitro system. The combined chemotherapeutics displayed synergistic cytotoxicity towards the granulosa-derived COV434 cells, and the reduction in viable cells to 34 ± 5% of control was accompanied by a 770 ± 84% increase in ROS. In this study, the viability of the granulosa tumour-derived COV434 cell line was more sensitive to the clinically relevant combination of Dox + 4-Cyc than the two breast cancer cell lines. More experiments are needed to determine if primary-derived physiologically normal granulosa cells display the same sensitivity, but since immortalised cancer-derived cell lines are generally more robust and resistant to cytotoxic agents than mortal primary-derived cells, we predict that proliferating follicular granulosa cells will be more sensitive to Dox + 4-Cyc than breast cancer cells.

Cancer-derived cell lines are different from tumours in vivo, and cell lines and tumours are different from physiologically normal, healthy cells. Tumour-derived COV434 granulosa cells, however, retain features of physiologically normal granulosa cells, and the ovarian epithelial OVCAR cells are different from the mammary epithelial cells because they, like normal granulosa and COV434 cells, retain a steroidogenic pathway. The steroidogenic pathway and associated intracellular antioxidants imply differences in redox status between the two ovarian and two breast cancer cell lines. This frames our hypothesis that antioxidant tocopherols might reduce chemotherapeutic-induced ROS and their associated damage in granulosa-like steroidogenic cells, whilst maintaining the anti-cancer cytotoxicity of the chemotherapeutics against breast cancer cells. Therefore, one of the aims of this study was to discover the effect of αToc and γToc on ROS induced by exposure to the combination of Dox + 4-Cyc.

In our earlier study, we did not find a concentration of αToc that killed MCF7 breast cancer cells, and selected the highest concentration tested. In the present study, this relatively high concentration of αToc alone did not kill any cell type compared to the DMSO control, had no significant effect on the proportions of condensed nuclei, and did not affect ROS generation in three of the four cell lines, although both tocopherols stimulated MCF7 ROS generation somewhere between 3 and 24 h of exposure. The addition of αToc to Dox + 4-Cyc for 24 h decreased T47D cell viability while increasing condensed nuclei, and significantly increased ROS in all four cell lines. The COV434 cells differed from the other three cell lines in that αToc reduced chemotherapeutic-stimulated ROS in the first 3 h of exposure, but in the subsequent 3–24 h of exposure the combination of Dox + 4-Cyc + αToc caused significantly more ROS, condensed nuclei and loss of viable COV434 cells than Dox + 4-Cyc without αToc, and additionally caused COV434 nuclei to shrink. Steroid hormones and tocopherols are lipophilic and miscible with cell membrane lipid bilayers. The chemotherapeutics Dox and 4-Cyc are soluble in water, and this caused us to speculate that if the relatively high, non-cytotoxic concentration of αToc used in this study were to affect the fluidity and plasticity of the cell membranes, it may have improved chemotherapeutics' access to the interior of the cell in some way. Another difference between COV434 cells and the other three cell lines is their size; COV434 cells are smaller and probably have a higher surface area to volume ratio, which would increase the importance of membrane changes and chemotherapeutic uptake relative to the other three cell lines. Our finding that αToc increased the cytotoxicity of Dox + 4-Cyc against the COV434 granulosa tumour cell line, in an estrogen-depleted in vitro system, provides a rationale for further investigation.

A 24 h exposure to the concentration of γToc that killed 25% of MCF7 cells in a previous study had no effect on condensed nuclei but significantly increased ROS and killed approximately 25% of the two breast cancer cell lines. In contrast, the same 24 h exposure to the MCF7 EC25 value of γToc did not kill the two ovarian cell lines and the lower percentages of condensed nuclei corresponded to the higher numbers of viable cells. Although 24 h exposure to γToc stimulated MCF7 ROS, it had no effect on ROS in any other cell line. It is interesting that, in the two ovarian cell lines, γToc alone not only prevented an increase in ROS levels but no condensed nuclei at all were observed in COV434 cells, and there were significantly fewer condensed OVCAR nuclei than in the DMSO control (*p* < 0.05). These differences between the two steroidogenic cell lines may reflect their different sizes, or the different steroid hormones they synthesise.

Twenty-four hours exposure to Dox + 4-Cyc (without γToc) increased ROS, reduced cell viability and generally increased the percentage of condensed nuclei in all four cell lines. The addition of γToc to Dox + 4-Cyc nearly halved chemotherapeutic-induced ROS in COV434 cells within 1 h and maintained this inhibition for 3 h but had no effect on ROS in the other three cell lines. This early inhibition of ROS in COV434 cells may have been the reason there was no increase in condensed nuclei after 24 h, and why there were significantly more viable cells after 24 h exposure to Dox + 4-Cyc + γToc than after exposure to Dox + 4-Cyc without γToc. The different responses to the combination of chemotherapeutics and γToc displayed by COV434 and the other three cell lines are unlikely to be explained by antioxidant or REDOX status and are more likely to be related to the interaction of γToc with apoptotic pathways within the cells. MCF7 and T47D cells have different apoptotic pathways, which may have resulted in significant increases in condensed nuclei in T47D but not MCF7 cells. OVCAR are resistant to Dox, and probably to 4-Cyc too, and this may account for the maintenance of viable cell numbers and lack of increase in condensed cell nuclei, despite significant increases in ROS. Nevertheless, the finding that γToc reduced COV434 ROS for 3 h and prevented nucleus condensation and loss of cell viability for 24 h, whilst stimulating ROS and nucleus condensation and decreasing T47D breast cancer cell viability during a 24 h exposure, supports our hypothesis: γToc reduced chemotherapeutic-induced ROS and associated damage in granulosa-like steroidogenic cells, whilst maintaining the anti-cancer cytotoxicity of the chemotherapeutics against breast cancer cells.
