**3. Results**

The numbers of DAPI-stained nuclei with normal regular morphology in the images of cell culture medium control wells were T47D (415 ± 68, Figure 1I) > MCF7 (312 ± 18 Figure 1G) > OVCAR (173 ± 32) > COV434 (53 ± 19) and there were similar numbers of nuclei with normal morphology in the 0.8% DMSO controls. The OVCAR cells were derived from ovarian surface epithelial cells, and the T47D and MCF7 cells were derived from mammary epithelial cells. The normal DAPI-stained nuclei of these three cell lines in control wells were a similar size and morphology, with regular round or oval shapes (Figure 1E,G,I). The COV434 cells were derived from a granulosa cell tumour, and their nuclei were much smaller and often shaped like a croissant (crescent, Figure 1C). DAPI-intense 'bright' objects were observed in two forms: a relatively low number of apoptotic bodies (Figure 1A,D) and much higher numbers of what appeared to be small, irregularly shaped condensed nuclei (Figure 1A,H). Some nuclei were clearly not 'normal', but neither were they bright nor small enough to qualify as 'condensed'. These were labelled 'uncertain' and were frequently observed in the COV434 cell line (8.9%, Figure 1B), but were found less frequently in the MCF7 (2.4%), OVCAR (1.4%) and T47D cells (1.2%).

**Figure 1.** Cell nucleus 6-diamidino-2-phenylindole (DAPI) staining and scoring: COV434 (**A**–**D**), OVCAR (**E**,**F**), MCF7 (**G**,**H**) or T47D (**I***,***J**) cells adhered to the glass microscope slides for 24 h before exposure to doxorubicin (Dox, **J**) or 4-cyclophosphamide (Cyc, **H**) or α Tocopherol (αT, **F**) or γ Tocopherol (γT, **B**) or Dox + 4-Cyc + αToc (**D**) or Dox + 4-Cyc + γToc (**A**) or cell culture medium as a control for the chemotherapeutics (**G**,**I**), or 0.8% DMSO as a control for treatments containing tocopherols (**C**,**E**) were stained with DAPI before a fluorescence microscope was used to obtain digital images. The experiment was repeated on three separate occasions (*n* = 3) for each of the four cell types. Representative examples in each image shown of nuclei with normal morphology (solid circles), condensed DAPI-bright nuclei (dotted circles) or groups of apoptotic bodies (dotted circles and arrow) and nuclei with uncertain status (broken dash and dot circles). Scale bars 100 μm.

Although the numbers of nuclei with normal morphology were similar in cell culture medium and in medium containing 0.8% DMSO in all four cell lines (representative examples shown in Figure 1), a 24 h culture in 0.8% DMSO caused significantly more condensed COV434 nuclei than medium control (*p* < 0.05, Figure 2A), whereas the percentages of condensed nuclei were similar in media and DMSO controls in the other three cell lines (Figure 2). 4-Cyc significantly increased the percentage

of condensed COV434 nuclei (*p* < 0.05) compared to the cell culture medium control. The COV434 cell nuclei in control cell culture media were smaller than the nuclei of the other three cell lines, but exposure to the combination of Dox + 4-Cyc + αToc (Figure 1D) caused the nuclei which most resembled 'normal' to become even smaller. Nuclei exposed to Dox + 4-Cyc + αToc were difficult to score because it was not clear if they should be placed in the condensed or uncertain categories. It was clear, however, that there were very few normal nuclei; only 9.6 ± 1.5 normal (Figure S1A) and 12.3 ± 7.8 condensed nuclei (53 ± 16%, Figure 2A) per image, as opposed to 59.5 ± 3.5 normal nuclei in the COV434 DMSO control images (Figure 1C). Although exposure to αToc alone resulted in the same proportions of condensed nuclei as in the DMSO control, the addition of αToc to the chemotherapeutics significantly increased the percentage of condensed nuclei compared to the DMSO control and to Dox + 4-Cyc (*p* < 0.001, Figure 2A). Even though the COV434 DAPI-stained nuclei were difficult to score, it was still clear that αToc increased the cytotoxic activity of Dox + 4-Cyc.

Conversely, γToc reduced the background levels of the condensed COV434 nuclei to zero (Figure 2A) and the addition of γToc to the chemotherapeutics maintained the number of normal nuclei at 48 ± 9 (Figure 1B and Figure S1A), hence there were no statistical differences between the percentages of condensed nuclei in DMSO control and Dox + 4-Cyc + γToc treated COV434 cells (Figure 2A).

Neither of the chemotherapeutics affected the percentages of condensed MCF7 nuclei (Figure 2B). MCF7 cells exposed to αToc for 24 h had 375 ± 101 normal nuclei and 21 ± 4 condensed nuclei (6 ± 2.7%, Figure 2B), whereas exposure to Dox + 4-Cyc + αToc resulted in only 90 ± 20 normal nuclei (Figure S1D) and 0.48 ± 0.6% condensed nuclei (Figure 2B). This reduction in condensed nuclei occurred in the context of cell loss indicative of cell death earlier in the 24 h exposure and was unlikely to be caused by αToc protecting MCF7 against the cytotoxic effects of Dox + 4-Cyc. MCF7 cells cultured in DMSO control conditions or exposed to Dox + 4-Cyc + γToc were like the COV434 cells in that there were no significant differences in the numbers of normal nuclei, nor the percentages of condensed nuclei.

Dox + 4-Cyc increased the number of condensed T47D nuclei (*p* < 0.01) compared to the medium control, and the addition of each of the tocopherols to Dox + 4-Cyc increased the percentages of condensed nuclei compared to the DMSO control (*p* < 0.05, Figure 2C). Addition of the tocopherols to the chemotherapeutics also increased the numbers of normal nuclei; there were 253 ± 13 normal nuclei after exposure to Dox + 4-Cyc, but 503 ± 290 and 650 ± 73 normal nuclei after exposure to the chemotherapeutics combined with αToc and γToc, respectively (Figure S1C).

The combination of Dox + 4-Cyc also increased the number of condensed OVCAR nuclei (*p* < 0.01, Figure 2D). Although γToc reduced the number of condensed nuclei compared to the DMSO control (*p* < 0.05), when the tocopherols were combined with Dox + 4-Cyc neither affected the proportions of condensed (Figure 2D) or normal nuclei (Figure S1B).

Crystal violet stains DNA. The proportion of condensed nuclei in all cases except one (COV434 Dox + 4-Cyc + αToc) was lower than 10% (Figure 2). It is therefore reasonable to assume that at least 90% of the crystal violet staining was indicative of viable cells containing nuclei with normal morphologies. Exposure to 0.8% DMSO, the vehicle for both tocopherols, did not affect the amount of crystal violet staining in any cell line, for example, there were 33,837 ± 1642 T47D cells after 24 h in medium control and 37,897 ± 495 in control medium containing 0.8% DMSO (Figure S1). Although 20,000 cells were initially seeded into the 96 well plates, after a total of 48 h in control conditions there were fewer COV434 cells than the other three cell lines—22,948 ± 1567 COV434 cells per well (Figure S1).

The MCF-7 derived EC25 values for 4-Cyc reduced viable cell numbers by approximately 25% in the present study. MCF-7 viable cell numbers were reduced to 68 ± 9% of control, T47D to 71 ± 2%, COV434 to 66 ± 6% and OVCAR to 61 ± 15% (Figure 3). Cell line sensitivity to the cytotoxic effects of Dox was slightly different. COV434 cells were the most sensitive (57 ± 2%), whereas the three epithelial cell lines had similar sensitivities, ranging from 67 ± 6% (T47D) to 75 ± 5% (OVCAR). The combination of the EC25 value for 4-Cyc with the EC25 value for Dox was expected to cause death of 50% of the cells, but was less cytotoxic to MCF-7 (76 ± 14% of control) and T47D (61 ± 4%) but more

cytotoxic to COV434 cells (34 ± 5% of control). For OVCAR cells, the cytotoxicity caused by the two chemotherapeutics was additive; the combination reduced viable OVCAR cell numbers to 57 ± 11% of control.

**Figure 2.** DAPI-stained condensed cell nuclei: COV434 (**A**), OVCAR (**B**), T47D (**C**) or MCF7 (**D**) cells adhered to glass microscope slides for 24 h before a 24 h exposure to doxorubicin (Dox) or 4-cyclophosphamide (Cyc) or α Tocopherol (αToc) or γ Tocopherol (γToc) or combinations of these. Cell culture medium was a control for the chemotherapeutics and 0.8% DMSO in culture medium was a control for tocopherols. Cells were stained with DAPI before a fluorescence microscope was used to capture digital images. The numbers of normal and condensed (includes groups of apoptotic bodies) nuclei were scored in each image, and the condensed nuclei expressed as a percentage. The experiment was repeated on three separate occasions and the mean ± stdev of percentages is shown. Data were subjected to 1 Way ANOVA with a Tukey post-test \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001.

**Figure 3.** Effect of chemotherapeutics and tocopherols on cell viability and reactive oxygen species (ROS). Human COV434 (**A**), OVCAR (**B**), T47D (**C**) and MCF7 (**D**) cells were cultured for 24 h before being loaded with DCFDA and exposed to Doxorubicin (Dox) or 4-Cyclophosphamide (Cyc) or α Tocopherol (αToc) or γ Tocopherol (γToc), or combinations of these, for 24 h in triplicate wells. ROS were measured before measuring the number of adherent cells per well in a crystal violet assay (Cells). The average cell number, or ROS, obtained from triplicate wells, was expressed as a percentage of control for the same experimental replicate. Controls were culture medium, or culture medium containing DMSO which was the vehicle for tocopherols. Data are shown as mean ± stdev of percentages obtained in three independent experiments (*n* = 3). The original 'Cells' or 'ROS' data (i.e., not percentages) were subjected to 1 Way ANOVA with a Tukey post-test. Within either Cells or ROS, significant difference from control \* *p* > 0.05, \*\* *p* > 001, \*\*\* *p* > 0.001, oravb *p* > 0.05,bvc *p* < 0.01,avc *p* > 0.001.

A cytotoxic dose of αToc was not found, and the relatively high non-cytotoxic concentration of αToc used in the present study was not cytotoxic to any cell line (Figure 3). The MCF7-derived EC25 value of γToc reduced T47D cell viability to 64 ± 9% (*p* < 0.01, Figure 3C) and MCF7 cells to 70 ± 14% of DMSO control but had no effect on the viability of COV434 or OVCAR cells (Figure 3). γToc interacted with the chemotherapeutics and COV434 cells such that Dox + 4-Cyc reduced viable COV434 cell numbers to 34 ± 5% of medium control (*p* < 0.001), but when γToc was added to the combined chemotherapeutics COV434 cell viability was reduced to 54 ± 4% of the DMSO control (*p* < 0.01). Hence, when compared to Dox + 4-Cyc, γToc conferred a significant protective effect against the cytotoxicity caused by the chemotherapeutics (*p* < 0.05). However, γToc did not affect the cytotoxicity of the combined chemotherapeutics in the other three cell lines.

Although 0.8% DMSO, the vehicle control for tocopherols, had no effect on cell viability (Figure S1), COV434, OVCAR and T47D cells cultured in 0.8% DMSO generated significantly more ROS than in culture medium (*p* < 0.001, Figure 4), whereas MCF7 cells generated similar amounts of ROS in the two control media (Figure 4).

**Figure 4.** Reactive oxygen species (ROS) generation under in vitro control conditions. Cells were cultured for 24 h before being loaded with DCFDA, then cultured for 24 h in either cell culture medium (Medium Control) or cell culture medium containing 0.8% DMSO (DMSO Control). Relative Fluorescent Units (RFU per culture well) indicative of reactive oxygen species (ROS) generated by the cells in that well were measured using a 96-well plate spectrofluorometer. Data are shown as mean ± stdev of three independent experiments (*n* = 3). The RFU per well values were analysed by 2-Way ANOVA with Bonferroni post-test. Within each cell line, significant difference between two controls: NS not significant, \*\*\* *p* > 0.001.

After 24 h exposure, 4-Cyc caused all four cell lines to generate more ROS than Dox (Figure 3). The addition of the MCF7-derived EC25 4-Cyc to the EC25 Dox did not double ROS compared to each chemotherapeutic alone, but ROS levels after exposure to the combined chemotherapeutics were always higher than after exposure to 4-Cyc alone. Neither αToc nor γToc affected ROS production by COV434, OVCAR or T47D cells, but it was surprising that each tocopherol significantly increased ROS in MCF7 cells (*p* < 0.001 Figure 3D). The addition of tocopherols to Dox + 4-Cyc was unable to prevent ROS generation by MCF7, T47D or COV434 cells, but αToc reduced and γToc completely prevented the chemotherapeutic-stimulated increase in ROS.

Although ROS levels did not change in cell culture medium, ROS levels increased in the cell culture medium containing 0.8% DMSO during the first 3 h of culture (Figure 5). Acute, time-dependent, significant increases in ROS levels were detected in MCF-7, T47D and OVCAR-3 cells during the first 3 h exposure to the MCF-7 EC25 value (21.23 μM) of 4-Cyc (Figure 5). Dox caused a lower, but still significant increase in ROS production by the same cell lines. In COV434 granulosa cells, ROS levels increased after 1 h exposure to Dox. Unlike the other cell lines, 4-Cyc and Dox stimulated the same amount of ROS during the first 3 h; the amount of ROS generated by 4-Cyc was lower than in the other three cell lines (Figure 5). α and γToc did not stimulate ROS generation after 1, 2 or 3 h in any of the cell lines.

**Figure 5.** Effect of 3 h exposure to chemotherapeutics and tocopherols on ROS production. (**A**) COV434, (**B**) OVCAR, (**C**) T47D, and (**D**) MCF-7 cells were loaded with DCFDA then exposed to Doxorubicin (Dox), 4-Cyclophosphamide (4-Cyc), α Tocopherol (aToc) or γ Tocopherol (gToc) or combinations of these for 3 h at concentrations that reduced MCF-7 viability by 25% (EC25). Fluorescence was read every hour for 3 h. Means ± SD of three independent experiments are shown. Relative fluorescent units (RFU) were subjected to Two-way ANOVA with Bonferroni post-hoc test \* *p* ≤ 0.05, \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001 significant difference from same exposure control, bars show significant difference compared to combination of Dox and 4-Cyc at same exposure.

The combination of Dox + 4-Cyc also caused time-dependent, significant increases in ROS within the first 3 h of exposure (Figure 5). The addition of either tocopherol to the combination of chemotherapeutics had no effect on acute ROS generation by three of the cell lines, but significantly reduced ROS generation by COV434 cells. γToc was more effective than αToc at reducing Dox + 4-Cyc generated ROS in COV434 cells within the first 3 h of exposure.
