Resistance to Bleomycin in Cancer Cell Lines Is Characterized by Prolonged Doubling Time, Reduced DNA Damage and Evasion of G2/M Arrest and Apoptosis

Background To establish, characterize and elucidate potential mechanisms of acquired bleomycin (BLM) resistance using human cancer cell lines. Seven BLM-resistant cell lines were established by exposure to escalating BLM concentrations over a period of 16-24 months. IC50 values and cell doubling times were quantified using a real time cytotoxicity assay. COMET and γ-H2AX assays, cell cycle analysis, and apoptosis assessment further investigated the mechanisms of BLM resistance in these cell lines. Results Compared with parental cell lines, real time cytotoxicity assays revealed 7 to 49 fold increases in IC50 and a mean doubling time increase of 147 % (range 64 %-352%) in BLM-resistant sub-clones (p<0.05 for both). Higher maintenance BLM concentrations were associated with higher IC50 and increased doubling times (p<0.05). Significantly reduced DNA damage (COMET and γ-H2AX assays), G2/M arrest, and apoptosis (p<0.05 for each set of comparison) following high-dose acute BLM exposure was observed in resistant sub-clones, compared with their BLM-sensitive parental counterparts. Three weeks of BLM-free culturing resulted in a partial return to BLM sensitivity in 3/7 BLM-resistant sub-clones (p<0.05). Conclusion Bleomycin resistance may be associated with reduced DNA damage after bleomycin exposure, resulting in reduced G2/M arrest, and reduced apoptosis.


Introduction
Bleomycin (BLM) is a glycopeptide antibiotic isolated from Streptomyces verticillis [1,2]. As a chemotherapeutic agent, it is used in the treatment of multiple tumors, including but not limited to testicular carcinomas, lymphomas, and head and neck cancers [3,4]. Although the full pathway of the drug's mechanism of action has not been elucidated, BLM does bind to iron and oxygen to produce reactive oxygen species (ROS) [5] that induces single-and double-strand DNA breaks, with the latter being primarily responsible for its anti-tumor effects [6,7].
It also causes lipid peroxidation and mitochondrial DNA damage [8]. Extended cell-cycle arrest/senescence, apoptosis and mitotic cell death are the most common cellular responses to BLM treatment [9].
BLM was found to induce G2/M cell cycle arrest in cancer cell lines [10,11]. This may be explained by a G2/M checkpoint response to DNA damage. The G2/M checkpoint is important for genomic stability, for it ensures that chromosomes are intact and ready for separation before cells enter mitosis [12]. Unlike the G1 checkpoint, G2/M checkpoint genes are often not mutated in cancer cells [13].
The development of BLM resistance serves as an important mechanism for the evasion of chemotherapeutic eradication in cancer cells. However, the mechanisms responsible for acquired BLM resistance in human tumor cells have not been well investigated. In this study, we established BLM-resistance in seven human cancer cell lines, including lines of tumor types currently treated with BLM and others known to be either sensitive or resistant to BLM. Moreover, we characterized these cell lines with regard to their level of BLM-resistance, BLM-induced DNA damage, doubling time, cell cycle distribution, and degree of apoptosis (before and after BLM treatment) to increase our understanding of the potential mechanisms of resistance.

Establishment of bleomycin-resistant sub-clones from parental (control) cell lines
To develop BLM-resistance, cells were continually exposed to stepwise increases in the concentration of BLM over a period of 16 to 24 months. Briefly, cells were seeded at a density of ~5 ×10 5 /ml in a T75 cell culture flask with 10ml complete growth medium. After 4-6 hours of incubation, relatively low concentrations of BLM (ranging from 0.01 to 0.1µg/ml depending on the innate BLM-sensitivity), dissolved in phosphate-buffered saline (PBS) without Ca 2+ and Mg 2+ , were added into the medium. Cells were left in BLM for 2 to 4 weeks or until a stable cell re-population formed. Regular medium replenishment was performed throughout this period. The BLM concentration was then increased by 0.5 to 2 fold. This stepwise dose escalation continued for 16 to 24 months until the BLM concentration reached at least ten times the starting concentration. Thereafter, all BLM-resistant cell lines ("BLMresistant sub-clones") were maintained in their highest achieved BLM concentration ("maintenance dose"). At the same time, regular passage of the parental cell lines were performed in parallel with the BLM-resistance establishment process.

BLM sensitivity test and doubling time analysis
Prior to the BLM resistance/sensitivity assessment (cytotoxicity assay), a cell proliferation assay was carried out on the xCELLigence system (Roche, USA) to identify the optimal conditions under which the real-time cytotoxicity assay should be running. Specifically, the proliferation assay was performed: a) to identify the optimal seeding density for the cytotoxicity assay for each of the cell lines; and b) to determine the duration of the cytotoxicity assay.
The proliferation assay was carried out by seeding various numbers of cells into a 96-well plate (E-plate 96, Roche, USA) in quadruplicate, followed by real time monitoring of cellular growth for up to 7 days. Twenty-four hours after the seeding, half of the wells on the plate were treated with BLM to determine the cellular response. The proliferation assay was repeated twice on each cell line. Optimal seeding densities for each line were selected on the basis of dramatic changes in proliferation at 72-96 hours after BLM treatment and small variations across replicate wells.
For cytotoxicity assays, the cell count was first performed, and cells were then seeded into triplicate wells with 180µl of BLM-free cell culture medium on a 96-well plate. Twenty-four hours after initial plating, 20µl of cell culture medium containing serially diluted BLM ranging from 0 to 800 µg/ml were added into each well. The number of viable live cells was monitored every 15 minutes, for a total of at least 120 hours. IC 50 (integrated software, xCELLigence system) and fold differences in IC 50 between BLM-resistant and parental (control) cell lines (IC 50 BLM-resistant / IC 50 control ) were subsequently calculated. The fastest growth period observed for each of the cell lines in the proliferation assay was isolated for doubling time determination and its percentage change was calculated using xCELLigence system software.

Comet assay assessment of BLM-induced DNA damage
BLM is known to cause DNA damage in cells [6,7]. To determine initial (baseline) and DNA strand breaks after high dose BLM expose, alkaline Comet assays (single-cell gel electrophoresis) were performed [23] for each of the parental and resistant sub-clones. Olive Tail Moment (OTM) values of one hundred cells were scored at random per slide using fluorescence microscope with KOMET 5.0 software (Kinetic Imagine).

BLM-induced γ-H2AX foci formation
DNA double-strand breaks (DSBs) triggers the cellular formation of γ-H2AX foci (phosphorylated H2AX protein) [24]. To confirm the cellular DNA damage response to BLM through the Comet assay, quantitative analysis of γ-H2AX foci formation following high dose BLM exposure was performed on a subset of four parental/resistant pairs (HOP, ACHN, NCCIT, and H322M) [25] using Phospho-Histone H2AX pSer139 Monoclonal Antibody and Alexa Flour 488-conjugated antiphospho-H2AX (BioLegend, San Diego, CA, USA). A minimum of 10000 events were counted on flow cytometer for each measurement; the intensity of γ-H2AX, which directly correlates with cytometry counts, was analyzed using Cell Quest software (BD, USA).

Cell cycle distribution analysis
Cell cycle distributions of each of pair of seven parental and resistant sub-clones were tested pre-and post-24 hours of high dose BLM exposure at ten times the resistant sub-clones' maintenance concentration. Then, cells in the mono-dispersed suspension were fixed with ethanol, followed by propidium iodide (PI) staining (PI, Sigma, USA) and analyzed using the

Annexin V/PI assay for BLM-induced apoptosis
To determine cell apoptosis pre-and post-BLM treatment, a representative subset of four parental/resistant pairs (HOP, ACHN, NCCIT, and H322M) was treated with 24 hours of highdose BLM. The cells were then stained with Annexin V-FITC and PI, and evaluated for apoptosis by flow cytometry according to the manufacturer's protocol (BD PharMingen, San Diego, CA, USA). Both early (Annexin V-positive, PI-negative) and late apoptotic cells (Annexin V-positive, PI-positive) were counted as apoptotic cells.

Statistical Analysis
To assess treatment significance or difference, paired T-tests or unpaired T-tests (depending on the experimental specifications) were performed with a two-sided significance level of 0.05. Normality assumptions were assessed via Shapiro-Wilks tests. When the normality assumption could not be held, paired or unpaired Wilcoxon rank-sum tests were performed instead. For Comet assay assessment between parental and resistant sub-clones after high-dose BLM treatment, p-values were calculated using a t statistic for nonequal variances, testing the null hypothesis of equality on log ratios. Logistic regression was performed to assess baseline G2/M distribution differences between parental and resistant sub-clones. To investigate correlation between various measures (IC 50 control , IC 50 change following high-dose BLM treatment, doubling time, and cell cycle distribution), linear regression analyses were performed. All analyses were conducted using SAS version 9.2 or SPSS version 13.0.

BLM-resistant cell lines maintained on BLM stably displayed higher IC 50 values and prolonged doubling times
All seven BLM-resistant sub-clones demonstrated greater IC 50 than their parental counterparts ( Figure 1). Cytotoxicity assays showed between 7-49 fold increases of IC 50 in BLMresistant sub-clones. A positive correlation was observed between the maintenance BLM concentration and IC 50 values (p<0.001, R 2 =0.58). After prolonged BLM exposure, cell lines with greater parental sensitivity to BLM (mean IC 50 , 0.1µg/ml) exhibited a greater increase in resistance (mean of 48 fold) compared to parental lines that were less sensitive (mean IC 50 , 9µg/ml, 15 fold; p<0.05 comparing parental sensitive to less sensitive lines).
It was observed that BLM-resistant sub-clones grew slower than their parental cell lines. Two cell lines, when maintained in higher concentrations of BLM, such as MB231 3.0 and H322M 2.5 (subscripts denote maintenance BLM concentration), also exhibited enlarged and flattened cell morphology resembling that of cell senescence compared to their parental lines, but only after many generations. In contrast, acute exposure to high doses of BLM did not result in morphological changes. The slower cellular growth was confirmed by cell doubling time calculated with the xCELLigence system. All BLM-resistant sub-clones displayed statistically significant doubling time prolongation with a mean doubling time increase of 147% (range: 64%-352%) when compared with their parental cell lines (Figure 2, p<0.05). There was no correlation between cell doubling time and IC 50 values, and none between the percentage increase in doubling time and fold increase in IC 50 .
To test the stability of BLM resistance in BLM-resistant subclones, comparisons in IC 50 values and doubling times were made between normally maintained BLM-resistant sub-clones and the same resistant sub-clones which were subsequently cultured in BLM-free medium for three weeks. After three weeks of BLM-free culturing, three of the originally resistant sub-clones (including both testicular cell lines NT2 0.1 , NCCIT 1.5 and the lung cancer cell line HOP 0.05 ) exhibited a significant IC 50 reduction ( Figure 3) and doubling time reduction ( Figure  4), when compared to regularly maintained BLM-resistant subclones. There were no statistically significant changes in IC 50 and doubling time in the remaining four lines.

BLM resistance may be dose-dependent
Given that a general correlation exists between IC 50 values and the maintenance BLM concentrations across 7 cell lines (Figure 1), the possibility of dose-dependent BLM resistance was tested. For the single cell line ACHN, IC 50 values were obtained from ACHN 0 (parental line), and its two resistant subclones, ACHN 0.1 and ACHN 0.25 . A positive correlation was found between the maintenance BLM concentrations (0, 0.1 and 0.25µg/ml) and their IC 50 values (0.01, 0.29, and 0.74µg/ml respectively, p<0.05). Moreover, a positive correlation was also observed between BLM maintenance concentrations and doubling times (0µg/ml-12hrs, 0.1µg/ml-16.5hrs, 0.25µg/ ml-23.5hrs, p<0.05). This finding was not tested or confirmed in any of the other cell lines.

BLM-resistant sub-clones had less BLM-induced DNA damage in Comet assays
Quantification of DNA damage in all seven parental/resistant pairs using Comet assay (measured in OTM) showed that prior to BLM treatment, six of the seven resistant cell lines had higher basal DNA damage compared with control (the exception was HOP 0.05 , p<0.05). This generally correlated with the prolonged basal cell doubling time observed in these resistant sub-clones. Following high dose BLM treatment, five of seven resistant sub-clones (SF 0.4 , HOP 0.1 , NT2 0.1 , NCCIT 1.5 , and H322M 2.5 ) had lower DNA damage than their parental lines. No increase in DNA damage after BLM exposure was observed in five of seven resistant lines (SF 0.4 , NT2 0.1 , NCCIT 1.5 , H322M 2.5 , and MB231 3.0 ). In contrast, all parental cell lines had greater DNA damage post-BLM than pre-BLM (p<0.05 for each comparison; Figure 5). Further, all seven parental lines displayed significantly greater DNA damage

BLM-resistant sub-clones had reduced γ-H2AX levels compared to their parental lines following high dose BLM treatment
As a second measure of cellular response to DNA damage, γ-H2AX was also assessed in a subset of four cell lines (HOP, ACHN, NCCIT and H322M). Following 24 hours of high dose BLM treatment, γ-H2AX intensities increased in all parental cell lines. In the resistant sub-clones, increased γ-H2AX intensities were only observed in two of four lines (ACHN 0.25 and HOP 0.05 , Figure 6). This is in agreement with the Comet assays. Three (HOP 0.05 , NCCIT 1.5 , and H322M 2.5 ) of the four resistant sub-clones exhibited significantly less change in γ-H2AX intensity (γ-H2AX intensity post-treatment minus pre-treatment) compared with their parental sub-clones post-BLM treatment (p<0.05). This trend was borderline significant in the fourth line (H322M 2.5 , p=0.054).

BLM-resistant cell lines had a lower percentage of G2/M arrest following high dose BLM exposure
Since cell cycle arrest at G2/M phase was a characteristic general cellular response to BLM exposure, the ability of BLMresistant sub-clones to suppress BLM-induced G2/M arrest was evaluated. As shown in Figure 7

G2/M arrest becomes prominent after 8 hours of high dose BLM treatment
To evaluate the timing of G2/M arrest after high dose BLM exposure, four cell lines (the innately BLM sensitive HOP and ACHN lines, the NCCIT testicular line and the innately BLM resistant H322M, the same lines evaluated for the γ-H2AX experiment) underwent a time course analysis. Although there was increasing G2/M arrest in both parental and BLM-resistant sub-clones (Figure 8), this arrest was most prominently seen between 8 and 12 hours after BLM treatment in parental cells.

Reduced apoptosis was found in BLM-resistant subclones
Using the same four paired sub-clones, we examined whether cells underwent apoptosis after high dose BLM treatment. After 24 hours of BLM treatment, the percentage of apoptotic cells increased in all four parental lines tested ( Figure  9). In contrast, no resistant sub-clones exhibited statistically significant increases in apoptosis following BLM treatment. This was in agreement with the Comet assay (DNA damage) analysis. Moreover, three of four resistant sub-clones (HOP 0.05 , NCCIT 1.5 , and H322M 2.5 ) exhibited significantly less increase in apoptosis (% apoptosis after minus % apoptosis before BLM treatment) compared with their parental lines following BLM

Discussion
In this study, we successfully established seven BLMresistant human cancer cell lines from commercially available cancer cell lines of various organ origins (lung, testicle, breast, kidney, as well as the central nervous system). After sudden acute, short-term exposure to BLM, these BLM-resistant subclones exhibited less DNA damage, had longer doubling times, had a lower proportion of cells in G2/M arrest, and had reduced apoptosis, when compared to their more BLM-sensitive parental cell lines. BLM-resistant cell lines were developed by progressively increasing the incubating BLM concentration over an extended period of 16-24 months. Previous studies may have utilized similar methods in cultivating BLM resistant subclones. However, few of them reached the high level of BLMresistance observed in this study (which was 7-49 fold increase in IC 50 between resistant and parental sub-clones, when compared to a 3-20 fold increase in another study [15]). Moreover, through analysis of BLM-induced DNA damage, cell cycle distribution, and percentages of apoptosis, several putative mechanisms of BLM resistance/sensitivity were evaluated.
BLM is known to cause extended G2/M arrest and/or cell apoptosis [9] in BLM sensitive cells. This can be mediated by ATM/ATR [26,27], the upstream proteins of a DNA repair and signaling pathway that triggers G2/M arrest or cell apoptosis via a range of downstream gene products such as chk1/2, cdc 25 [28], p53, and p21 WAF1/CIP1 , where the latter two are critical for sustaining G2/M cycle arrest [29]. Moreover, histone H2AX was found to be necessary for the activation of the G2/M check point [30]. Comet and γ-H2AX results revealed less BLM-induced DNA damage in the resistant lines, suggesting that the resistant subclones may have an enhanced ability to prevent and/or reduce DNA damage caused by BLM. This may be due to reduced cellular uptake of BLM, and/or enhanced BLM elimination/ detoxification, mediated by cell surface receptors or transporters [31], antioxidant molecules/enzymes that reduce BLM-generated ROS [11,32]; or enzymes that inactivate BLM [33,34]. Future studies need to further study these mechanisms.
In this study, BLM resistance also resulted in less G2/M arrest and cell apoptosis, consistent with results from a previous study on a single BLM-resistant line [11]. The slight increase in G2/M arrest at baseline (prior to high dose BLM treatment) for these BLM resistant sub-clones could be explained by chronic exposure to BLM. The fact that BLMresistant cells were able to proliferate at a BLM concentration that was not viable for their parental lines suggests that evasion of cell cycle blockage could be another mechanism of resistance. Taken together, our results suggest that BLM resistant cells may have acquired enhanced ability to prevent/ After removing maintenance BLM concentrations from all seven resistant sub-clones, there was partial reversal of IC 50 values and doubling time in three of seven cell lines. The absence of reversibility in the other four cell lines could be the result of individual cell line differences, or perhaps these cell lines needed longer breaks from BLM exposure to develop BLM sensitivity again. To our knowledge, the literature on reversing BLM-resistance in cancer cell lines has been extremely limited. This observation speaks to the potential importance of continued BLM exposure when developing resistant sub-clones. Moreover, the result suggests that reversible mechanisms such as epigenetic rather than (permanent) genetic changes may be playing a role in maintaining some of the BLM-resistance. This study has limitations. Firstly, an in vitro model may not explain resistance in patients, but remains a critical first step in studying BLM resistance mechanism. Secondly, the amount of BLM treatment for each cell line corresponded to ten times the maintenance concentration for each BLM resistant sub-clone. There is the possibility that the treatment is not "acute" or high enough to elicit significant cellular response following the treatment. However, significant responses were seen in parental cells. Thirdly, individual cell line differences were not well-accounted, given the range of cell line origins. Since each cell line may have unique mechanisms that contribute to BLM resistance, this may explain some of the variations observed across experiments in this study. Fourthly, the cells used were not monoclonal. This may result in cells behaving differently upon BLM treatment. Yet, given that it was technically difficult to repopulate cells from a single cell that survived the dose escalation, we adopted this conventional approach of resistance development.
In summary, this study described some of the relationships between BLM resistance, BLM-induced DNA damage, cell growth rate, cell cycle distribution, and apoptosis. The reduced DNA damage, reduced G2/M arrest, and reduced apoptosis observed in BLM-resistant sub-clones following high dose BLM exposure suggest that acquired BLM resistance involves effective DNA damage reduction and G2/M cell cycle evasion. The seemingly reversible resistance observed in at least some of the BLM resistant sub-clones suggests that some of the BLM-resistance in our cell lines models may have utilized non- permanent mechanisms such as epigenetic changes to cope with chronic BLM exposure. Our results provide the foundation for future research in biomarkers of BLM resistance, which may ultimately lead to an improved rationale for personalized chemotherapy selection.