Characterisation of cytotoxicity and DNA damage induced by the topoisomerase II-directed bisdioxopiperazine anti-cancer agent ICRF-187 (dexrazoxane) in yeast and mammalian cells

ICRF-187 sensitivity of yeast cells depends on their homologous recombination status, albeit to a lesser extent than for m-AMSA sensitivity

The products of the three genes RAD50, MRE11 and XRS2 together form the Rad50/Mre11/Xrs2 hetero-trimer protein complex that has catalytic and structural functions in many kinds of DNA metabolic processes including HR as reviewed in [18]. We observed that Δrad50, Δmre11, and Δxrs2 single knockout strains were extremely hypersensitive towards m-AMSA, while they displayed considerably less hypersensitivity towards ICRF-187 (additional file 1 and table 2).

We also tested the effect of deleting a number of genes exclusively involved in HR namely RAD51, RAD52, RAD54, RAD55, RAD57, RAD59, DMC1 and SAE2 [18]. Deleting RAD51, RAD52, RAD54, RAD55, RAD57 and SAE2 had a profound effect on the sensitivity of the yeast cells towards m-AMSA while having a smaller, but significant, effect on the sensitivity of these cells towards ICRF-187 (additional file 1 and table 2), again pointing to the HR pathway as being most important for the repair of DNA damage caused by cleavage complex stabilising drugs. We found that deleting RAD59 had no effect on drug sensitivity, confirming reported data that RAD59 only becomes functionally important in the absence of functional Rad51 protein [19]. We also observed no effect of deleting DMC1 (additional file 1 and table 2). This may be explained by the fact that Dmc1p is primarily involved in meiotic recombination [20].

NHEJ represents another DNA repair-pathway. In yeast, this repair pathway is generally less important than HR for the repair of DNA breaks [21]. In accordance with this we observed no effect of deleting the NHEJ genes YKU70 and YKU80 on the sensitivity towards m-AMSA. We did however, observe some hypersensitivity of Δyku70 cells towards ICRF-187, while Δyku80 cells were not hypersensitive (additional file 1 and table 2). This is a surprising result, because Yku70p and Yku80p have been demonstrated to play equally important roles for NHEJ activity in yeast [21]. These results suggest that the effect of deleting YKU70 is unrelated to its DNA repair functions.

The DNA binding Rad18p forms a hetero-dimer with Rad6p that is involved in post replication repair (PRR) [22]. We found that although Δrad18 cells were clearly hypersensitive towards m-AMSA, Δrad6 cells were markedly more sensitive. Δrad6 cells were actually among the most sensitive towards m-AMSA (figure 1, additional file 1 and table 2). Interestingly, the sensitivity of Δrad6 and Δrad18 cells towards ICRF-187 is indistinguishable from that of wild-type cells (figure 1). The vast difference in the sensitivity of Δrad 6 cells towards ICRF-187 and m-AMSA confirms the notion that the DNA lesions induced by these drugs are different in nature. The finding that Δrad6 cells are much more sensitive towards m-AMSA than Δrad18 cells is surprising, and may indicate that Rad6p functions unrelated to DNA repair affect cellular sensitivity towards m-AMSA. Rad6p has ubiquitin conjugating activity [22], and therefore such Rad18p-unrelated functions could involve protein degradation via the 26S proteasome pathway. To test this hypothesis, we analysed the drug sensitivity of four yeast strains with impaired protein degradation; Δubc4, Δubc13, Δdoa4 and Δqri8. These deletion strains were not hypersensitive towards m-AMSA (or ICRF-187) (additional file 1 and table 2), suggesting that Δrad6 cells are hypersensitive towards m-AMSA due to impaired PRR activity. The involvement of both HR repair and PRR in determining the sensitivity of yeast cells towards the topoisomerase II cleavage complex stabilising drugs mitoxantrone and idarubicin has previously been reported [23]. The observed lack of hypersensitivity of the Δrev1 and Δrev3 strains (additional file 1 and table 2) suggests that trans-lesion DNA synthesis plays no role in determining the sensitivity towards ICRF-187 or m-AMSA.

We also analysed the effect of deleting genes belonging to the nucleotide excision repair (NER) pathway – RAD1 and RAD14, the base excision repair (BER) pathway – APN1, and the mismatch repair (MMR) pathway – MLH1, PMS1, MSH2 and MSH3. None of these deletions caused cells to become more sensitive towards ICRF-187 and m-AMSA, indicating that these pathways are not involved in repairing DNA damage induced by these drugs. Interestingly, deleting genes involved in the MMR and NER pathways caused cells to become somewhat resistant to ICRF-187, and to a lesser extent towards m-AMSA (additional file 1 and table 2). The products of these genes have been implicated to have anti-recombination activities [24]. Increased levels of recombination in these cells could therefore be responsible for the observed low-level resistance towards ICRF-187.

Deletion of DNA damage checkpoint genes has little effect on both ICRF-187 and m-AMSA sensitivity of yeast cells

While deleting genes involved in DNA repair caused cells to be hypersensitive towards both drugs tested, we observed little effect of deleting the DNA damage checkpoint genes MEC3, DDC1, RAD17, TEL1, RAD9 and CHK1 (additional file 1 and table 2). Our finding that checkpoint control regulation plays no important role for bisdioxopiperazine sensitivity supports earlier data showing that arresting yeast cells in G1 phase did not protect against ICRF-193 cytotoxicity [7]. The lack of importance of checkpoint function in determining sensitivity towards m-AMSA is also in accordance with published observations [23], where sensitivity towards the cleavage complex stabilising topoisomerase II poisons mitoxantrone, idarubicin, daunorubicin and doxorubicin were only marginally affected by inactivating the RAD9, RAD17, MEC1, and RAD53 genes, while the sensitivity of yeast cells towards the topoisomerase I poison camptothecin showed a strong dependency on these pathways.

ICRF-187 is a possible substrate for the Pdr5 ABC transporter in yeast

In mammalian cells resistance towards various structurally-unrelated anti-neoplastic agents is often associated with over-expression of ABC-type drug efflux transporters such as p-glycoprotein and multi-drug resistance protein (MRP) as reviewed in [25]. Among the three yeast ABC transporters assessed in our study, Pdr5 is by far the best characterised [26]. While deleting YOR1 and ATR1 had no effect on drug sensitivity, Δpdr5 cells were clearly hypersensitive towards ICRF-187 but not towards m-AMSA (additional file 1 and table 2), suggesting that ICRF-187 is a substrate for the Pdr5 pump in yeast. It has to be emphasized, that over-expression of drug efflux pumps has not been associated with resistance towards bisdioxopiperazines in mammalian cells.

Transcriptional profiling of yeast cells after exposure to equitoxic concentrations of ICRF-187 and m-AMSA

Genes induced by both drugs

Both compounds induced the expression of a number of genes known to be up-regulated by DNA damage. The expression of four well-established DNA-damage inducible genes; RNR2, RNR3 [27, 28] and RAD51, RAD54 [29] was thus induced by both drugs. Both compounds also stimulated the expression of HUG1 recently shown to be up-regulated by DNA damage and replication arrest [30] (See Table 3, additional files 3 and 4). Furthermore, both drugs stimulated expression of the stress-inducible XBP1 gene whose protein product is a transcription factor. XBP1 expression is reportedly induced in response to heat shock, high osmolarity, oxidative stress, glucose starvation and DNA damage, and induces a slow-growth phenotype with lengthening of the G1 cell cycle phase [31]. The PCL9 gene product has cyclin-dependent protein kinase regulator activity suggesting a role for Pcl9p in cell cycle regulation [32]. Repression of PCL9 by exposure to both drugs (table 3, additional files 3 and 4) may thus be indicative of drug-induced cell cycle arrest in accordance with the XBP1 expression data. Finally, both drugs induced the expression of general stress-induced HSP genes as expected (table 3, additional files 3 and 4). Although expression of the RNR3 and HUG1 genes was up-regulated by both drugs, pMJ1-transformed cells having RNR3 or SML1 deleted (the latter is a functional non-inducible homolog of HUG1) have wild-type sensitivity towards both drugs (table 2, additional file 1), showing that although these genes are induced by both drugs, they are probably not involved in determining their cytotoxicity.

Genes specifically induced by ICRF-187

We found that ICRF-187 specifically induced the expression of two genes encoding the ABC efflux transporters Pdr12 and Pdr15, while m-AMSA had no effect on the expression of these genes (table 3, additional files 3 and 4). Transcription of the stress-inducible WSC4 gene was likewise enhanced by exposure to ICRF-187 (table 3, additional files 3 and 4). Knocking out WSC4 in yeast cells has been found to enhance their sensitivity towards various stresses including heat, ethanol and DNA damage [33]. Recently, the SWI/SNF complex was directly shown to repress transcription in S. cerevisiae cells [34]. We found that SWI1 was specifically induced by ICRF-187 (table 3, additional files 3 and 4). Finally, we found that ICRF-187 specifically repressed the expression of the stationary phase-induced SNZ1 gene [35] (table 3, additional files 3 and 4).

Exposure of yeast cells to ICRF-187 causes less transcriptional induction of DNA damage-inducible genes than exposure to m-AMSA at equitoxic drug concentrations

To verify the array data we performed real-time PCR to assess the expression of the RNR3, HUG1, RAD51 and RAD54 genes after exposure to the two drugs using the actin gene ACT1 as internal control (figure 2). Real-time PCR confirmed induction of these established DNA damage-inducible genes by both drugs assessed. Furthermore, exposure of the cells to m-AMSA resulted in a higher level of induction than did exposure to ICRF-187 for the four genes tested, especially for HUG1. These data suggest that when yeast cells are exposed to equitoxic concentrations of the two drugs, m-AMSA generates more extensive DNA damage than ICRF-187.

ICRF-187 sensitivity of mammalian hamster cells depends on their homologous recombination status, albeit to a lesser extent than seen for m-AMSA sensitivity

The yeast clonogenic assays presented above point to an important role of HR in the repair of m-AMSA-induced DNA damage, while the importance of this pathway in the repair of ICRF-187-induced DNA damage is less so. Because HR is the major repair pathway in yeast [21], while both NHEJ and HR are important for the repair of DNA breaks in mammalian cells [36], we next turned to assess the importance of these pathways in mammalian cells having reduced levels of HR and NHEJ. In this analysis we used a panel of four hamster cell lines; AA8 cells (wild-type), irs1SF cells [37] (recombination defective caused by non-functional XRCC3), CXR3 cells [37] (recombination proficient due to ectopic expression of human XRCC3), and V3-3 cells [38] (reduced level of NHEJ due to non-functional DNA-PK_cs).

We observed a strong dependence on HR for the sensitivity towards m-AMSA (figure 3A). Thus, only 1 % relative survival was seen for the irs1SF cells (recombination defective) at 6 nM of this drug, while wild-type AA8 cells were only slightly sensitive to15 nM m-AMSA. Furthermore, ectopic expression of human XRCC3 fully reversed the m-AMSA hypersensitivity as CXR3 cells were no more hypersensitive than AA8 wild-type cells, confirming the notion that HR plays a role in the repair of topoisomerase II-induced DNA breaks in mammalian cells. We also observed that irs1SF cells were hypersensitive towards ICRF-187 (figure 3B), but the degree of hypersensitivity was much less than observed for m-AMSA, as also seen for recombination deficient yeast cells. Again, ectopic expression of the human XRCC3 homolog reversed the hypersensitivity as CXR3 cells displayed near wild-type sensitivity towards ICRF-187.

DNA-PK_cs deficient hamster cells show slightly increased sensitivity towards both m-AMSA and ICRF-187

To assess the effect of NHEJ on drug sensitivity we also employed the V3-3 cell line (DNA-PK_cs deficient, with concomitant reduced level of NHEJ). The results of these experiments are depicted in figure 3A and 3B. The V3-3 cells were slightly hypersensitive towards both drugs suggesting a role for NHEJ in the repair of DNA lesions induced by both drugs.

AA8, irs1SF, CXR3 and V3-3 cells have similar levels of topoisomerase II catalytic activity

The sensitivity of cells towards topoisomerase II directed drugs depends both on their levels of topoisomerase II catalytic activity, and on their capability to repair topoisomerase II-induced DNA damage. We therefore determined the level of topoisomerase II catalytic (DNA strand passage) activity in crude protein extracts isolated from the four cell lines used in clonogenic assays, by applying a radioactive decatenation assay. No significant difference in the level of topoisomerase II DNA strand passage activity was recorded between the four cell lines (additional file 5). This result rules out the possibility that varying levels of topoisomerase II catalytic activity in these cells is responsible for their differential drug sensitivity.

ICRF-187 induces lower levels of homologous recombination in hamster cells than m-AMSA at equitoxic concentrations

The hypersensitivity of the recombination defective irs1SF cells towards both drugs suggests that HR is involved in repairing DNA lesions induced by both drugs. To address this directly we applied a mammalian recombination assay to measure stimulation of HR by ICRF-187 and m-AMSA by using SPD8 hamster cells [39]. This assay measures the repair of a defective chromosomal hprt gene by the activity of HR. From figure 4A it is evident that both drugs stimulated the level of HR in a dose dependent manner. When recombination frequency is expressed as a function of surviving cells (figure 4C) it becomes evident that the recombination frequency increases with increasing cell mortality for both drugs tested. From figure 4C it is also evident that at equitoxic concentrations of the two drugs, m-AMSA stimulated HR to much higher levels than did ICRF-187. Thus, at 50 % survival, no induction of HR was seen with ICRF-187 (in three independent experiments), while m-AMSA caused an approximately 10-fold induction at equitoxic doses.

ICRF-187 induces only low levels of H2AX phosphorylation in human SCLC cells as compared to m-AMSA

Induction of γH2AX is a well-established marker for topoisomerase-induced DNA double strand breaks in mammalian cells [40–42]. We therefore assessed the effect of exposing human SCLC OC-NYH cells to 10 μM m-AMSA and 1 mM of ICRF-187 at increasing time points (figure 5A and 5B). Exposure to 10 μM m-AMSA quickly resulted in γH2AX induction. Thus, induction was evident after 30 min, and after 24 hours more than 10-fold induction was observed. In contrast, when cells were exposed to 1 mM ICRF-187, much less γH2AX induction was observed, and after 24 hours the level of induction was less that three-fold.

Drugs

ICRF-187 (Cardioxane, Chiron group) was dissolved in sterile water at 20 mg/ml and kept at – 80°C. To avoid hydrolysis of the drug, fresh aliquots were used for each experiment. m-AMSA (Pfizer) was diluted in DMSO and stored at – 80°C at 1 mg/ml. L-azaserine and thymidine (both from Sigma) were added directly to tissue culture medium. 6-thioguanine and hypoxanthine (both from Sigma) were dissolved in 5 M NaOH and immediately added to the tissue culture medium.

Yeast strains and constructs

BY4741 haploid Saccharomyces cerevisiae cells (MAT a his3 Δ1 leu2 Δ0 met15 Δ0 ura3 Δ0) and a panel of single-gene deletion derivatives hereof (table 1) were purchased from EUROSCARF, Institute of Microbiology, Johann Wolfgang Goethe University Frankfurt, Germany. The construction of BY4741 and its deletion derivatives have been described [65]. JN362A_t2–4 cells with the relevant genotype (MAT a ura3–52 leu2 trp1 his7 ade1–2 ISE2 top2-4) were kindly provided by Dr. John L. Nitiss, St. Jude Children's Research Hospital, Memphis TN, USA. This strain and the construct for functional expression of human topoisomerase II α in yeast pMJ1 (URA3) have been described previously [66]. All yeast strains were transformed with pMJ1 to functionally express human topoisomerase II α in a cell cycle independent fashion. BY4741 wild-type and Δrad6, Δrad50, Δrad52, Δsae2 and Δyku70 cells were also transformed with an empty URA3 vector (pYX112). The ICRF-187 and m-AMSA sensitivity of pYX112-transformed cells was assessed to assure that the drug sensitivity of the pMJ1-transformed cells (Table 2, S1) is related to the ectopic expression of human topoisomerase II α in the cells, which was the case. Transformation and selection was carried out according to standard procedures using lithium acetate cell wall permeabilisation and PEG-mediated DNA uptake by using single-stranded DNA as carrier as described [67]. Selection was done on SC-URA plates. Three independent pMJ1-transformed yeast clones were selected and propagated for each transformation. All strains were propagated at 30°C, to be subsequently used in clonogenic assays at 34°C.

Yeast clonogenic assay

The clonogenic sensitivity of the yeast cells towards ICRF-187 and m-AMSA was determined using a clonogenic assay essentially performed as described in [7]. Briefly, overnight cultures of the strains were grown in SC-URA medium at 34°C at 200 rpm. Cells in log phase were diluted to 2 × 10⁶ cells/ml in pre-warmed YPD medium, and 3 ml cultures were exposed to different concentrations of drug at 34°C for 22.5 hours. After drug exposure the samples were diluted up to 10⁵ times (depending on the combination of strain and drug used) in distilled sterile water. Yeast cells that were not diluted before plating were spun down by brief centrifugation, and re-suspended in the same volume of sterile water. Next, 200 μl of diluted cells were transferred to SC-URA plates, which were incubated for 5 days at 30°C before counting. 200 to 600 colonies were typically counted for each drug concentration in each single experiment. Finally, the relative survival at the different drug concentrations as compared to the no drug sample was calculated to generate dose-response curves. For each combination of yeast strain and drug, at least three dose-response curves were generated using pMJ1-transformed cells from at least two independent clones (mostly from three).

Yeast microarray gene expression analysis

Microarray experiments were performed with yeast strain JN362_t2–4 transformed with pMJ1 to functionally express human topoisomerase II α. Fresh colonies were inoculated into YPD medium and grown overnight at 34°C, 180 rpm. The cultures were then diluted to obtain an OD₆₀₀ of 0.2. Cultures of 50 ml in YPD medium were first grown for two hours to assure exponential growth of the cells. 1 mg/ml ICRF-187 or 50 μg/ml m-AMSA (equitoxic concentrations) were then added to the cell cultures (a no-drug sample was also included), and the cells were grown for an additional two hours. Each treatment was performed in duplicate. The used concentration of both drugs resulted in a reduction in the clonogenecity of the cells of 50 %. After treatment cells were harvested by centrifugation. Total RNA was isolated by the hot acidic phenol method [69]. All the steps for cDNA synthesis, cRNA synthesis, biotin labeling and array hybridization to Affymetrix S98 yeast arrays were performed as described in the Affymetrix GeneChip Expression Analysis Technical Manual (Affymetrix), and performed at the microarray core facility at Rigshospitalet, Copenhagen Denmark. Briefly, cDNA was synthesized from 5 μg RNA using a (dT)₂₄ primer containing a T7 RNA polymerase promoter sequence and SuperScript II reverse transcriptase (Invitrogen) for 1 h at 42°C followed by second-strand synthesis using DNA polymerase I and RNase H digestion followed by isolation of cDNA using GeneChip Sample Cleanup Module (Affymetrix). The cDNA was used as template for synthesis of biotin-labeled cRNA by incubation with biotin-labeled ribonucleotides and T7 RNA polymerase for 5 h at 37°C. Biotin-labeled cRNA was purified using GeneChip Sample Cleanup Module. Biotinylated cRNA was fragmented and 15 μg used for hybridization to Affymetrix Yeast Genome S98 arrays at 45°C for 16 h as described in the Affymetrix users' manual. Washing and array staining with streptavidin-phytoerythrin were performed using the GeneChip Fluidics Station 400 and scanning was performed with a Gene Array Scanner G25 (Agilent technology). Data was analyzed using the DNA-Chip Analyzer (dChip) software [70].

Real-time PCR analysis

The RNA preparations used for microarray analyses were also used for real-time PCR. This analysis was performed on an ABIPrism 7900HT (Applied Biosystems). RNA samples were DNase treated using the DNA-free™ DNase treatment and removal kit (Ambion), and RNA concentrations were measured before conversion to cDNA using the TaqMan RT kit (Applied Bioscience). Priming was performed by random hexamers converting 2 μg RNA pr 100 μl reaction volume, to make 20 ng/μl cDNA. Primers were designed for coding sequences from the Saccharomyces genome database http://www.yeastgenome.org using the Primer 3 input program http://www.broad.mit.edu/cgi-bin/primer/primer3_www.cgi. All primers were purchased at DNA Technology A/S, with melting temperatures close to 60°C. Reaction mixtures containing the following components at the indicated end-concentrations were prepared. To make a total of 40 μl in sterile water, 20 μl 1x SYBR^® green PCR master mix (Applied Biosystems), 250 nM forward primer, 250 nM reverse primer, and 5 ng template was mixed. Cycling conditions: 95°C for 10 min, followed by 40–45 cycles of 95°C for 15 s and 60°C for 60 s. Relative values of gene expression were calculated with untreated samples as calibrator, and normalized to levels of actin, according to the 2^-ΔΔCt method [71] and (User bulletin #2, AbiPrism 7700 Sequence Detection System, Applied Biosystems) after primer optimisation and target efficiency evaluation. The following primers were used:

HUG1-forward, AGGCCTTAACCCAAAGCAAT;

HUG1-reverse, TCTTGTTGACACGGTTGCTC;

RNR3-forvard, ATGCATCTCCAGTTCCATCC;

RNR3-reverse, GGGGCAACACTATCTTCCAA;

RAD51-forward, GTGGCGGTGAAGGTAAGTGT;

RAD51-reverse, GTCTAATCCGAACCGCTGAG;

RAD54-forward, CTAAAGCAGGTGGGTGTGGT;

RAD54-reverse, CTTGTTGATCAGCAGCAGGA;

ACT1-forward, CGGTGATGGTGTTACTCACG;

ACT1-reverse, GGCCAAATCGATTCTCAAAA.

Mammalian cells

The CHO cell lines AA8, irs1SF, CXR3, V3-3, and the hamster lung fibroblast cell line SPD8 were kindly provided by Dr Thomas Helleday, University of Sheffield, UK. AA8 is a wild-type cell line. The AA8-derived irs1SF cell line is XRCC3-defective and has reduced levels of HR [37]. CXR3 is a human-XRCC3-cosmid complemented strain of irs1SF, which is proficient in HR [37]. The V3-3 cell line is DNA-PK_cs-deficient and consequently deficient in NHEJ [38]. The SPD8 cells carry a non-functional hprt gene that can be repaired by HR [39]. HPRT⁺ cells can then be selected on HAsT medium containing hypoxanthine, L-azaserine and thymidine. When SPD8 cells were not used in the recombination assay they were propagated in medium supplemented with 6-thioguanine to select against spontaneous reversion to the HPRT⁺ phenotype. Human SCLC OC-NYH cells have been described [72]. Hamster cells were propagated in DMEM medium and OC-NYH cells were propagated in RPMI-1640 medium. All cell culture media were supplemented with 10 % fetal calf serum and 100 U/ml penicillin-streptomycin. Cells were grown in a humidified atmosphere containing 5 % CO₂ in the dark at 37°C.

Determination of topoisomerase II activity in crude cell extracts

Topoisomerase II activity in crude extract was determined by using a decatenation assay previously described [68]. Briefly, 200 ng ³H labeled kDNA isolated from C. fasciculata was incubated with increasing amounts of crude extracts in 20 μl reaction buffer containing 10 mM TRIS-HCl pH 7.7, 50 mM NaCl, 50 mM KCl, 5 mM MgCl₂, 1 mM EDTA, 15 μg/ml BSA and 1 mM ATP for 20 min at 37°C. After addition of 5x stop buffer (5 % Sarkosyl, 0.0025 % bromophenol blue and 50 % glycerol), unprocessed kDNA network and decatenated DNA circles were separated by filtering, and the amount of unprocessed kDNA in each reaction was determined by scintillation counting. The amount of crude extract required to fully decatenate 200 ng of kDNA under these assay conditions (which is equivalent to 1 U of catalytic activity) was then determined, and the specific activity of the crude extract was calculated as U/μg protein.

Mammalian clonogenic assay

Four hours prior to continuous treatment with either ICRF-187 or m-AMSA, 250 cells of each of the hamster cell lines were plated onto 100 mm dishes. After 7 days colonies were fixed and strained in methylene blue in methanol (4 mg/ml), and colonies with more than 50 cells were counted. Finally, the relative survival compared to the no drug treatment was calculated, and plotted against drug concentrations to generate dose-response curves. 150 – 250 colonies were typically counted in the "no drug" dishes.

Mammalian homologous recombination assay

A mammalian recombination assay was performed as described [39]. Briefly, 1 × 10⁶ SPD8 cells were inoculated into 75 cm² flasks. When transferred, 6-thioguanine was omitted from the medium. Cells were trypsinised and resuspended in 10 ml medium at 100,000 cells/ml, and exposed to the indicated drug concentrations for 24 hours. To determine clonogenic survival, for each drug-treatment 500 cells were transferred to each of two 100 mm petri dishes containing 10 ml of non-selecting medium and the cells were cultured for 7 days. For selection of recombination events, 300,000 cells were transferred to each of three 100 mm petri dishes containing 10 ml medium supplemented with 50 μM hypoxanthine, 10 μM L-azaserine and 5 μM thymidine and selection was carried out for 10 days. Colonies were fixed by using methylene blue in methanol (4 mg/ml) and counted. Finally, the recombination frequency was determined as the plating efficiency in recombination selective medium divided by the plating efficiency in normal medium, for all concentrations of ICRF-187 or m-AMSA. To enable comparison of recombination frequency at equitoxic levels of m-AMSA and ICRF-187, the recombination frequency was plotted against the relative clonogenic survival of cells receiving only drug.

Histone purification

Human SCLC OC-NYH cells were grown to sub-confluence and histones were extracted as follows. After the relevant drug treatments, the cells were pelleted and washed in cold PBS, and lysed in lysis buffer (10 mM TRIS-HCl pH = 6.5, 50 mM Sodium Bisulphate, 1% Triton X-100, 10 mM MgCl, 8.6% sucrose) at 4°C by applying 20 strokes in a tight fitting Dounce homogenizer. Released nuclei were pelleted by centrifugation at 2500 g for 10 min at 4°C, and washed in lysis buffer followed by wash buffer (10 mM TRIS-HCl, 13 mM EDTA pH 7.4). The pellet was next resuspended in 100 μl ice-cold 0.4 M H₂SO₄, and incubated for 1 hour at 4°C prior to centrifugation. The supernatant was transferred to a clean tube and 1 ml ice-cold acetone was added followed by incubation overnight for histone precipitation. After centrifugation, the pellet was air-dried and resuspended in 40 μl H₂O, and the protein concentration was determined by Bradford protein assay (Bio-Rad Laboratories)

γH2AX western blot

Western blotting was performed by loading 10 μg of total histones on a 4–12% gradient gel (NuPageTM Bis-Tris Gel, Invitrogen). Separated proteins were transferred to nitrocellulose membranes (Bio-Rad) which were blocked in 10% skimmed milk (Fluka) and incubated overnight with anti γH2AX primary antibody diluted 1:500 (Upstate Technology, cat no 16–193) followed by detection with goat-anti-mouse (Amersham) 1:2000 for one hour. Detection with ECL Plus™(Amersham) was performed by scanning on STORM™ 840 (Molecular Dynamics Inc), on which the image was optimized and bands quantified by Image Quant™ version 5.0 (Molecular Dynamics).