BRCA2 is needed for both repair and cell cycle arrest in mammalian cells exposed to S23906, an anticancer monofunctional DNA binder

Abstract

Repair of DNA-targeted anticancer agents is an active area of investigation of both fundamental and clinical interest. However, most studies have focused on a small number of compounds limiting our understanding of both DNA repair and the DNA damage response. S23906 is an acronycine derivative that shows strong activity toward solid tumors in experimental models. S23906 forms bulky monofunctional DNA adducts in the minor groove which leads to destabilization of the double-stranded helix. We now report that S23906 induces formation of DNA double strand breaks that are processed through homologous recombination (HR) but not Non-Homologous End-Joining (NHEJ) repair. Interestingly, S23906 exposure was accompanied by a higher sensitivity of BRCA2-deficient cells compared to other HR deficient cell lines and by an S-phase accumulation in wild-type (wt), but not in BRCA2-deficient cells. Recently, we have shown that S23906-induced S phase arrest was mediated by the checkpoint kinase Chk1. However, its activated phosphorylated form is equally induced by S23906 in wt and BRCA2-deficient cells, likely indicating a role for BRCA2 downstream of Chk1. Accordingly, override of the S phase arrest by either 7-hydroxystaurosporine (UCN-01) or AZD7762 potentiates the cytotoxic activity of S23906 in wt, but not in BRCA2-deficient cells. Together, our findings suggest that the pronounced sensitivity of BRCA2-deficient cells to S23906 is due to both a defective S-phase arrest and the absence of HR repair. Tumors with deficiencies for proteins involved in HR, and BRCA2 in particular, may thus show increased sensitivity to S23906, thereby providing a rationale for patient selection in clinical trials.

Abbreviations

ATR

Ataxia telangiectasia- and RAD3-related

DSBs

Double Strand Breaks

FA

Fanconi Anemia

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

HR

Homologous Recombination

HU

Hydroxyurea

ICLs

Inter-strand Crosslinks

NER

Nucleotide Excision Repair

NHEJ

Non-Homologous End-Joining

TCR

Transcription-Coupled Repair

UCN-01

7-hydroxystaurosporine.

Introduction

Cytotoxic agents remain the backbone of most, if not all, drug combinations used for cancer treatment, and there is a real need for the identification of novel cytotoxic compounds with different mechanisms of action. “Alkylating agent” is a generic term englobing different subclasses of widely used anticancer drugs which share the capacity to form covalent DNA adducts. This includes true alkylators like temozolomide, cross-linking agents like cisplatin and mitomycin C and, more recently, compounds forming bulky, monofunctional DNA adducts.^1,2 The latter group includes many compounds derived from natural sources such as irofulven, trabectedin (ecteinascidin-743, ET-743, yondelis), hedamycin and S23906.³ Unexpectedly, mechanistic studies have revealed many differences in the way these compounds react with the repair machinery and other macromolecular complexes. For example, irofulven and structurally related compounds are the only agents identified so far that are exclusively repaired by transcription-coupled repair (TCR) without recognition by the global genome repair.^4–7 While trabectedin is also exclusively recognized by TCR, this interaction is accompanied by increased, rather than decreased, cytotoxicity, probably due to formation of non-functional toxic DNA repair intermediates.^8–12 Finally, the cytotoxic effects of hedamycin are mediated through induction of secondary DNA damage in form of double strand breaks (DBSs), apparently in a replication-independent manner.^13,14 A better understanding of the factors controlling the induction and removal of DNA damage by these compounds may not only facilitate their clinical use but would also increase our general understanding of the structural and biological features governing adduct processing in mammalian cells.

S23906 is an anticancer agent derived from acronycine, an acridone alkaloid originally extracted from the bark of an Australian scrub.¹⁵ S23906 forms bulky adducts with the exocyclic amino group of guanine residues in the minor groove of DNA.¹⁶ A unique property of S23906, compared to other DNA-targeting anticancer agents, is its “helicase-like” activity which leads to the destabilization of the duplex structure in the vicinity of the adducts, thereby facilitating base unpairing and helix opening.¹⁷ Accordingly, S23906-induced bulky adducts are quickly processed through Nucleotide Excision Repair (NER) and acquired resistance to S23906 is associated with increased NER activity.¹⁸ S23906 shows remarkable activity in preclinical models^19,20 and has undergone phase I/II clinical trials. Unfortunately, its clinical development was recently arrested due to limited activity in most patients. The clinical failure exemplifies the need for identifying strong biomarkers during preclinical development in order to reduce drug attrition rate. Indeed, biomarkers should help, even in early clinical trials, to select patients with a defined profile likely to respond to a novel anticancer drug. This approach is expected to be more efficient than enrolling patients in clinical trials independently of their genetic or phenotypic status. This strategy has been successful in the development of trabectedin. Indeed, the DNA repair pathways identified as playing an important role in trabectedin cytotoxicity in preclinical models^8,10,12,21 have permitted improved patient outcomes in the clinic.^22,23 Thus, elucidating the precise mechanism of action and the molecular/cellular consequences of chemotherapeutic agents should turn “conventional” DNA alkylating agents into targeted therapies toward predictable sensitive tumors thereby improving their therapeutic index (toxicity to cancer cells versus normal cells).²⁴

Today, newly synthesized acronycine derivatives are under preclinical investigation.²⁵ Interestingly, among the new acronycine serie, monoesters of the pyranediol were the most potent derivatives exhibiting sub-micromolar activities. Intriguingly, the most promising compound corresponded to the aza derivative of S23906.²⁵ Thus, identifying strong biomarkers to predict S23906 efficacy should help the rational development of its promising derivatives.

To that end, we have previously shown that both XPC and CSB, and, to a lesser extent XPA, but not ERCC1, which are all involved in the NER machinery, can serve as biomarkers for predicting the sensitivity of tumor cells to bulky monofunctional alkylators like S23906.¹⁸ In the absence of functional NER, unrepaired S23906 adducts are converted into toxic DSBs, coherent with the marked influence of NER on cell survival.^18,26 Interestingly, DSBs formation induced by S23906 exposure is mainly observed in S phase and is inhibited by aphidicolin suggesting a role of DNA synthesis in the conversion of DNA adducts into highly toxic lesions.²⁶ In agreement, a recent work showed that the S23906-induced γ-H2AX foci were associated with sites of DNA synthesis, suggesting that the early induction of the DNA damage is limited to actively replicating DNA regions.²⁷ S23906 exposure induced a rapid and specific recruitment of RPA, suggesting that ssDNA regions are formed at the stalled replication forks. Consistent with the formation of ssDNA, the initial DNA-damage response is mainly ATR-dependent and the activation of the ATR/Chk1 pathway is accompanied by an early S-phase arrest whose disruption led to an increased cytotoxic activity of S23906.²⁷ These observations are important since several ATR or Chk1 inhibitors are currently undergoing preclinical and clinical development.^28–30 Alternatively, the deficiency in either ATR or Chk1 could be considered as potent predictive biomarkers to DNA targeting compounds acting like S23906. Interestingly, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) has recently been reported to bind to S23906-alkylated single stranded, double-stranded and telomeric sequences in a drug-dependent and DNA sequence/structure-dependent manner and to play a role in the cellular response to S23906.³¹ These results demonstrate the importance to precisely characterize the molecular mechanism of action of anticancer drugs to identify initially unexpected biomarkers and/or to uncover original biological processes.

We here report that cells deficient in proteins required for homologous recombination repair are unable to process the S23906-induced DSBs and showed increased sensitivity to the drug, which was particularly pronounced for cells lacking the tumor suppressor protein BRCA2. The marked sensitivity of BRCA2-deficient cells was linked to both the absence of homologous recombination repair and aberrant S-phase progression in the presence of S23906 lesions. Importantly, the activated Ser317-phosphorylated form of Chk1 is equally induced by S23906 in both wt and BRCA2-deficient cells and its pharmacological inhibition by the checkpoint kinase inhibitors 7-hydroxystaurosporine (UCN-01) or AZD7762 has only little influence on S-phase progression and viability in S23906-treated BRCA2-deficient cells but increases the cytotoxic activity of S23906 almost 8-fold in the wt cells. These findings suggest that tumors with deficiencies of proteins involved in recombination repair such as BRCA2 may be particularly sensitive to S23906 and provide a rationale for patient selection and drug combinations in clinical trials.

Results

Cytotoxicity of S23906 toward cells deficient in recombination repair

DSBs can be repaired by 2 major pathways, homologous recombination (HR) that is active during the S and G2 phase of the cell cycle, and non-homologous end joining (NHEJ), that is preferentially active during G0/G1.³² To establish the involvement of the 2 pathways, we determined the cytotoxic effects of S23906 toward well-characterized cell lines deficient for homologous recombination repair (BRCA2, RAD51C, XRCC2, XRCC3) or NHEJ (Ku80, DNA-PK). The results showed increased sensitivity of cells deficient in homologous recombination repair, which was particularly pronounced for BRCA2-deficient cells that were almost 20-fold more sensitive to S23906 (Fig. 1A). In comparison, cells deficient for RAD51C and its 2 paralogues, XRCC2 and XRCC3, were 4–5 fold more sensitive to S23906, compared to the corresponding repair-proficient parental cells (Fig. 1 B-D). In contrast, cells deficient for Ku80 or DNA-PK showed at the most 2-fold increased sensitivity to S23906 (Fig. 2). These results indicate that homologous recombination repair has an important influence on the sensitivity to S23906, which is particularly striking for BRCA2-deficient cells.

Figure 1.

Cytotoxic activity of S23906 toward recombination-deficient cells. Cells proficient or deficient for homologous recombination repair (BRCA2, RAD51C, XRCC2 and XRCC3,) were exposed to S23906 for 4 generation times and the viability determined by the MTT assay. (A), wt V79 (•) and BRCA2-deficient V-C8 cells (∘). (B), wt V79B (•) and RAD51C-deficient CIV4B cells (∘). (C), wt V79 (•) and XRCC2-deficient Irs1 cells (∘). (D), wt AA8 (•) and XRCC3-deficient Irs1 SF cells (∘). All values are averages of at least 3 independent experiments, each done in duplicate. Standard deviations are indicated by error bars and are indicated when they exceed symbol size.

Figure 2.

Cytotoxic activity of S23906 toward non-homologous end-joining -deficient cells. Cells proficient or deficient for non-homologous end-joining (Ku80 and DNA-PK) were exposed to S23906 for 4 generation times and the viability determined by the MTT assay. (A), wt V79B (•) and Ku80-deficient XRV15B cells (∘). (B), DNA-PK-proficient FUS1 (•) and DNA-PK deficient FUS9 cells (∘). All values are averages of at least 3 independent experiments each done in duplicate. Standard deviations are indicated by error bars when they exceed symbol size.

Formation and removal of DSBs in BRCA2-proficient and -deficient cells

Next, wt and BRCA2-deficient cells were exposed to S23906 (5 μM) for 1 hour followed by post-incubation in drug-free media. To directly identify the S23906-induced DNA strand breaks, cells were subjected to single cell electrophoresis under neutral conditions (the neutral comet assay), that almost uniquely detects DSBs.^33,34 The results showed that 1 hour exposure to S23906 was accompanied by formation of comparable levels of strand breaks in the 2 cell lines (Fig. 3A) with the average levels of DNA in the comet being 14% for the parental cells compared to 15% for the BRCA2-deficient cells (data not shown). In contrast, striking differences were revealed during the post-incubation period (Fig. 3A). The parental cells were able to repair the DSBs with the average levels of DNA in the comet reduced from 14% to 11% and 6%, respectively, after 0, 1 and 6 hours post-incubation in drug-free media. In clear contrast, no repair was observed for the BRCA2-deficient cells where the average levels of DNA in the comet remained around 15% even after 6 hours post-incubation (Fig. 3A). Accordingly, the median values that can be retained to characterize each distribution³⁵ show a clear trend toward a decrease of lesions in BRCA2-proficient cells while no such trend can be observed for BRCA2-deficient cells (Fig. 3A).

Figure 3.

Influence of S23906 on wt and BRCA2 mutant cells. (A), BRCA2-proficient V79 (∘) or BRCA2-deficient V-C8 cells (•) were exposed to 5 μM S23906 for 1 hour followed by post-incubation in drug-free media for the indicated times. The results indicate the levels of DNA damage in individual cells from a typical experiment. Bars represent the median values. (B), BRCA2-proficient and deficient cells were exposed to the indicated concentrations of S23906 for 1 hour followed by post-incubation in drug-free media for 6 or 24 hours. The proportion of γ-H2AX-positive cells was determined by flow cytometry analysis after immunostaining. (C), Karyotype analysis. Left, typical metaphase in untreated V79 cells. Right, V79 cells treated with S23906 for 1 hour followed by post-incubation in drug-free media for 24 hours. The arrows indicate chromatid gaps (G), breaks (B) or radials (R). (D), Typical apoptotic features of S23906-exposed V79 cells.

The long-term evolution of the S23906-induced DSBs was further evaluated by flow cytometry of BRCA2-proficient and -deficient cells exposed to S23906 for 1 hour followed by post-incubation in drug-free media for up to 24 hours. The results revealed that both parental (Fig. 3B left) and BRCA2-deficient (Fig. 3B right) cells showed extensive dose-dependent γ-H2AX formation after 6 hours post-incubation in drug-free media. By 24 hours, the  γ-H2AX signal was reduced in the parental cells for all doses below 1 μM. This diminution is most likely a reflection of ongoing DNA repair as illustrated in Fig. 3A. In clear contrast, no decrease in the fraction of γ-H2AX positive cells was observed for the BRCA2-deficient cells, whatever the dose, suggesting an absence of DSB repair even after prolonged post-incubation in drug-free media.

Induction of chromosome damage in BRCA2-proficient and -deficient cells

In addition to the S phase, DSBs are particularly dangerous to cells during mitosis, where the strand breaks can result in formation of chromosome breaks and gaps. To determine the influence of BRCA2 status on the chromosome integrity of S23906-treated cells, BRCA2-proficient and -deficient cells were exposed to different doses of S23906 for 1 hour followed by post-incubation in drug-free media for 24 hours and karyotype analysis (Table 1). The results show that S23906-exposure was accompanied by a dose-dependent increase in the fraction of cells with chromosome abnormalities, the most common aberrations being chromatid breaks, gaps and fragments (Fig. 3C and Table 1). At 100 nM S23906, 34% metaphases of BRCA2-deficient cells showed chromosome aberrations, which increased in a dose-dependent manner to reach 94% at 1 μM (Table 1). In comparison, exposure of repair-proficient parental cells to 1 μM S23906 was associated with chromosome abnormalities in only 6% of the cells (Table 1). However, at 2 μM S23906, 32% of metaphases showed chromosome aberrations. Therefore, although only BRCA2-deficient cells showed chromosomal aberrations at low S23906 concentrations, BRCA2 proficient-and deficient cells showed comparable levels of chromosome aberrations at isotoxic doses.

Table 1.

Analysis of chromosome aberrations in repair-proficient V79 parental cells and BRCA2-deficient V-C8 cells after 1 hour exposure to S23906 followed by 24 hours post-incubation in drug-free media. One hundred metaphases per treatment condition were evaluated

	Wt			BRCA2−
S23906 (nM)	Undamaged Chromosomes	Damaged Chromosomes	End-to-end association or radials	Undamaged Chromosomes	Damaged Chromosomes	End-to-end association or radials
0	100	0	0	98	2	0
50	—	—	—	84	16	1
100	—	—	—	66	34	8
200	—	—	—	36	64	9
500	—	—	—	16	84	10
1000	94	6	0	6	94	10
2000	68	32	3	—	—	—
5000	8	92	27	—	—	—

Induction of apoptotic cell death

DSBs are known to be potent inducers of apoptotic cell death. In agreement, S23906-exposure of both parental and BRCA2-deficient cells was accompanied by the appearance of cells with typical apoptotic nuclear morphology (Fig. 3D) starting 24 hours after drug-exposure.

Taken together, these results indicate that the cytotoxic effects of S23906 are closely associated with induction of DSBs resulting in chromosomal abnormalities and apoptotic cell death. They further demonstrate that the pronounced sensitivity of BRCA2-deficient cells is linked to a prolonged presence of unrepaired DSBs.

Cell cycle progression in BRCA2-proficient and -deficient cells

An important difference between BRCA2 and other factors needed for homologous recombination repair is that BRCA2 is also implicated in the regulation of cell cycle progression during S-phase. In particular, BRCA2-deficient rodent cells as well as primary fibroblasts derived from individuals with mutations in Fanconi anemia complementation group FANCD1/BRCA2 show radioresistant DNA synthesis following exposure to ionizing irradiation.^36,37

To determine if BRCA2 also plays a role in the cell cycle response to S23906, parental and BRCA2-deficient cells were exposed to different concentrations of S23906 for 1 hour followed by post-incubation in drug-free media for 6 hours and flow cytometry analysis. Exposure of parental cells was accompanied by an enrichment of the S phase fraction which increased from 24% in untreated controls to 32% in cells treated with 100 nM and 42% in cells treated with 2,5 μM S23906, that is close to the IC₅₀ concentration (Fig. 4A upper panels). In comparison, S23906-exposure had modest effect on the cell cycle distribution of BRCA2-deficient cells (Fig. 4A lower panels). By 6 hours the S-phase fraction had increased from 25% in untreated control cells to 28% in cells treated with 100 nM S23906 and 30% in cells treated with 2,5 μM S23906. To establish that this observation was not limited to CHO cell lines, BRCA2-deficient human cells were also exposed to different concentrations of S23906 for 1 hour followed by post-incubation in drug-free media and flow cytometry analysis (Fig. S1A). Again, exposure of BRCA2-proficient cells to S23906 was accompanied by a striking enrichment of the S phase fraction (Fig. S1A upper panels) while the drug had no effect on the cell cycle distribution of BRCA2-deficient cells (Fig. S1A lower panels). These results show that the loss of BRCA2 function is accompanied by a defective S-phase response following S23906 exposure, in agreement with the findings reported for ionizing radiation.

Figure 4.

Influence of S23906 on the cell cycle progression in wt and BRCA2 mutant cells. (A), BRCA2-proficient V79 or -deficient V-C8 cells were exposed to the indicated concentrations of S23906 for 1 hour followed by 6 hours post-incubation in drug-free media and cell cycle analysis. (B), BRCA2-proficient V79 or -deficient V-C8 cells were exposed to 0, 250, 500 or 1000 nM S23906 for 1 hour followed by 6 hours post-incubation in drug-free media and immunoblotting analysis.

Chk1 phosphorylation in BRCA2-proficient and -deficient cells

In a recent work, we have shown that S23906 exposure is accompanied by rapid Chk1 activation dependent on ATR (Ataxia telangiectasia- and RAD3-related). Importantly, loss of ATR function or pharmacological inhibition of Chk1 was accompanied by abrogation of the S-phase arrest.²⁷ To determine whether Chk1 is also activated in BRCA2-deficient cells, parental and BRCA2-deficient cells were exposed to different concentrations of S23906 for 1 hour followed by immunoblot analysis (Fig. 4B). The results unambiguously show that Chk1 is equally activated in both cell lines as indicated by the formation of phosphorylated Chk1 on the Ser-317 residue (Fig. 4B). Again, this observation was not restricted to CHO cell lines since similar results were obtained for human wt or BRCA2-deficient cells (Fig. S1B).

Together, our data demonstrate that the initial response to S23906 is not influenced by BRCA2 deficiency in contrast to its late effect on cell cycle progression is. This suggests that BRCA2 acts downstream to ATR/Chk1 to promote S-phase slow-down following S23906 exposure.

Cytotoxicity of S23906 combined with checkpoint kinases inhibitors toward BRCA2-proficient and -deficient cells

We have previously shown that the pharmacological inhibition of the checkpoint kinases potentiate the cytotoxic activity of the S23906 by abrogating the S-phase arrest.²⁷ To determine whether checkpoint kinase inhibition could also potentiate S23906 cytotoxic activity in BRCA2-deficient cells, parental and BRCA2-deficient cells were exposed to either S23906 alone or combined with UCN-01 or AZD7762 (2 inhibitors of the Chk1 and Chk2 checkpoint kinases) used at non-toxic concentrations (data not shown). Interestingly, while both UCN-01 and AZD7762 did individually potentiate the cytotoxic activity of S23906 in parental cells (Fig. 5A left), neither potentiated the cytotoxic activity of S23906 in BRCA2-deficient cells (Fig. 5A right). In agreement, although UCN-01 alone (100 nM for 7 hours) had little influence on the cell cycle distribution, it almost completely abrogated the accumulation of S-phase cells following S23906 exposure in BRCA2-proficient (Fig. 5B upper panels) but not in BRCA2-deficient cells (Fig. 5B lower panels). Interestingly, the absence of effect of checkpoint kinase inhibitors on the S23906 cytotoxic activity seems not to be generalizable to all HR-deficient cells since UCN-01 still improves S23906 efficacy in Rad51-deficient cells (Fig. S2).

Figure 5.

Checkpoint kinase inhibitors abrogates S23906-induced cell cycle arrest and increase S23906 cytotoxic activity in wt BRCA2 cells but have no effect on BRCA2 mutant cells. (A), Influence of UCN-01 and AZD7762 on the cytotoxic effects of S23906. BRCA2-proficient V79 (left) or BRCA2-deficient V-C8 (right) cells were exposed to the indicated concentrations of S23906 for 1 hour in the absence (∘) or presence of either (∘) 100 nM UCN-01 or (•) 100 nM AZD7762 (1 hour co-exposure, 6 hours post-exposure) followed by further incubation in drug-free media for 4 generation times and determination of cellular viability by the MTT assay. All values are averages of at least 2 independent experiments each done in duplicate. Standard deviations are indicated by error bars and are indicated when they exceed symbol size. (B), BRCA2-proficient V79 or -deficient V-C8 cells were cultured in drug-free media (untreated), in the presence of UCN-01 (100 nM, 7 hours), in the presence of S23906 (500 nM for 1 hour followed by 6 hours post-incubation in drug-free media) or in the presence of both S23906 and UCN-01 (1 hour co-incubation followed by 6 hours post-incubation with UCN-01) and flow cytometry analysis.

Together, these results suggest that the pronounced difference in the sensitivity of parental and BRCA2-deficient cells toward S23906 is linked to the attenuation of both repair and S phase control.

Discussion

In this work, we establish a mechanistic link between the initial S23906-adducts and the subsequent downstream events leading to cell death, and identify HR as an important modulator of S23906 action.

We have recently reported that S23906 exposure induces the formation of DNA adducts that are primarily processed by the NER machinery.¹⁸ If left unrepaired, the adducts interfere with the progression of the replication fork in S phase leading to the formation of replicative stress foci.²⁷ This leads to the early activation of the ATR/Chk1 pathway that controls the stabilization of stalled replication forks, S-phase progression and the resumption of DNA synthesis.³⁸ Ultimately, the initial lesions are transformed into DSBs possibly due to the collapse of stalled replication forks.^18,26,27,39

DSBs can be repaired by 2 major pathways, HR and NHEJ. To establish the involvement of the 2 pathways, we here characterized the cytotoxic effects of S23906 toward cell lines deficient for HR (RAD51C, XRCC2, XRCC3, BRCA2) or NHEJ (Ku 80, DNA-PK). The results showed an important role for HR that was particularly pronounced for BRCA2-deficient cells. In contrast, NHEJ seemed to play a lesser role.

To further characterize the role of HR, BRCA2-proficient and -deficient cells were exposed to S23906 for 1 hour followed by post-incubation in drug-free media for up to 6 hours. Comet analysis under neutral conditions revealed no differences in the initial levels of DSBs between BRCA2-proficient and -deficient cells. In contrast, the DSBs were effectively repaired in the wt cells during the post-incubation period while no repair was observed for the BRCA2-deficient cells. Similar findings were observed by flow cytometry analysis of γ-H2AX positive cells following 6 and 24 hours post-incubation in drug-free media. A dose-dependent increase in the fraction of γ-H2AX positive cells was observed at all doses of S23906 after 6 hours post-incubation for both parental and BRCA2-deficient cells. However, after 24 hours post-incubation, wt cells showed an important diminution in the fraction of γ-H2AX positive cells at all doses below 1 μM while no decrease was observed for the BRCA2-deficient cells.

The presence of unrepaired DSBs may lead to apoptosis^40,41 and to chromosomal abnormalities such as chromosome breaks and chromatid fusions which would lead to micronuclei formation and anaphase bridging.⁴² Accordingly, S23906-treatment was followed by the appearance of typical apoptotic nuclear morphology as well as mitotic chromosomes with chromatid aberrations in terms of gaps, breaks and radials. Radials were first described following exposure to DNA interstrand cross-linking agents such as mitomycin C⁴³ and it is currently debated whether these structures represent cross-linked DNA or abnormal recombination intermediates. The formation of radials after exposure to monofunctional agents such as trabectedin²¹ and S23906 (this report) provides experimental evidence in favor of the second hypothesis, although the precise mechanism remains uncertain.⁴⁴

Whereas all HR-deficient cell lines show increased sensitivity to S23906, this was particularly pronounced for the BRCA2-deficient cells, which were almost 20-fold more sensitive to S23906 compared to the 4–5 fold increased sensitivity for the other HR-deficient cells (RAD51C, XRCC2, XRCC3). Interestingly, beside its function in repair, it has been reported that BRCA2 also plays a role in stabilizing replication forks following treatment with hydroxyurea (HU), an agent that depletes the nucleotide pool, causing genome-wide DNA polymerase stalling.^45–47 In the absence of BRCA2, newly replicated strands are degraded at stalled replication forks in a process dependent on MRE11 nuclease activities.⁴⁶ This is followed by replication fork collapse and DSBs formation.^45,48 However, if the pronounced sensitivity of the BRCA2-deficient cells toward S23906 was due to the accumulation of unprotected replication forks in BRCA2-deficient cells, one could expect an increased accumulation of DSBs in BRCA2-deficient cells compared to wt cells following S23906 exposure. This is not what our results show. Moreover, after continuous exposure to HU, BRCA2-deficient cells showed only a modest sensitivity toward HU relative to cells expressing wild-type BRCA2, suggesting that fork degradation has little effect on cell survival.⁴⁶ In addition, it has been reported that BRCA2 deficient-cells efficiently maintain S phase arrest during HU treatment making it unlikely that the replicative protective function of BRCA2 is a key determinant of S23906 sensitivity.⁴⁵ Interestingly, it has been recently shown that the recruitment of the PALB2–BRCA2 complex at the stalled replication forks was mediated by the phosphorylated form of RPA2.⁴⁷ This observation reinforces the likely non-decisive immediate protective role of BRCA2 at stalled replication fork following S23906 exposure since RPA2 is only lately phosphorylated in S23906 treated cells.²⁷

A particularity of both BRCA2-deficient rodent cells and primary fibroblasts derived from patients with mutations in Fanconi anemia complementation group FANCD1/BRCA2 is that the cells are not only deficient in HR, but also show an impaired S-phase checkpoint as indicated by radio-resistant DNA synthesis following exposure to ionizing irradiation.^36,37 In extension of these findings, we now show that S23906 triggers a prominent S-phase response in parental cells while the cell cycle progression is practically unchanged in hamster and human BRCA2-deficient cells, whatever the dose. Triggering of S-phase checkpoints in response to DNA damage is a complex process.³⁸ Chk1 has been implicated as the major mediator of S phase arrest following DNA damage and has been involved in the cellular response to S23906.^27,49,50 To establish the influence of S-phase progression in response to S23906, parental and BRCA2-deficient cells were exposed to UCN-01, a checkpoint abrogator with selective activity toward Chk1 in the low nM dose range.⁵¹ Interestingly, UCN-01 almost completely inhibited the S23906-induced S-phase slowdown in parental cells but had only a marginal effect on the BRCA2-deficient cells. This was accompanied by a dramatic 8-fold increased sensitivity of parental cells to S23906 while no significant effect was observed for BRCA2-deficient cells. FA (Fanconi anemia) proteins have been involved in S-phase arrest following inter-strand crosslinks (ICLs) formation.^52–54 In particular, it has been shown that ICLs induce 2 independent pathways (ATR/Chk1 and ATR/NBS1–FANCD2 that requires the core FANC complex) who cooperate to fully activate the S-phase checkpoint in response to ICLs.⁵⁴ Interestingly, in response to ICLs, Chk1 inhibition led only to a partial deficiency in the S phase checkpoint, not to a complete absence as observed following S23906 exposure. This, conjugated with the absence of effect of Chk1 inhibition on BRCA2-deficient cells sensitivity toward S23906, suggests that Chk1 and BRCA2 do not act independently like for ICLs but rather act through a common pathway in response to S23906 exposure. Again, these results underline the need of a precise characterization of the cellular context to rationally develop new anticancer compounds and identify a molecular signature (“BRCA2 low”) where combining S23906 with checkpoint kinase inhibitors is likely not to prove useful.

Together, our findings suggest that the pronounced sensitivity of BRCA2-deficient cells to S23906 is due to aberrant S-phase progression in the presence of S23906 lesions as well as to the absence of homologous recombination repair. In this model, the S23906-induced DNA-adducts would interfere with the progression of the replication fork leading to the downstream activation of the ATR/Chk1 pathway.²⁷ Chk1 would then participate in the activation of HR likely by phosphorylating both BRCA2 and Rad51.^55,56 However, our results suggest that BRCA2 also functions downstream of Chk1 in the control of the cell cycle progression (Fig. 5B lower panels). This process seems independent of the other HR regulators since UCN-01 still improves S23906 efficacy in Rad51-deficient cells although to a lesser degree than in WT cells (Fig. 5B upper panels and Fig. S2). Interestingly, a similar RAD51-independent role of BRCA2 has been reported for maintenance of the G2 checkpoint induced in response to ionizing irradiation.⁵⁷ Finally, even if the replicative protective function of BRCA2 may not to be a key determinant of S23906 sensitivity, its possible role at the replication fork once RPA2 becomes phosphorylated could explain why BRCA2-deficient cells show an extreme sensitivity toward S23906 which cannot be fully recapitulated by Chk1 inhibition alone. Interestingly, RPA2 is phosphorylated independent of Chk1 thereby making this process insensitive to Chk1 inhibition.⁵⁸ Our findings may have important clinical implications, since defects in DNA repair genes are believed to be common in human cancers. In particular, mutations in BRCA1 or BRCA2 in women are associated with cancers of the breast, ovary, fallopian tube, and peritoneum, while male carriers of BRCA1 and BRCA2 mutations are at increased risk of cancers of the breast, prostate, and pancreas.⁵⁹

In summary, our findings provide a model for the mechanism of S23906 action from the initial DNA adducts to the induction of cell death, identify a molecular signature (NER low, HR and, particularly, BRCA2 low) that predicts tumor cell response to S23906 and its related derivatives, and provide a rationale for patient selection in future clinical trials.

Material and Methods

Chemicals

The acronycine derivative S23906 was obtained from Institut de Recherches Servier while nocodazole was purchased from Sigma (http://www.sigmaaldrich.com/catalog/product/sigma/m1404). AZD7762 was purchased from Axon Medchem (http://www.axonmedchem.com/product/1399) and UCN-01 (7-hydroxystaurosporine) was kindly provided by Edward A. Sausville (University of Maryland, MD).

Antibodies

The phospho-Ser317Chk1-directed antibody (http://www.cellsignal.com/products/primary-antibodies/2344) as well as the Chk1-directed antibody (clone 2G1D5, http://www.cellsignal.com/products/primary-antibodies/2360) were purchased from Cell Signaling Technology. The β-actin-directed antibody (clone AC-15, http://www.sigmaaldrich.com/catalog/product/sigma/a5441) was from Sigma–Aldrich while the γ-H2AX-directed antibody (http://www.emdmillipore.com/US/en/product/,MM_NF-05-636) was purchased from Millipore. HRP (horseradish peroxidase) and fluorescent dye-conjugated antibodies were all obtained from Jackson ImmunoResearch.

Cells

Recombination-deficient hamster cells lines and their corresponding repair-proficient parental cells (AA8 and V79B) were generously provided by Malgorzata Z. Zdzienicka (Leiden, the Netherlands). The cells include irs1, defective in XRCC2;^60,61 irs1SF, defective in XRCC3,^62,63 CL⁻V4B, defective in RAD51C,⁶⁴ V-C8, defective in BRCA2/XRCC11/FANCD1³⁶ and XR-V 15B, defective in Ku80/XRCC5.^65,66 DNA-PK deficient Fus9 human M059J glioblastoma cells and DNA-PK proficient Fus1 cells⁶⁷ were kindly provided by Bernard Salles (Toulouse, France). All cell lines were regularly tested for Mycoplasma contamination by PCR analysis.

Single cell electrophoresis

Cells for comet analysis were exposed to the indicated drug-concentrations at 37°C in the dark and analyzed immediately according to previously published procedures.^21,33,68,69 Cells were stained with ethidium bromide (2 μg/ml) and the slides were examined at 400x magnification using a fluorescent microscope (Nikon TS 100) without prior knowledge of the treatment. Image analysis was performed by using the Komet 5.5 software (Kinetic Imaging Ltd, Nottingham, United Kingdom). At least 100 cells were analyzed per sample. Results are expressed as % of total nuclear DNA present in the comet tail and are depicted for all cells analyzed in a representative experiment. Alternatively, the values shown represent the average levels of DNA damage from at least 2 independent experiments.

Growth inhibition and viability assays

The cytotoxic activity of S23906 was measured using the MTT colorimetric assay as previously described.¹² Briefly, cells proficient or deficient for specific repair genes were exposed to S23906 for 4 generation times and the viability determined. It has to be noted that the cell lines used in this study did not all proliferate with a similar doubling time. AA8, V79, CL⁻V4B, VC-8 and XR-V15B doubled every 14–16 hours while Irs1 and irs1SF doubled every 17 and 20 hours, respectively. DNA-PK deficient Fus9 human M059J glioblastoma cells doubled every 40 hours while DNA-PK proficient Fus1 cells doubled in approximately 24 hours. AA8, V79, CL⁻V4B, VC-8, XR-V15B and Irs1 were therefore exposed to S23906 for 66 hours while irs1SF were exposed to S23906 for about 80 hours. Fus1 and Fus9 human M059J glioblastoma cells were exposed to S23906 for 4 and 7 days, respectively. All values are averages of at least 3 independent experiments each done in duplicate.

Cell cycle analysis and Histone H2AX phosphorylation

Cell cycle analysis was carried out as described previously.^6,70 The phosphorylation of histone H2AX was determined by flow cytometry analysis after immunolabeling with an anti-phospho-histone-γ-H2A.X (ser139) murine monoclonal antibody as described.^21,26

Immunoblotting

Cells were incubated with different concentrations of S23906 at 37°C for 1 hour, washed in PBS, counted and lysed for 30 min at 4°C in SDS/PAGE loading buffer. Proteins were resolved on linear-gradient SDS/PAGE (5–15%) and blotted on nitrocellulose membranes (Bio-Rad). Membranes were saturated by TBST-milk [50 mM Tris/HCl (pH 8.0), 150 mM NaCl, 0.5% Tween 20 and 5% dehydrated skimmed milk] and the antigens were revealed by immunolabelling. Antigens were detected using an enhanced chemiluminescence kit (Amershan Biosciences).

Karyotype analysis

V79 parental cells and V-C8 mutant cells (BRCA2⁻) were exposed for 1 hour to the indicated doses of S23906. Cells were washed with PBS and post-incubated in drug-free medium for 24 hours, and chromosome spreads were prepared as described.^21,33 One hundred metaphases per treatment condition were evaluated.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr. Malgorzata Zdzienicka for generously providing us with the recombination-deficient cells.

Funding

Daniele Grazziotin Soares was supported by a fellowship from Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior (CAPES), Brasil. Hana Bouzid is supported by a fellowship from La Ligue Contre le Cancer, France.