**3. Repair processes for DNA destabilizing lesions**

DNA adducts are critical lesions for cell proliferation and survival. Single or multiple DNA repair machineries could be implicated in the removal of these damages, as for example

DNA Helix Destabilization by Alkylating Agents: From Covalent Bonding to DNA Repair 105

(Newlands et al., 1997) and cell sensitivity to TMZ treatment depends on multiple DNA repair mechanisms (**2**). The major one is the recognition of methyl lesions from O6 position of guanines by the O6-methylguanine DNA methyltransferase (MGMT) protein which directly converts the methylated DNA to its normal, undamaged state (**3**). MGMT enzymatic activity is crucial for TMZ resistance in vivo suggesting that MGMT expression may predict the response of patients to TMZ treatment (Everhard et al., 2006; McCormack et al., 2009). However, other repair mechanisms are also implicated since some cell lines with low MGMT expression still evidence significant resistance to TMZ (Fukushima et al., 2009). When O6-methyguanine is not repaired by MGMT, it may lead to an O6 methylguanine:thymine mismatch during DNA replication. The following DNA replication cycle can then pair thymine with adenine in place of the original guanine, thus leading to transition mutations (**4**). However, the cytotoxic property of TMZ is mostly linked to MMR pathway through O6-methylguanine:thymine mismatch recognition and repair by this system **(5**). MMR is not involved in TMZ chemo-resistance but in TMZ cytotoxicity, associated with cell cycle blockade at G2 checkpoint (Caporali et al., 2004), activation of p53 and ATM, leading to cell death (**6**). The MRN (Mre11/Rad50/Nbs1) complex was evidenced as the earliest sensor of TMZ-induced damage (Mirzoeva et al., 2006). It undergoes a series of conformational changes that activates the protein sensor ATM (ataxia telangiectasia mutated) which, subsequently, activates Chk1 and Chk2 to block cell cycle. TMZ induces p53-mediated apoptosis in MMR-proficient but not in MMR-deficient cells (D'Atri et al., 1998). Thus, deficient MMR is another mechanism for resistance to TMZ (Cahill et al., 2007). Besides MGMT and MMR, BER is also implicated in TMZ lesion repair. More than 80% of N7-methylated purines are recognized and excised by the BER enzyme N-methylpurine DNA glycosylase (MPG) (Trivedi et al., 2008; J. Zhang et al., 2010) (**7**). As a consequence, disruption of BER system sensitizes MMR-deficient and proficient cells (Liu et al., 1999). The major MPG-dependent repair occurs via short-patch BER, a mechanism whereby only the damaged nucleotide is excised. So, BER pathway is another contributor of cell resistance to TMZ and its efficacy depends on specific BER gene expression and activity (Fishel et al., 2008). DNApol β or MPG-deficient cells are more sensitive than wild-type cells to TMZinduced cell death, whereas MPG over-expression increases TMZ-induced cytotoxicity (Tang et al., 2011; Trivedi et al., 2008). Similarly, inhibition of poly(ADP-ribose) polymerase-

1 partially restored sensitivity to TMZ (J. Zhang et al., 2010).

well as the implication of Fanconi anemia FANC-D1 (Kondo et al., 2011).

**3.2 DNA repair process and implication in ET-743 expressing cytotoxicity** 

ET-743 is a tetrahydroisoquinoline alkaloid isolated from the tunicate *Ecteinascidia turbinata* which is approved as an orphan drug against advanced soft tissue sarcoma and, in

Both methylated DNA lesions can lead to SSBs in a DNA repair-dependent manner (BER, MMR). If unrepaired before replication, SSBs convert in DSBs, a more mutagenic and lethal lesion (Newlands et al., 1997). However, DSBs could be processed by the conservative HR pathway to give back undamaged double stand DNA or by NHEJ repair machinery potentially resulting in chromosomal rearrangements between chromatide or deleterious genomic rearrangements as other toxic lesions (**8**). Other inter-crossings between repair pathways are not presented in this scheme: a role of some MMR proteins in the NHEJ pathway to repair DSB during G1 phase of the cell cycle or in HR pathway through the regulation of the early G2 checkpoint and inhibition of DSB repair (Y. Zhang et al., 2009) as

BER, GG-NER (global genome) or TC-NER (transcription-coupled), MMR, HR or NHEJ. Only few data are published about the consequences of non-covalent DNA destabilizing agents on protein/DNA binding from the repair machineries. These data on BisA function reported that insertion of BisA could flip the mispaired thymine to an extrahelical base subsequently inducing a sterical blockage of DNA glycosylases binding (David, 2003). The present section will therefore focus on alkylating compounds. As examples, we will shortly present the repair processes for the well-studied temolozomide-induced lesions in the major groove and for the DNA stabilizing drug ET-743, as an original minor groove alkylating agents that "poison" the NER machinery to exert its anti-tumor properties, before presenting the current knowledge on DNA repair of DNA destabilizing lesions.

#### **3.1 Repair of temolozomide-induced DNA lesions**

Temolozomide (TMZ, Temodar®, Figure 5) is a monofunctional alkylating agent chemically related to dacarbazine. It is active in vitro and in vivo against a wide variety of tumor type and particularly efficient in malignant glioma (Newlands et al., 1997). Contrasting with dacarbazine, TMZ does not require to be activated by enzymatic oxidation, but spontaneously hydrolyses to 5-(3-methyltriazen-l-yl)-imidazole-4-carboximide (MITC) at pH above 7. MITC is then broken down to (*i*) the reactive methyldiazonium cation which next loses the methyl group in the presence of DNA or proteins and (*ii*) the inactive 5 aminoimidazole-4-carboxyamide moiety (AIC) (**1**). TMZ treatment leads to different adducts on the double helix DNA: N3-methyladenine, N7-methylguanine and O6-methylguanine

Fig. 5. DNA repair pathways for TMZ-induced damage.

BER, GG-NER (global genome) or TC-NER (transcription-coupled), MMR, HR or NHEJ. Only few data are published about the consequences of non-covalent DNA destabilizing agents on protein/DNA binding from the repair machineries. These data on BisA function reported that insertion of BisA could flip the mispaired thymine to an extrahelical base subsequently inducing a sterical blockage of DNA glycosylases binding (David, 2003). The present section will therefore focus on alkylating compounds. As examples, we will shortly present the repair processes for the well-studied temolozomide-induced lesions in the major groove and for the DNA stabilizing drug ET-743, as an original minor groove alkylating agents that "poison" the NER machinery to exert its anti-tumor properties, before

Temolozomide (TMZ, Temodar®, Figure 5) is a monofunctional alkylating agent chemically related to dacarbazine. It is active in vitro and in vivo against a wide variety of tumor type and particularly efficient in malignant glioma (Newlands et al., 1997). Contrasting with dacarbazine, TMZ does not require to be activated by enzymatic oxidation, but spontaneously hydrolyses to 5-(3-methyltriazen-l-yl)-imidazole-4-carboximide (MITC) at pH above 7. MITC is then broken down to (*i*) the reactive methyldiazonium cation which next loses the methyl group in the presence of DNA or proteins and (*ii*) the inactive 5 aminoimidazole-4-carboxyamide moiety (AIC) (**1**). TMZ treatment leads to different adducts on the double helix DNA: N3-methyladenine, N7-methylguanine and O6-methylguanine

presenting the current knowledge on DNA repair of DNA destabilizing lesions.

**3.1 Repair of temolozomide-induced DNA lesions** 

Fig. 5. DNA repair pathways for TMZ-induced damage.

(Newlands et al., 1997) and cell sensitivity to TMZ treatment depends on multiple DNA repair mechanisms (**2**). The major one is the recognition of methyl lesions from O6 position of guanines by the O6-methylguanine DNA methyltransferase (MGMT) protein which directly converts the methylated DNA to its normal, undamaged state (**3**). MGMT enzymatic activity is crucial for TMZ resistance in vivo suggesting that MGMT expression may predict the response of patients to TMZ treatment (Everhard et al., 2006; McCormack et al., 2009). However, other repair mechanisms are also implicated since some cell lines with low MGMT expression still evidence significant resistance to TMZ (Fukushima et al., 2009).

When O6-methyguanine is not repaired by MGMT, it may lead to an O6 methylguanine:thymine mismatch during DNA replication. The following DNA replication cycle can then pair thymine with adenine in place of the original guanine, thus leading to transition mutations (**4**). However, the cytotoxic property of TMZ is mostly linked to MMR pathway through O6-methylguanine:thymine mismatch recognition and repair by this system **(5**). MMR is not involved in TMZ chemo-resistance but in TMZ cytotoxicity, associated with cell cycle blockade at G2 checkpoint (Caporali et al., 2004), activation of p53 and ATM, leading to cell death (**6**). The MRN (Mre11/Rad50/Nbs1) complex was evidenced as the earliest sensor of TMZ-induced damage (Mirzoeva et al., 2006). It undergoes a series of conformational changes that activates the protein sensor ATM (ataxia telangiectasia mutated) which, subsequently, activates Chk1 and Chk2 to block cell cycle. TMZ induces p53-mediated apoptosis in MMR-proficient but not in MMR-deficient cells (D'Atri et al., 1998). Thus, deficient MMR is another mechanism for resistance to TMZ (Cahill et al., 2007).

Besides MGMT and MMR, BER is also implicated in TMZ lesion repair. More than 80% of N7-methylated purines are recognized and excised by the BER enzyme N-methylpurine DNA glycosylase (MPG) (Trivedi et al., 2008; J. Zhang et al., 2010) (**7**). As a consequence, disruption of BER system sensitizes MMR-deficient and proficient cells (Liu et al., 1999). The major MPG-dependent repair occurs via short-patch BER, a mechanism whereby only the damaged nucleotide is excised. So, BER pathway is another contributor of cell resistance to TMZ and its efficacy depends on specific BER gene expression and activity (Fishel et al., 2008). DNApol β or MPG-deficient cells are more sensitive than wild-type cells to TMZinduced cell death, whereas MPG over-expression increases TMZ-induced cytotoxicity (Tang et al., 2011; Trivedi et al., 2008). Similarly, inhibition of poly(ADP-ribose) polymerase-1 partially restored sensitivity to TMZ (J. Zhang et al., 2010).

Both methylated DNA lesions can lead to SSBs in a DNA repair-dependent manner (BER, MMR). If unrepaired before replication, SSBs convert in DSBs, a more mutagenic and lethal lesion (Newlands et al., 1997). However, DSBs could be processed by the conservative HR pathway to give back undamaged double stand DNA or by NHEJ repair machinery potentially resulting in chromosomal rearrangements between chromatide or deleterious genomic rearrangements as other toxic lesions (**8**). Other inter-crossings between repair pathways are not presented in this scheme: a role of some MMR proteins in the NHEJ pathway to repair DSB during G1 phase of the cell cycle or in HR pathway through the regulation of the early G2 checkpoint and inhibition of DSB repair (Y. Zhang et al., 2009) as well as the implication of Fanconi anemia FANC-D1 (Kondo et al., 2011).

#### **3.2 DNA repair process and implication in ET-743 expressing cytotoxicity**

ET-743 is a tetrahydroisoquinoline alkaloid isolated from the tunicate *Ecteinascidia turbinata* which is approved as an orphan drug against advanced soft tissue sarcoma and, in

DNA Helix Destabilization by Alkylating Agents: From Covalent Bonding to DNA Repair 107

removal of both 5'-GG, 5'-AG and 5'-GNG cisplatin intra-strand crosslinks, with a preference for the latter site. The induced-kink, being greater for 5'-GNG than 5'-GG or 5'-AG alkylated sites, seems to be of major relevance for NER recognition (**1**, in Figure 7). Particularly, platinum adducts are recognized by the global genome-NER XPC/hHR23B "sensor complex" (Neher et al., 2010) and XPC expression or polymorphism predicts the response to cisplatin treatment in lung cancers (Lai et al., 2011; L.B. Zhu et al., 2010). Lesions induced by cisplatin, oxaliplatin and JM216 are similarly repaired whereas transplatin-induced lesions, which

MMR is also important to remove platinated lesions (**2**). Facilitated by cisplatin-induced kink, MSH2 binding is associated with a 60° angle generated through intercalation of its Phe39 at the lesion site. MSH2/MSH6 complex (Mut-S) recognizes cisplatin crosslinks (Castellano-Castillo et al., 2008; Fourrier et al., 2003) but not transplatin mono-adducts from [Pt(dien)Cl]+. Translesion bypass is also implicated in cisplatin toxicity. Interestingly, oxaliplatin lesions are more bypassed by DNA polymerases than cisplatin, in relation with their difference in DNA bending/destabilization potencies. Mutants FANC-C and –D of

Of major concern, cisplatin adducts are also recognized by HMG proteins (**3**). Similarly to MutS complex recognition, the large induced bend is crucial for this recognition and fits perfectly with the L-shaped structure of HMG DNA binding domain (HMG-box) to reduce the "cost" of DNA bending for HMG-box (Privalov et al., 2009). Insertion of Phe37 between

poorly affect 3D structure of DNA, are poorly repaired by NER.

Fig. 7. DNA repair pathways for platinated DNA.

Fanconi anemia pathway also sensitize cells to Pt-drug (Kachnic et al., 2010).

association with doxorubicine, in refractory cisplatin-sensitive ovarian cancers. This DNA minor groove binder (Pommier et al., 1996) bends DNA toward the major groove (Hurley & Zewail-Foote, 2001). ET-743 (Figure 6) is composed of three subunits: A and B are involved in DNA binding at specific sites (David-Cordonnier et al. 2005; García-Nieto et al., 2000; Pommier et al., 1996) and C protrudes out of the double helix thus facilitating the interaction of ET-743 with nuclear proteins such as transcription factors or DNA repair proteins (**1**). The formation of such protein/ET-743-DNA complex prevents the transcription of different genes (Friedman et al., 2002; Jin et al., 2000) and induces a rapid degradation of transcribing RNA polymerase II in TC-NER proficient, but not deficient, cells (Aune et al., 2008).

By contrast with other DNA damaging agents, NER-deficient cell lines are resistant to ET-743, and restoration of NER functions sensitizes cells to the drug. Indeed, the TC-NER complex is trapped during the repair process of ET-743-DNA damage (Damia et al., 2001; Takebayashi et al., 2001) through the formation of a stable XPG/DNA 'cytotoxic complex' (Herrero et al., 2006)(**2**). In a replication-independent manner, the MRN complex is recruited (**3**) and induces DSBs subsequently recognized by DNA-PK from the HR machinery. DNA-PK then phosphorylates H2AX and activates ATM (Damia et al., 2001) and Chk1 to bypass G2/M and S phases checkpoints and promote cell death (Herrero et al., 2006).

Protein recognition of ET-743-DNA adducts also induces the formation of DSBs through replication fork collapse (Soares et al., 2007; Takebayashi et al., 2001)(**4**), as well known for topoisomerase/drug/DNA poisoning complexes. Such DSBs are repaired by HR (acting mainly in G2-M phases) but not by NHEJ (Soares et al., 2007; Tavecchio et al., 2008)(**5**).

Fig. 6. DNA repair pathways for ET-743-induced DNA damage.

#### **3.3 DNA repair for cisplatin and other transition-metal antitumor agents**

Regarding DNA repair, local destabilization of the double helix, base-flipping, DNA bending and poor base-stacking following cisplatin alkylation are determinant for recognition of DNA lesions by repair proteins (C.G. Yang et al., 2009; W. Yang, 2006). Several repair machineries are implicated in metal-drug-induced DNA adduct recognition, removal and cytotoxicity (Basu & Krishnamurthy, 2010; S. Ahmad, 2010). First, NER is an important actor for the

association with doxorubicine, in refractory cisplatin-sensitive ovarian cancers. This DNA minor groove binder (Pommier et al., 1996) bends DNA toward the major groove (Hurley & Zewail-Foote, 2001). ET-743 (Figure 6) is composed of three subunits: A and B are involved in DNA binding at specific sites (David-Cordonnier et al. 2005; García-Nieto et al., 2000; Pommier et al., 1996) and C protrudes out of the double helix thus facilitating the interaction of ET-743 with nuclear proteins such as transcription factors or DNA repair proteins (**1**). The formation of such protein/ET-743-DNA complex prevents the transcription of different genes (Friedman et al., 2002; Jin et al., 2000) and induces a rapid degradation of transcribing

RNA polymerase II in TC-NER proficient, but not deficient, cells (Aune et al., 2008).

G2/M and S phases checkpoints and promote cell death (Herrero et al., 2006).

Fig. 6. DNA repair pathways for ET-743-induced DNA damage.

**3.3 DNA repair for cisplatin and other transition-metal antitumor agents** 

Regarding DNA repair, local destabilization of the double helix, base-flipping, DNA bending and poor base-stacking following cisplatin alkylation are determinant for recognition of DNA lesions by repair proteins (C.G. Yang et al., 2009; W. Yang, 2006). Several repair machineries are implicated in metal-drug-induced DNA adduct recognition, removal and cytotoxicity (Basu & Krishnamurthy, 2010; S. Ahmad, 2010). First, NER is an important actor for the

By contrast with other DNA damaging agents, NER-deficient cell lines are resistant to ET-743, and restoration of NER functions sensitizes cells to the drug. Indeed, the TC-NER complex is trapped during the repair process of ET-743-DNA damage (Damia et al., 2001; Takebayashi et al., 2001) through the formation of a stable XPG/DNA 'cytotoxic complex' (Herrero et al., 2006)(**2**). In a replication-independent manner, the MRN complex is recruited (**3**) and induces DSBs subsequently recognized by DNA-PK from the HR machinery. DNA-PK then phosphorylates H2AX and activates ATM (Damia et al., 2001) and Chk1 to bypass

Protein recognition of ET-743-DNA adducts also induces the formation of DSBs through replication fork collapse (Soares et al., 2007; Takebayashi et al., 2001)(**4**), as well known for topoisomerase/drug/DNA poisoning complexes. Such DSBs are repaired by HR (acting mainly in G2-M phases) but not by NHEJ (Soares et al., 2007; Tavecchio et al., 2008)(**5**).

removal of both 5'-GG, 5'-AG and 5'-GNG cisplatin intra-strand crosslinks, with a preference for the latter site. The induced-kink, being greater for 5'-GNG than 5'-GG or 5'-AG alkylated sites, seems to be of major relevance for NER recognition (**1**, in Figure 7). Particularly, platinum adducts are recognized by the global genome-NER XPC/hHR23B "sensor complex" (Neher et al., 2010) and XPC expression or polymorphism predicts the response to cisplatin treatment in lung cancers (Lai et al., 2011; L.B. Zhu et al., 2010). Lesions induced by cisplatin, oxaliplatin and JM216 are similarly repaired whereas transplatin-induced lesions, which poorly affect 3D structure of DNA, are poorly repaired by NER.

MMR is also important to remove platinated lesions (**2**). Facilitated by cisplatin-induced kink, MSH2 binding is associated with a 60° angle generated through intercalation of its Phe39 at the lesion site. MSH2/MSH6 complex (Mut-S) recognizes cisplatin crosslinks (Castellano-Castillo et al., 2008; Fourrier et al., 2003) but not transplatin mono-adducts from [Pt(dien)Cl]+. Translesion bypass is also implicated in cisplatin toxicity. Interestingly, oxaliplatin lesions are more bypassed by DNA polymerases than cisplatin, in relation with their difference in DNA bending/destabilization potencies. Mutants FANC-C and –D of Fanconi anemia pathway also sensitize cells to Pt-drug (Kachnic et al., 2010).

Fig. 7. DNA repair pathways for platinated DNA.

Of major concern, cisplatin adducts are also recognized by HMG proteins (**3**). Similarly to MutS complex recognition, the large induced bend is crucial for this recognition and fits perfectly with the L-shaped structure of HMG DNA binding domain (HMG-box) to reduce the "cost" of DNA bending for HMG-box (Privalov et al., 2009). Insertion of Phe37 between

DNA Helix Destabilization by Alkylating Agents: From Covalent Bonding to DNA Repair 109

to that of the other conformers, contributes to its higher mutagenic potential (Mocquet et al., 2007). Lesion recognition by XPC requires DNA bending facilitated by local conformational flexibility (Clement et al., 2010) and destabilization of the base-pairing (Brown et al., 2010). Such recognition is driven by Trp690 and Thp733 amino-acids identified as "aromatic sensors" (Maillard et al., 2007). Upon treatment with BaP, human bronchial epithelial 16HBE cells expressed higher levels of the heat shock protein 70 and the NER proteins XPA and XPG, both three proteins co-localizing in the nucleus, suggesting that Hsp70 is also implicated in the DNA repair response to BPDE-DNA adducts (J. Yang et al., 2009). The highly mutagenic (*+*)-(7R,8S,9S,10R)-7,8-dihydroxy-9,10-epoxy-7,8,9,10-tetrahydrobenzo[*a*]pyrene-DNA lesion leads to different repair processes depending on sequence context, associated with the destabilization potency. Indeed, for an identical BaP-DNA lesion leading to differently orientated bulky lesions, sequence-dependent effect was observed: DNA destabilized at 5'-CG\*GC site is more rapidly excised in cell-free human HeLa extracts than DNA bent at 5'-CGG\*C site (Rodríguez et al., 2007). As the DNA helix is readily opened upon alkylation, recognition of the lesion by repair protein (including induction of base flipping) is less energetic and, thus, is quicker for DNA already destabilized at 5'-CG\*GC site than for duplex DNA bent at 5'-CGG\*C site, clearly evidencing the importance of DNA sequence/global structure context for an efficient repair of BPDE-DNA adducts (Yuqin et al., 2009). Moreover, interesting data arise from comparison of the 3D conformation and the NER excision efficiencies for dA adducts formed using the bay region BPDE and the fjord region benzo[*c*]phenanthrene diol epoxide (B[*c*]PhDE) (M. Wu et al., 2002). The bay region of B[*a*]P is more extended, planar and rigid than the B[c]Ph fjord region, being twisted and curved. Consequently, B[*a*]P-dA adducts are associated with greater backbone distortion, unwinding, intercalation potency and

Fig. 8. DNA repair pathways for BPDE-induced DNA damage.

the two platinated guanines in 5'GG dinucleotide stabilizes the binding but is regulated in a redox manner. Indeed, the formation of a disulfure bond between the thiol groups of Cys22 and Cys44 on helix II and III, respectively, of HMG-box infers with the correct planar insertion of Phe37 between the two guanines at crosslink site (Park & Lippard, 2011). Binding of HMG-B1 (and HMG-B2) stabilizes the cisplatin-induced bent and supercoiling of the DNA helix, increases the sensitivity of the cells to cisplatin and shields the platinated adducts from repair by the human DNA excision machinery (J.C. Huang et al., 1994). As a consequence of the degree of kink of the DNA, HMG proteins poorly bind to oxaliplatin adducts which induce relatively small DNA-bending and DNA destabilization (Figure 3), and so poorly protects them from DNA repair (Kasparkova et al., 2008b). This difference correlates with the lower level of DNA lesions in oxaliplatin- versus cisplatin-treated cells. If HMG-B1 and –B2 binding participates in platinated-agent-induced cytotoxicity (Sharma et al., 2009), bent platinated-DNA is also a good substrate for transcription factors from HMGbox family such as SRY, LEF-1 and UBF-1, resulting in the transcriptional changes observed in treated cells (Chvalova et al., 2008; Treiber et al., 1994; Trimmer et al., 1998). For the repair of other platinum derivative-induced DNA damages, JM108 evidenced higher level of protein/DNA cross-links such as DNA-PtII-NF-κB cross-linked complexes (**4**). Those lesions are less efficiently removed from DNA by the cell repair system (Kostrhunova et al., 2010). Other studies described the binding of PARP-1 protein to cisplatin adduct at 5'-GG and 5'- GNG intra-strand crosslinks on duplex DNA with a preference for 5'-GG platinated site to protect it from DNA repair and thus to increase cytotoxicity (G.Y. Zhu et al., 2010), particularly in MSH3-deficient cells (Takahashi et al., 2011) (**5**). Such side effect of PARP-1 orientates current phase I/II clinical trials using PARP inhibitors (CEP-6800, AZD2281 or ABT-888) as sensitizing agents in combination with cisplatin and carboplatin. A recent paper suggests that PARP is a pharmacological target of platinum- and other metal-based drugs showing PARP inhibition using Pt- (cisplatin), Ru- (RAPTA-T, NAMI-A) or Au- (Auphen, Aubipy) derived drugs (Mendes et al., 2011).

In a general manner, NER process of DNA lesions induced by ruthenium-drug appears to be less efficient than for platinum adducts. Ru-CYM and Ru-THA destabilize the DNA helix via different enthalpic effects and differ in terms of their DNA base-pair intercalation propensities. Comparison of their DNA repair processes has been used as a model for understanding the link between DNA destabilization and repair. Interestingly, Ru-CYM adducts (that destabilize the DNA helix much more than Ru-THA adducts) are excised more efficiently than Ru-THA complex adducts. Such observation is in good agreement with lower binding of RPA helicase to Ru-THA- than to Ru-CYM- adducts (Nováková et al., 2005). Ru-THA is also more cytotoxic than Ru-CYM, suggesting that DNA destabilization plays a major role in the cytotoxicity of these series of compounds.

#### **3.4 DNA repair for the carcinogen BaP (BPDE) and 4-OHEN adducts**

In prokaryote, the NER sensor protein UvrB recognizes BPDE/DNA adduct (**1** in Figure 8). Lesion-induced local thermodynamic destabilization and associated nucleotide flipping facilitate this recognition (Jia et al., 2009) with excision efficiencies changing up to a factor of 3 with stereoisomery (i.e. (+) *vs*. (-), *cis*- *vs*. *trans*-orientation)(Zou & Van Houten, 1999).

By contrast, the BaP-induced lesions are recognized in eukaryotic higher cells by the NER machinery's "sensor" protein XPC, associated with HR23B to initiate DNA repair (**2**). Weaker recognition by XPC/HR23B complex of the (+)-*trans*-B[*a*]P-N2-dG adduct, relatively

the two platinated guanines in 5'GG dinucleotide stabilizes the binding but is regulated in a redox manner. Indeed, the formation of a disulfure bond between the thiol groups of Cys22 and Cys44 on helix II and III, respectively, of HMG-box infers with the correct planar insertion of Phe37 between the two guanines at crosslink site (Park & Lippard, 2011). Binding of HMG-B1 (and HMG-B2) stabilizes the cisplatin-induced bent and supercoiling of the DNA helix, increases the sensitivity of the cells to cisplatin and shields the platinated adducts from repair by the human DNA excision machinery (J.C. Huang et al., 1994). As a consequence of the degree of kink of the DNA, HMG proteins poorly bind to oxaliplatin adducts which induce relatively small DNA-bending and DNA destabilization (Figure 3), and so poorly protects them from DNA repair (Kasparkova et al., 2008b). This difference correlates with the lower level of DNA lesions in oxaliplatin- versus cisplatin-treated cells. If HMG-B1 and –B2 binding participates in platinated-agent-induced cytotoxicity (Sharma et al., 2009), bent platinated-DNA is also a good substrate for transcription factors from HMGbox family such as SRY, LEF-1 and UBF-1, resulting in the transcriptional changes observed in treated cells (Chvalova et al., 2008; Treiber et al., 1994; Trimmer et al., 1998). For the repair of other platinum derivative-induced DNA damages, JM108 evidenced higher level of protein/DNA cross-links such as DNA-PtII-NF-κB cross-linked complexes (**4**). Those lesions are less efficiently removed from DNA by the cell repair system (Kostrhunova et al., 2010). Other studies described the binding of PARP-1 protein to cisplatin adduct at 5'-GG and 5'- GNG intra-strand crosslinks on duplex DNA with a preference for 5'-GG platinated site to protect it from DNA repair and thus to increase cytotoxicity (G.Y. Zhu et al., 2010), particularly in MSH3-deficient cells (Takahashi et al., 2011) (**5**). Such side effect of PARP-1 orientates current phase I/II clinical trials using PARP inhibitors (CEP-6800, AZD2281 or ABT-888) as sensitizing agents in combination with cisplatin and carboplatin. A recent paper suggests that PARP is a pharmacological target of platinum- and other metal-based drugs showing PARP inhibition using Pt- (cisplatin), Ru- (RAPTA-T, NAMI-A) or Au- (Auphen,

In a general manner, NER process of DNA lesions induced by ruthenium-drug appears to be less efficient than for platinum adducts. Ru-CYM and Ru-THA destabilize the DNA helix via different enthalpic effects and differ in terms of their DNA base-pair intercalation propensities. Comparison of their DNA repair processes has been used as a model for understanding the link between DNA destabilization and repair. Interestingly, Ru-CYM adducts (that destabilize the DNA helix much more than Ru-THA adducts) are excised more efficiently than Ru-THA complex adducts. Such observation is in good agreement with lower binding of RPA helicase to Ru-THA- than to Ru-CYM- adducts (Nováková et al., 2005). Ru-THA is also more cytotoxic than Ru-CYM, suggesting that DNA destabilization

In prokaryote, the NER sensor protein UvrB recognizes BPDE/DNA adduct (**1** in Figure 8). Lesion-induced local thermodynamic destabilization and associated nucleotide flipping facilitate this recognition (Jia et al., 2009) with excision efficiencies changing up to a factor of 3 with stereoisomery (i.e. (+) *vs*. (-), *cis*- *vs*. *trans*-orientation)(Zou & Van Houten, 1999). By contrast, the BaP-induced lesions are recognized in eukaryotic higher cells by the NER machinery's "sensor" protein XPC, associated with HR23B to initiate DNA repair (**2**). Weaker recognition by XPC/HR23B complex of the (+)-*trans*-B[*a*]P-N2-dG adduct, relatively

Aubipy) derived drugs (Mendes et al., 2011).

plays a major role in the cytotoxicity of these series of compounds.

**3.4 DNA repair for the carcinogen BaP (BPDE) and 4-OHEN adducts** 

to that of the other conformers, contributes to its higher mutagenic potential (Mocquet et al., 2007). Lesion recognition by XPC requires DNA bending facilitated by local conformational flexibility (Clement et al., 2010) and destabilization of the base-pairing (Brown et al., 2010). Such recognition is driven by Trp690 and Thp733 amino-acids identified as "aromatic sensors" (Maillard et al., 2007). Upon treatment with BaP, human bronchial epithelial 16HBE cells expressed higher levels of the heat shock protein 70 and the NER proteins XPA and XPG, both three proteins co-localizing in the nucleus, suggesting that Hsp70 is also implicated in the DNA repair response to BPDE-DNA adducts (J. Yang et al., 2009). The highly mutagenic (*+*)-(7R,8S,9S,10R)-7,8-dihydroxy-9,10-epoxy-7,8,9,10-tetrahydrobenzo[*a*]pyrene-DNA lesion leads to different repair processes depending on sequence context, associated with the destabilization potency. Indeed, for an identical BaP-DNA lesion leading to differently orientated bulky lesions, sequence-dependent effect was observed: DNA destabilized at 5'-CG\*GC site is more rapidly excised in cell-free human HeLa extracts than DNA bent at 5'-CGG\*C site (Rodríguez et al., 2007). As the DNA helix is readily opened upon alkylation, recognition of the lesion by repair protein (including induction of base flipping) is less energetic and, thus, is quicker for DNA already destabilized at 5'-CG\*GC site than for duplex DNA bent at 5'-CGG\*C site, clearly evidencing the importance of DNA sequence/global structure context for an efficient repair of BPDE-DNA adducts (Yuqin et al., 2009). Moreover, interesting data arise from comparison of the 3D conformation and the NER excision efficiencies for dA adducts formed using the bay region BPDE and the fjord region benzo[*c*]phenanthrene diol epoxide (B[*c*]PhDE) (M. Wu et al., 2002). The bay region of B[*a*]P is more extended, planar and rigid than the B[c]Ph fjord region, being twisted and curved. Consequently, B[*a*]P-dA adducts are associated with greater backbone distortion, unwinding, intercalation potency and

Fig. 8. DNA repair pathways for BPDE-induced DNA damage.

DNA Helix Destabilization by Alkylating Agents: From Covalent Bonding to DNA Repair 111

(Q. Wu et al., 2005; 2008). Psoralen monoadducts are good substrates for 3-Methyladenine DNA glycosylase (MPG) (Maor-Shoshani et al., 2008) and the human oxidative DNA glycosylase, NEIL1, which catalyses the ,-elimination at AP site, leaving a 3'-P termini at the resulting SSB (**5**) (Couvé-Privat et al., 2007). Fanconi anemia pathway was also implicated in the repair process, in link with NEIL1 stability and NER efficiency (Macé-

S23906-1 alkylates the DNA in the minor groove and induces a strong destabilization of the DNA helix. Two reactive acetate groups are positioned on asymmetric carbons leading to four pure enantiomers: 2 *cis* (1R;2R and 1S;2S) (the *cis*-racemate being S23906-1) and two *trans* (1R;2S and 1S;2R) isomers. Both pure enantiomers react with DNA and destabilize the DNA helix but at different extends. The most potent DNA destabilizing ones (1S;2S and 1S;2R) being those presenting the most active anti-tumour activities in animal models (Depauw et al., 2009). Therefore, the rate of DNA destabilization is different depending on the orientation of the core of the adducts regarding the opened drug/DNA structure, and correlates with different cellular and anti-tumour effects. Such strong destabilisation could affect single-stranded endonuclease and DNA repair activities. There is currently only partial knowledge on the repair of S23906-1 DNA adducts. The NER proteins XPC and CSB are involved in cell sensitivity to S23906-1, associated with both global genome repair and transcription-coupled NER (Rocca et al., 2010). ATR coordination, RPA recognition and Chk1 activation were also implicated in responses to S23906-1 DNA damages (Soares et al., 2011). Process of the lesions is associated with DSB as secondary DNA lesions important for cytotoxicity of S23906-1, associated with histone H2AX phosphorylation (Léonce et al., 2006). Of major interest, the most potent destabilizing isomer of S23906-1 was evidenced to be also the most cytotoxic on cellular models and the most efficient on xenografted animal models (Depauw et al., 2009). Current ongoing research is identifying proteins implicated in S23906-1/DNA adduct recognition and evaluating their impact on S23906-1 cytotoxic activity (personal communication). Locally destabilized DNA could favour the recognition of DNA lesion by "DNA repair sensors" thus increasing the efficiency/kinetic of the

Destabilization of the DNA helix that is induced by drugs is an important aspect of the antitumor mechanism of action of this series of compounds besides they represent just few droplets in an ocean of DNA-interacting compounds that mainly stabilize the double helix. As evidenced here, stabilizing *vs.* destabilizing compounds differs in terms of molecular and cellular processes: DNA repair, transcription or replication. From the different series (platinum, ruthenium, BPDE, benzoacronycines), the level of DNA destabilization correlates with the efficiency of protein recognition and anti-tumor/cytotoxic activities. Therefore, we believe that it is important not to consider DNA destabilization as a unique process but in relation with potential associated bending of the DNA helix (as evidenced using oxaliplatinand cisplatin-induced distortions or the different isomers of BPDE) and with the size of the locally destabilized DNA (for instance, portions of DNA opened by benzoacronycines are strongly sensitive to single-strand-specific nucleases). The most recent and ongoing studies

Aimé et al., 2010).

removal of the DNA lesion.

**4. Conclusion** 

**3.6 DNA repair for benzoacronycine-DNA adducts** 

disturbed Watson-Crick hydrogen bonding than B[*c*]Ph-dA adducts, in correlation with stronger excision efficiency by NER machinery. The fjord region B[*c*]Ph-dA adducts being poorly excised lead to more tumorigenic activities. HMG-1 and -2 proteins are also implicated in bulky BPDE-adducts recognition (Lanuszewska & Widlak, 2000) but the consequences on repair or cell death are unknown (**3**). HMG binding might protects adduct recognition by repair proteins as for platinated DNA, but this needs further evaluation. Excision of bulky 4-OHEN-DNA adducts by NER proteins also depends on both the nature of the alkylated base, its stereo-isomery and the sequence context. For instance, 4-OHEN-dC adducts are more efficiently excised from the DNA than the 4-OHEN-dA adducts (D. Chen et al., 2006). It was reported in male zebrafish that 17a-ethinylestradiol, as a source of 4-

OHEN, induces a decrease in NER activity as part of a decrease of the expression level of some NER genes such as XPC, XPA, XPD and XPF, but not of HR23B (Notch et al., 2007).
