Cells have evolved intricate DNA repair mechanisms to circumvent genomic instability. Thousands of spontaneous single-strand breaks (SSBs) occur in cellular DNA each day 1 and, if they persist, can convert to potentially lethal double-strand breaks (DSBs). Accordingly, highly efficient and diverse mechanisms for SSB repair (SSBR) have evolved (Table 1). SSBs commonly have damaged or blocked termini that lack the conventional 5′-phosphate and the 3′-hydroxyl required for polymerase activity and strand ligation. Thus, further processing of blocked DNA ends is required prior to DNA synthesis (3′-end-tailoring) and ligation (5′-end-tailoring), and such tailoring of blocked termini is frequently the rate-limiting step in a repair cascade 2, 3.

Table 1 Repair of cellular SSBs

Repair of IR-induced DNA damage

Endogenous sources of SSBs include attack by reactive oxygen and alkylating species, as well as breaks that arise from the inherent instability of DNA. Direct SSBs can also be induced after sugar damage and disintegration of the DNA backbone following absorption of ionizing radiation (IR), or through IR-mediated formation of reactive oxygen species (ROS) 4. SSBs resulting from IR commonly have 3′-phosphate or 3′-phosphoglycolate groups that are substrates for apurinic/apyrimidinic endonuclease (APE) 5, 6 or the oculomotor apraxia type 1 gene product aprataxin 7 (Table 1). Additionally during repair of IR-induced SSBs, blocking groups can be removed by the bifunctional enzyme polynucleotide kinase (PNK) (3′-phosphate) 8 or by tyrosyl DNA phosphodiesterase (Tdp1) (3′-phosphoglycolate) 9, 10 (Table 1). When DNA ligases attempt to repair non-ligatable blocked ends induced by ROS, abortive intermediates are formed with an adenylate group bound to the 5′-phosphate. Aprataxin has a hydrolase activity able to release the adenylate groups and producing a 5′-phosphate 11.

The initiation of BER of IR- and other ROS-mediated oxidative base damage can result in formation of indirect SSBs generated as intermediates of BER. In general, oxidized bases are excised by bifunctional glycosylases that have an associated AP lyase activity able to nick the DNA strand 3′ to the abasic site following base removal. In the case of glycosylase-catalyzed β-elimination associated with 8-oxoguanine-DNA glycosylase (OGG1) for example 12, APE is required to remove the blocking 3′-deoxyribosephosphate (dRP) group prior to gap-filling synthesis 5 (Table 1). Abasic site cleavage by glycosylase-associated β,δ-elimination, for example by NEIL1 or 2 13, 14, generates 3′-phosphate termini that can be removed by PNK in an APE-independent repair pathway 15 (Table 1). Alternatively, the AP site can be 5′-incised by APE, circumventing the glysosylase-associated AP lyase step 16. Oxidized abasic sites result in formation of 5′-oxidized dRP that can be removed by flap endonuclease 1 (FEN1) as part of the alternate long-patch BER sub-pathway 17 (Table 1).

Monofunctional glycosylase-initiated BER

Indirect SSBs similarly arise when BER is initiated by a monofunctional glycosylase. For example, N-methylpurine-DNA glycosylase (MPG) removes damaged bases (e.g. N7-alkyl guanine) occurring following exposure to simple alkylating agents such as methyl methanesulfonate (MMS). The resulting abasic site undergoes 5′-strand incision by APE. Removal of the 5′-dRP blocking group is catalyzed by the dRP lyase activity of DNA polymerase β (pol β), and, in the preferred single-nucleotide BER sub-pathway, pol β also performs single-nucleotide gap-filling synthesis (Table 1) 2. When a modified dRP group is not a substrate for the lyase activity, pol β-dependent strand displacement synthesis, in conjunction with FEN1-mediated flap cleavage, can conduct long patch BER 18, 19. It should be noted that pol β-independent single-nucleotide and long patch BER pathways have also been well documented 20, 21.

Formation of topoisomerase I-DNA complex

Another source of SSBs is the transient single-strand nicking by topoisomerase I (Top1) ahead of DNA replication. Under normal conditions, Top1 religates the breaks once relaxation of DNA supercoiling has occurred. This religation process is inhibited by Top1 inhibitors such as camptothecin, resulting in trapping of Top1 cleavage complexes and formation of protein-linked SSBs 22. Following processing by the proteosome, Tdp1-mediated hydrolysis removes the truncated form of Top1 that is covalently linked to DNA. This produces SSBs with a 3′-phosphate and a 5′-hydroxyl that can be removed by PNK, and in some cases subsequently extended by pol β prior to ligation (Table 1) 9, 23. Cytotoxic DSBs arise by collision between replication forks and the stabilized Top1 cleavage complexes 24.

Role of pol β in DNA repair

The X-family polymerase, pol β, is organized into two distinct domains, a 31-kDa polymerase domain and an 8-kDa dRP lyase domain. The polymerase activity of pol β has been implicated in gap-filling synthesis during repair of all the DNA lesions described above 23, 25, 26 (Table 1), but this activity is generally not rate-determining in repair. In contrast, the dRP lyase activity normally contributed by pol β in cells is essential and rate-limiting during monofunctional glycosylase-initiated BER 2. The hallmark MMS-hypersensitivity phenotype of pol β deficiency is thought to be a result of failure to repair the cytotoxic 5′-dRP intermediate of BER 27, 28.

Role of XRCC1 in DNA repair

X-ray cross-complementing group 1 (XRCC1) is a 70-kDa protein comprising three functional domains; an N-terminal DNA binding domain, a centrally located BRCT I and a C-terminal BRCT II domain. It has no known enzymatic activity. Since it specifically interacts with nicked and gapped DNA in vitro 29, 30, and rapidly and transiently responds to DNA damage in cells, it may serve as a strand-break sensor 31, 32. However, complementation experiments in XRCC1-deficient CHO cells suggest that the interaction of XRCC1 with DNA is not critical for efficient SSBR .

In addition, since XRCC1 interacts with many proteins known to be involved in BER and SSBR, it has been proposed that XRCC1 functions as a scaffold protein able to coordinate and facilitate the steps of various DNA repair pathways 32, 34. For example, XRCC1 interacts with several DNA glycosylases involved in repair of both oxidative and alkylated base lesions, and stimulates their activity 35, 36. XRCC1 also interacts with the N-terminal of APE stimulating both its AP endonuclease and 3′-dRPase activities 37. Binding of XRCC1 to PNK enhances its capacity for damage discrimination, and binding of XRCC1 to DNA enables displacement of PNK from the phosphorylated product 34 thus accelerating SSBR of damaged DNA 38. XRCC1 associates with Tdp1 and enhances its activity required for repair of Top1-associated SSBs. It may act to recruit Tdp1 to these damaged sites 23. Biochemical and NMR experiments have demonstrated protein-protein interaction between the N-terminal domain of XRCC1 and the polymerase domain of pol β 39, 40, 41, 42. Additionally, stabilization of DNA ligase IIIα is dependent on its interaction with the BRCT II domain of XRCC1 43. Aprataxin also interacts with XRCC1 and functions to maintain XRCC1 stability, thus further linking the neurological degeneration associated with ataxia to an inefficiency of SSBR 44, 45, 46.

Poly(ADP-ribose)-mediated recruitment to repair foci

Poly(ADP-ribose) polymerase (PARP)s-1 and 2, members of a family of at least 18 proteins with poly(ADP-ribosyl)ating activity, interact with the BRCT I domain of XRCC1 47, 48. Upon detection of DNA nicks and binding to damaged DNA, the activity of PARP-1 in particular is rapidly stimulated and, using NAD+ as substrate, it poly(ADP-ribosyl) ates multiple proteins including itself 49. As a consequence of self poly(ADP-ribosyl)ation, PARP-1 loses affinity for DNA and is released from its binding site permitting access of repair proteins 50, 51. Binding and activation of PARP-1 may function to sequester other DNA repair proteins to sites of SSBs. For example, XRCC1 preferentially interacts with automodified PARP-1 47, 52. In this way, activated PARP-1 recruits XRCC1 to sites of oxidative and methylated DNA damage suggesting that formation of repair foci may be mediated by poly(ADP-ribose) (PAR) 53, 54. In addition, XRCC1 is a substrate for PARP-1-mediated poly(ADP-ribosyl)ation thus confirming the functional interaction between these two proteins 47. DNA ligase IIIα associates with poly(ADP-ribosyl)ated PARP-1 providing another possible mechanism for recruitment of the XRCC1-ligase IIIα complex to sites of SSBs 55. Consistent with the idea that these proteins together are essential for SSBR, PARP-1, like XRCC1, also interacts with aprataxin 45.

Maintenance of genomic stability

Both XRCC1 and pol β play a significant role in maintaining chromosomal stability. An elevated level of sister chromatid exchange, widely used as an indicator of genetic damage, is characteristic of XRCC1 and pol β deficiency 56, 57, 58, 59. Genomic instability is proposed to be a factor in tumor initiation 60. Polymorphisms of XRCC1, and possibly pol β, have been associated with a higher risk of cancer 61, 62, 63, 64. Approximately 30% of human tumors express pol β variants, some of which are associated with a mutator phenotype (summarized in 65), and overexpression of pol β is a common event in tumorigenesis 66. It is therefore interesting to compare and contrast the phenotypes and repair defects associated with the absence of XRCC1 and pol β in cells.

Available XRCC1- and pol β-deficient cell lines

The XRCC1-mutant strains of Chinese hamster ovary (CHO) cells, EM7 and EM9, were isolated from the parental strain, AA8, based on hypersensitivity to the alkylating agent ethyl methanesulfonate (EMS) 67. Alternate mutant strains (EM-C11 and EM-C12) also lacking functional XRCC1 were similarly isolated from CHO-9 cells 57, 68. Defects in EM9 cells could be corrected by transfection with a human or mouse DNA repair gene designated XRCC1 since it was the first mammalian gene isolated that affects cellular sensitivity to X-rays 69, 70, 71. No full length XRCC1 protein nor truncated forms could be detected in any of the mutant CHO cell extracts 68, 72, and sequence analysis revealed debilitating point mutations in the cDNAs of all the mutant cell lines 68. Alterations in the encoded amino acid sequence are presumed to affect protein folding and protein-protein interactions, or result in prevention of translation of full-length functional protein 68.

Gene targeting in mice to produce a null mutation in the XRCC1 gene resulted in early embryonic lethality and accumulation of endogenous DNA damage 73. Creation of XRCC1-p53 double knock-out embryos resulted in a delay of lethality, but embryos failed to survive to term. However, it was possible to isolate XRCC1−/− mouse embryonic fibroblasts 73. Both the embryonic lethality, and the hypersensitivity to DNA damage observed in XRCC1-deficient cells, could be reversed by even low-level expression of XRCC1 protein 73, 74.

A homozygous pol β gene deletion similarly resulted in embryonic lethality 75 suggesting there is a requirement for BER and/or SSBR during early mouse development. Mouse embryonic fibroblasts could be established from these homozygous deletion mice at 10 days gestation and were transformed by expression of the SV40 large-T antigen 26. In another mouse system, targeted disruption of the pol β gene resulted in growth retardation, developmental defects of the nervous system and death from respiratory failure immediately after birth 76. There was dramatically decreased neuronal apoptosis in a double knock-out model (pol β−/−, p53−/−), but developmental defects were still evident and the mice died shortly after birth 77.

Characteristics of XRCC1- and pol β-deficient cells: hypersensitivity and repair deficiency following treatment with IR

The CHO cell mutant EM9 demonstrates sizeable hypersensitivity to the alkylating agent EMS, but is reported to be only 2-fold hypersensitive to X-rays 56 (reviewed in Table 2). Another CHO cell strain showed even lesser hypersensitivity to IR 57. Consistent with results obtained in XRCC1-deficient CHO cells, XRCC1 genetic deficiency in mouse fibroblasts also resulted in minimal hypersensitivity to IR (Figure 1A) 73. In addition, no hypersensitivity was observed to the radiomimetic agent, bleomycin (Table 2). Similarly, significant hypersensitivity to IR was not observed in fibroblasts deficient in pol β (Figure 1B) 26, 78.

Table 2 DNA damage-induced hypersensitivity phenotypes associated with XRCC1 and pol β deficiency
Figure 1
figure 1

Survival of mouse embryonic fibroblasts following exposure to IR and repair of IR-induced DNA damage. (A) XRCC1+/+ (closed circles) and XRCC1−/− (open circles) and (B) Pol β+/+ (closed triangles) and pol β−/− cells (open triangles) were exposed to IR and survival was determined by growth inhibition assays as described 119. The dashed line indicates the dose of IR used in the experiments presented in panels C and D. Data are from representative experiments where the values are the mean ± SEM of 4-5 determinations. (C and D) DNA damage (median OTM) was analyzed in non-treated cells (NT), and cells irradiated (5 Gy) then allowed to repair for 15 and 30 min as shown. The XRCC1 73 and pol β 26 cell lines have been described previously.

Irradiation of cells has been shown to generate DSBs and SSBs (ratio of 1:25 79), that can be detected in individual cells by the alkaline comet assay 80, 81. IR-mediated DSBs are generated from locally multiply damaged sites arising due to ionization tracks and radiolysis of water, and DSBs are believed to be responsible for most IR-induced lethality 82. In addition, IR-induced modification of bases and sugars (ratio of oxidized bases:SSBs is 3:1 79) results in formation of DNA nicks and alkali-labile sites following initiation of BER. Repair of IR-induced DNA damage in wild-type CHO cells 70, 83, as well as mouse fibroblasts, is extremely rapid (Figure 1C and 1D). A reduced rate of repair of DNA SSBs formed in irradiated XRCC1 mutant CHO cells has been identified by alkaline elution and the alkaline comet assay 56, 70, 83. Similarly, plots of median Olive Tail Moment (OTM) 81 demonstrate there is a delay in repair of IR-induced damage in XRCC1−/− mouse fibroblast cells compared with wild-type cells following exposure to 5 Gy of irradiation (Figure 1C). Thus the comet assay data confirm that there is repair deficiency associated with the absence of XRCC1 protein.

XRCC1 has several roles in the repair of DSBs, as well as SSBs, thus making a repair deficiency of IR-induced DNA damage likely in the absence of XRCC1. For example, PNK interacts with CK2-phosphorylated XRCC1 84, and XRCC1 is known to stimulate the activities of PNK required for end-processing during SSBR (Table 1) as well as during repair of DSBs by non-homologous end-joining (NHEJ) 85. Further, a recently described PARP-1-dependent back-up pathway of NHEJ involves the XRCC1-ligase IIIα complex as well as PNK 86, 87. In addition, XRCC1 is known to interact with the catalytic subunit of DNA-dependent protein kinase (DNA-PK) via its BRCT I domain. DNA-PK facilitates recruitment of proteins required for NHEJ, and XRCC1 is phosphorylated by DNA-PK in response to IR-induced DNA damage 88, thus providing evidence for its involvement in DSB repair. Interestingly, repair measured in XRCC1−/− cells by the comet assay, although delayed, is still relatively fast (<65% repair of damage by 30 min; Figure 1C) and only a minimal IR hypersensitivity phenotype is observed (Figure 1A). It seems that XRCC1−/− cells have a significant capacity to repair IR-induced cytotoxic damage even in the absence of DSB repair mechanisms in which XRCC1 has been implicated.

In contrast to results obtained in XRCC1-deficient cells, a delay in rejoining of strand breaks in pol β−/− cells was not detected (Figure 1D). Thus, the absence of IR hypersensitivity in replicating pol β−/− cells (Figure 1B, Table 2) correlates well with these repair data. The simplest explanation for these results is that pol β is not critical in cells for repair of IR-induced DNA damage (both cytotoxic and non-cytotoxic). Pol β has been implicated in gap-filling during BER and SSBR, but not in repair of DSBs by homologous recombination or NHEJ. Inefficient repair of SSBs is expected to result in an increased incidence of DSBs, however, since pol β is generally not needed for DNA gap-tailoring during the repair of oxidized bases and sugars, or SSBs (Table 1), overall repair may remain efficient in its absence. Alternatively, there might be robust mechanisms for repair of IR-induced DNA damage that, in the absence of pol β, can efficiently substitute for pol β-dependent pathways. A recent study has described IR hypersensitivity in non-dividing pol β null cells suggesting that pol β-dependent repair is masked by alternate replication associated repair pathways in cycling cells 89.

Characteristics of XRCC1- and pol β-deficient cells: hypersensitivity and repair deficiency following exposure to MMS

Despite being named for their X-ray cross-complementing gene deficiency, CHO cells with mutated XRCC1 are significantly more sensitive to the monofunctional alkylating agents, EMS and MMS, than to IR 56. Similarly, and in contrast to the low hypersensitivity observed for IR, XRCC1−/− mouse fibroblasts are at least 10-fold more sensitive than XRCC1+/+ cells to both EMS 73 and MMS (Table 2). Expression of wild-type XRCC1 protein complements the MMS and EMS hypersensitivity phenotype of both EM9 cells 33, 52, 53, 90, 91 and XRCC1−/− mouse fibroblasts 73 (Figure 2A). The high degree of EMS and MMS hypersensitivity indicates that XRCC1 is essential for resistance of cells to the cytotoxic effects of these simple alkylating agents. Strikingly, for both of these agents, the hypersensitivity of XRCC1−/− cells is considerably greater than the 2-3-fold differential in sensitivity observed in pol β+/+ and pol β−/− cells (Figure 3, Table 2). The hypersensitivity of pol β−/− mouse fibroblasts to monofunctional alkylating agents has been proposed to reflect accumulation of cytotoxic intermediates of BER in the absence of pol β-mediated repair 28. The more profound hypersensitivity of XRCC1−/− cells suggests that XRCC1 may be required for efficient BER by both pol β-dependent and -independent repair pathways.

Figure 2
figure 2

Complementation of XRCC1−/− hypersensitivity phenotypes by expression of the XRCC1 gene. XRCC1+/+ (closed circles) and XRCC1−/− mouse embryonic fibroblasts (open circles) or XRCC1−/− fibroblasts transfected with mouse XRCC1 cDNA (closed diamonds) were treated with (A) MMS for 1 h, (B) TMZ for 4 h, (C) hmdUrd for 24 h, or (D) camptothecin for 24 h. Cell sensitivity was determined by growth inhibition assays as described 119. The cell lines have been described previously 73.

Figure 3
figure 3

MMS hypersensitivity in XRCC1−/− compared with pol β−/− mouse embryonic fibroblasts. Sensitivity of XRCC1−/− (open circles) and pol β−/− cells (open triangles) to a 1 h exposure to MMS is compared with that of the respective isogenic wild-type (+/+) cell lines (closed symbols). Cell sensitivity was determined by growth inhibition assays as described 119. The cell lines have been described previously 26, 73. Data taken from ref. 106.

A defect in repair of MMS-induced SSBs has been demonstrated by the alkaline comet assay (which will detect cytotoxic abasic sites and SSB intermediates of BER) in XRCC1-deficient CHO cells 33, 52, 91. Following a short exposure of mouse embryonic fibroblasts to ice cold MMS, higher levels of DNA damage were seen than found in control untreated cells (NT), but levels were similar in the paired isogenic repair-deficient and wild-type cell lines (Figure 4, panels A and B). The survival of the cells following this same MMS treatment protocol is also presented (Figure 4, panels C and D). In both XRCC1 and pol β wild-type mouse fibroblasts, the level of DNA damage (median OTM) remained fairly constant during the repair incubation suggesting that the formation, and the repair, of strand breaks and alkali-labile sites are occurring at a similar rate (Figure 4, panels A and B). In contrast, elevated levels of DNA damage (median OTM) and diminished repair was observed in XRCC1−/− and, to a lesser extent, in pol β−/− cell lines compared with wild-type cells (Figure 4, panels A and B). Since a good correlation is seen between the MMS sensitivity of the different mouse fibroblast lines and their ability to repair DNA damage arising following exposure to MMS (Figure 4), the results support the hypothesis that XRCC1−/− cells, like pol β−/− cells, are deficient in cellular BER of MMS-induced damage.

Figure 4
figure 4

Increased DNA damage following exposure of repair-deficient mouse embryonic fibroblasts to MMS. (A and B) DNA damage (median OTM) after treatment of cells (as indicated) for 20 min with 10 mM MMS in ice-cold medium, and then during repair in warm medium for up to 120 min. NT are non-treated cells. (C) Sensitivity of XRCC1+/+ (closed circles) and XRCC1−/− (open circles) and (D) Pol β+/+ (closed triangles), pol β−/− (open triangles) cells to a 20 min exposure to MMS in ice-cold medium. Growth inhibition assays were conducted as described 119. Data are from representative experiments where the values are the mean ± SEM of 3 independent determinations. The dashed line indicates the concentration of cold MMS used in the experiments presented in panels A and B.

In the absence of pol β, cellular hypersensitivity to MMS is associated with a deficiency in the rate-limiting pol β-dependent dRP lyase activity, rather than its polymerase activity, in the predominant single-nucleotide BER pathway 27. Perhaps XRCC1 in some way facilitates 5′-dRP removal by pol β during monofunctional glycosylase-initiated BER, although such an effect has not been documented. XRCC1 and pol β proteins are known to physically interact 41, 42, and there is a rapid but lesser accumulation of pol β at sites of DNA base damage in cells in the absence of XRCC1 31. XRCC1 may also be required for accumulation of other proteins with dRP lyase activity (e.g., DNA polymerase λ). Since XRCC1-deficient cells are more sensitive to MMS than are pol β -deficient cells, XRCC1 clearly has a role in protection of cells against MMS-induced cytotoxicity in addition to mediating recruitment of pol β. In fact complementation of EM9 cells with a mutant XRCC1 (V86R) that disrupts the interaction with pol β is known to result in significant correction of the MMS hypersensitivity phenotype 33. The effects of XRCC1 may be additive since it can coordinate the entire BER pathway for MMS-induced DNA damage through protein-protein interactions and stimulation of the glycosylase, APE and ligase activities implicated in MMS-induced single-nucleotide BER 36, 37, 43. Interestingly, expression in EM9 cells of an XRCC1 BRCT II mutant protein that does not interact with ligase IIIα is able to fully complement hypersensitivity to MMS in the absence of a stabilizing effect on ligase IIIα 90, 91. Another possibility is that the protective role of XRCC1 is not entirely related to DNA repair, but that it modifies damage-dependent signaling resulting in cell death.

Characteristics of XRCC1- and pol β-deficient cells: hypersensitivity to DNA damage requiring BER

XRCC1−/− cells are more hypersensitive to the SN1 type methylating agent N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) (22-fold) than to the SN2 type agent MMS (Table 2) whereas lesser (2-fold) hypersensitivity was described in EM9 cells 56, 92. This discrepancy may be related to the ability of these cell types to directly repair MNNG-induced cytotoxic O6-methylguanine lesions by a process that is not dependent on cellular BER. XRCC1-deficient mouse fibroblasts are also significantly hypersensitive to the chemotherapeutic methylating agent temozolomide (TMZ) (Figure 2B, Table 2) and, like EM9 cells 93, they are extremely hypersensitive to 5-hydroxymethyl-2′-deoxyuridine (hmdUrd) (Figure 2C, Table 2). It is known that pol β-dependent BER is initiated following uracil-DNA glycosylase (SMUG1)-mediated removal of the abnormal base, hydroxymethyluracil, from cellular DNA 28. Transfection with XRCC1 cDNA is able to fully rescue these hypersensitivity phenotypes of XRCC1−/− cells (Figure 2B and 2C). Similar to MMS, XRCC1−/− fibroblasts are considerably more hypersensitive to these agents than are pol β null BER-deficient mouse fibroblasts (Table 2) 28. The hypersensitivity data obtained for hmdUrd in XRCC1−/− cells, as well as with the DNA methylating agents MMS, MNNG and TMZ, are consistent with the proposal that XRCC1 has a role in facilitating BER. In common between the methylating agents and hmdUrd is that BER of the resulting DNA lesions is initiated by a monofunctional glycosylase.

EM9 cells demonstrate only slight hypersensitivity (<2-fold) to bifunctional alkylating compounds that form DNA interstrand cross-links. Those tested include the clinically utilized agents mitomycin C and the nitrogen mustards melphalan and chlorambucil 56, 94, 95. In contrast, XRCC1−/− fibroblasts have significant hypersensitivity to monofunctional mustards that cannot cross-link DNA but instead result in formation of SSBs (Figure 5, panels A and B, Table 2). These compounds, 2-dimethylaminoethyl chloride (dimethyl mustard, DMM) and 2-diethylaminoethyl chloride (diethyl mustard, DEM) produce N7-alkylguanine adducts in DNA 96. Pol β−/− mouse fibroblasts also demonstrate hypersensitivity, but to a lesser extent (2-fold) (Table 2). The results suggest that these guanine adducts, although considerably more bulky than methylated DNA, are repaired at least in part by pol β-dependent BER. In support of this hypothesis, pamoic acid (PA), an inhibitor of pol β 97, was able to significantly sensitize wild-type, but not pol β−/− fibroblasts to DMM- and DEM-induced cytotoxicity (Figure 5C).

Figure 5
figure 5

XRCC1- and pol β-dependent hypersensitivity to monofunctional mustards. XRCC1+/+ (closed circles) and XRCC1−/− (open circles) mouse embryonic fibroblasts were treated with (A) DMM or (B) DEM for 1 h. (C) Pol β+/+ (closed symbols) and pol β−/− (open symbols) cells were treated with DMM for 1 h in the presence (squares) or absence (triangles) of pamoic acid (PA) (pretreatment with 300 μM for 7 h, then continued incubation following DMM for a total of 24 h). Data represent the mean ± SEM of at least 3 independent experiments. Cell sensitivity was determined by growth inhibition assays as described 119.

Other types of bulky DNA adducts, such as those formed after exposure to UV or cisplatin, are generally considered to be repaired by nucleotide excision repair rather than BER. There is recent evidence that the XRCC1-ligase IIIα complex is required for ligation of breaks arising during nucleotide excision repair of UV lesions, but is indispensable only in quiescent cells 98. However, neither XRCC1 mutant CHO cells 57, 95, nor XRCC1−/− fibroblasts, are hypersensitive to UV-induced damage or cisplatin exposure when studies are conducted in proliferating cells (Table 2).

Characteristics of XRCC1- and pol β- deficient cells: hypersensitivity to oxidative damage

Despite the known interactions of XRCC1 with a number of other repair proteins (e.g., APE and PNK) that contribute activities required in each step of BER and SSBR (Table 1), XRCC1−/− cells demonstrate only minimal hypersensitivity to the ROS-generating agents hydrogen peroxide and potassium bromate (Table 2). This result differs slightly from those obtained in XRCC1-defective CHO cells where a low (2-fold) hypersensitivity to hydrogen peroxide has been observed 99. The lack of hypersensitivity to oxidative DNA damage is in contrast to the sizeable hypersensitivity to MMS and other methylating agents, and strongly suggests that oxidized and methylated DNA bases are repaired by different BER sub-pathways in XRCC1−/− mouse fibroblasts. One potential difference will arise when oxidized bases are excised by bifunctional glycosylases, such as OGG1 with an associated lyase activity that cleaves the abasic site by β elimination 12, or NEIL1 and 2 that cleave the abasic site by β,δ-elimination 13, 14. During repair utilizing a bifunctional glycosylase, a 5′-dRP will not be formed and the dRP lyase activity of pol β will not be required for repair (Table 1). Pol β has been implicated in gap filling during repair of oxidative damage 100, 101, yet the initial characterization of pol β null fibroblasts did not reveal a hydrogen peroxide hypersensitivity phenotype 26. However, in older cells, low-level (<2-fold) hypersensitivity to ROS agents was observed 102, 103.

Characteristics of XRCC1- and pol β-deficient cells: hypersensitivity to Top1 inhibitors

Exposure of cells to camptothecin results in formation of SSBs with Top1 covalently linked to the 3′-termini. In CHO cells, XRCC1 has been shown to provide protection against the cytotoxicity of camptothecin 95, 104, and to enhance the repair of camptothecin-induced protein-linked SSBs 23. Like XRCC1-deficient CHO cells, XRCC1−/− mouse fibroblasts are moderately hypersensitive to camptothecin (Table 2) and expression of XRCC1 in XRCC1−/− fibroblasts can reverse the hypersensitivity phenotype (Figure 2D). This is in contrast to pol β−/− cells that demonstrate negligible hypersensitivity to this Top1 inhibitor (Table 2).

The cellular hypersensitivity observed in the absence of XRCC1, but not in the absence of pol β, is consistent with the proposed mechanism for reversal of the protein-DNA complex. Repair requires hydrolysis of the covalent Top1-DNA phosphotyrosyl bond by Tdp1, and then end-processing by PNK prior to DNA ligation, rather than pol β-dependent BER (Table 1). Tdp1 is critical for repair of this damage. It interacts both physically and functionally with XRCC1 23, and it is known that Tdp1 activity at Top1-SSBs is stimulated by the XRCC1/ligase IIIα complex in vitro 9. XRCC1 is also known to interact with PNK stimulating both the 3′-phosphatase and 5′-kinase activities at damaged termini, and acting to accelerate the overall rate of repair 38. Interestingly, XRCC1-deficient mouse fibroblast cells are not hypersensitive to etoposide, an inhibitor of topoisomerase II (Table 2).

Hypersensitivity phenotypes of XRCC1- and pol β-deficient cells: role of PARP activity

Characterization of XRCC1- and pol β-deficient cells has revealed an important role for both proteins in protection against the cytotoxic effects of agents resulting in DNA damage where repair is initiated by a monofunctional glycosylase (Table 2). Chemical inhibitors of PARP activity are able to enhance cell killing by these same agents that produce indirect strand breaks during BER. Specifically, exposure of wild-type and pol β−/− cells to the potent PARP inhibitor, 4-amino-1,8-napthalimide (4-AN), is known to extremely sensitize them to the cytotoxic effects of SN2 (e.g. MMS) and SN1 (e.g. MNNG and TMZ) monofunctional methylating agents, as well as hmdUrd 105 (reviewed in 106). For these agents, co-exposure to an inhibitor potentiates cytotoxicity in pol β−/− cells to a greater extent than in wild-type, thus resulting in an enhanced hypersensitivity phenotype. Results obtained in cell lines and animal models with the chemotherapeutic methylating agent TMZ have led to development of clinical trials of TMZ combined with a PARP inhibitor 107, 108, 109.

The BRCT I domain of XRCC1 capable of interacting with PARP-1 is critical for efficient SSBR and resistance to MMS-induced cytotoxicity 52, 91. It has been proposed that activation of PARP is required for recruitment of XRCC1 to sites of DNA damage. PAR synthesis by PARP-1 is known to recruit XRCC1 to repair foci in human and mouse fibroblasts, as well as HeLa cells 31, 53, 54. Interestingly, sensitization to EMS by the PARP inhibitor 3-aminobenzamide was only slightly reduced in XRCC1-deficient EM9 compared with wild-type AA8 cells 110. Similarly in mouse fibroblasts, PARP inhibition by 4-AN significantly sensitizes XRCC1−/− cells to methylating agents and to DMM, although to a lesser extent than wild-type cells (Figure 6, Table 3). These data suggest that PARP-mediated poly(ADP-ribosyl) ation plays a role in modulation of cytotoxicity beyond recruitment of XRCC1 to sites of DNA damage. Since XRCC1−/− cells still exhibit significant hypersensitivity to all of the agents compared with XRCC1+/+ cells when PARP activity is inhibited (i.e., in the presence of 4-AN), the results show that XRCC1 has a protective role in the cell in the absence of PARP activity (Figure 6).

Figure 6
figure 6

Hypersensitivity of XRCC1−/− mouse embryonic fibroblasts to DNA damaging agents and effect of PARP inhibition. XRCC1+/+ (closed symbols) and XRCC1−/− cells (open symbols) were treated with (A) TMZ for 4 h, (B) hmdUrd for 24 h, (C) DMM for 1 h or (D) camptothecin for 24 h. Shown are survival curves for 'control' cells in the absence of PARP inhibitor (circles), or cells treated in the presence of 4-AN (5 μM for 24 h) (squares). Data represent the mean ± SEM of at least 3 independent experiments. Cell sensitivity was determined by growth inhibition assays as described 119.

Table 3 Sensitization by PARP inhibitor (4-AN)

A small 4-AN-mediated potentiation of camptothecin cytotoxicity was observed in XRCC1+/+ cells, but the effect of 4-AN in XRCC1−/− cells was extremely low (Figure 6D, Table 3). Similar results were reported when camptothecin was combined with a PARP inhibitor in wild-type and XRCC1-deficient CHO cells 111. Sensitization to Top1 poisons by PARP inhibitors has been associated with persistence of DNA strand breaks as measured by the comet assay or alkaline elution 111, 112. Despite the relatively moderate sensitization to Top1 inhibitors compared with results obtained with monofunctional alkylating agents, there is interest in clinical evaluation of a topotecan/PARP inhibitor combination 109.

For IR, where there is only a minimal hypersensitivity phenotype in the absence of XRCC1 (Figure 1A), inhibition of PARP activity by 4-AN results in a comparatively low level (2-fold or less) sensitization even in XRCC1+/+ cells (Table 3). It has been reported that PARP inhibitors are able to slow, but not prevent, sealing of IR-induced strand breaks in cells 113. Presumably repair in the absence of PARP activity is sufficient to protect against IR-mediated cytotoxicity. In addition, only minimal PARP inhibition-mediated sensitization to oxidative DNA damaging agents was observed in wild-type mouse fibroblasts (reviewed in Ref. 106).

It appears that extreme 4-AN-mediated sensitization is associated with the requirement for pol β-dependent dRP lyase activity during BER, and is observed for agents that exhibit a hypersensitivity phenotype in both XRCC1 and pol β null cells (Table 2). This requirement for dRP lyase-mediated repair is consistent with the observation that PARP-1 is activated specifically by strand break intermediates of this BER pathway 114.

Characteristics of XRCC1- and pol β-deficient cells: repair deficiency following co-exposure to MMS and 4-AN

PARP-1 is the first protein shown to interact with nicked DNA during BER 115, and failure to automodify results in inactivated PARP-1 remaining bound to damaged DNA. This may hinder access of repair proteins to sites of damage resulting in inhibition of SSBR 115 and accumulation of SSBs, especially in repair-deficient cells. In addition, blocked replication in response to inactivated PARP-1 protein bound to damaged DNA, and the subsequent delayed resolution of stalled replication forks 116, will result in generation of DSBs 117, 118.

It is therefore anticipated that the combination of MMS with a PARP inhibitor would result in an increased level of DNA strand breaks that could be measured by the alkaline comet assay. Cells (XRCC1+/+ and XRCC1−/−, and pol β+/+ and pol β−/−) were treated briefly with 1 mM MMS in ice-cold medium in the presence or absence of 4-AN, and then allowed to repair in warm medium for up to 4 h. When MMS was used as a single agent, no strand breaks were detectable in any of the four cell lines over the time course of the experiment (Figure 7, panels C and D). When MMS was combined with 4-AN, there was measurable damage in all of the cell variants. XRCC1−/− cells are less sensitized to MMS by 4-AN than are wild-type cells (Figure 7A, Table 3), yet retain a low level of hypersensitivity under conditions of PARP inhibition. A slightly higher level of DNA damage was observed in the XRCC1-deficient compared with XRCC1+/+ cells in the presence of the PARP inhibitor (Figure 7C). Fibroblasts deficient in pol β are more highly sensitized to MMS by 4-AN than are wild-type cells (Figure 7D) 105. Levels of DNA damage were increased significantly more in the BER-deficient pol β−/− compared with wild-type cells following exposure to the combination of MMS + 4-AN (Figure 7D). The amplification of MMS-induced DNA damage in the presence of a PARP inhibitor is consistent with the level of PARP inhibition-mediated potentiation of MMS cytotoxicity in the cell lines studied.

Figure 7
figure 7

Enhanced cytotoxicity and increased DNA damage following co-exposure to MMS and 4-AN. (A) XRCC1+/+ (closed symbols) and XRCC1−/− (open symbols) and (B) Pol β+/+ (closed symbols) and pol β−/− (open symbols) mouse embryonic fibroblasts were treated with MMS in ice-cold medium for 20 min. Shown are survival curves for 'control' cells in the absence of PARP inhibitor (circles or triangles), or cells treated in the presence of 4-AN for a total of 24 h (squares or diamonds). Cell sensitivity was determined by growth inhibition assays as described 119. Data are from a representative experiment where the values are the mean of triplicate determinations. The dashed line indicates the concentration of cold MMS combined with 4-AN used in the experiments presented in panels C and D. (C) XRCC1+/+ and XRCC1−/− and (D) Pol β+/+ and pol β−/− cells were treated for 20 min with 1 mM MMS in ice-cold medium in the absence or presence of 4-AN (symbols as described for panels A and B). Plotted is DNA damage (median OTM) during cellular repair in warm medium ± 4-AN for up to 4 h.

Concluding Remarks

Both XRCC1−/− and pol β−/− cells are most hypersensitive to agents that produce DNA lesions repaired by monofunctional glycosylase-initiated BER. These agents include the methylating agents MMS and TMZ, as well as the thymidine analog hmdUrd. Comet assay results support the hypothesis that XRCC1−/− cells, like pol β−/− cells, are deficient in cellular BER of MMS-induced damage. XRCC1 interacts with many proteins known to be involved in BER and SSBR. The PARP-interacting BRCT I domain of XRCC1 is critical in cells for efficient repair and resistance to MMS-induced cytotoxicity. PARP-1 is activated by direct SSBs in DNA, and also by indirect SSBs that arise during BER, and chemical inhibitors of PARP activity are able to enhance cell killing. The hypersensitivity phenotypes, as well as the extreme PARP inhibitor-mediated sensitization, are associated with a requirement for pol β-dependent dRP lyase activity during repair. Even in the absence of XRCC1, PARP activity is a determinant of cell sensitivity to agents that produce DNA damage repaired by pol β-dependent BER. Thus PARP-mediated poly(ADP-ribosyl)ation plays a role in modulation of cytotoxicity beyond recruitment of XRCC1 to sites of DNA damage.