Mismatch Repair

Mismatches can occur in DNA due to the incorrect incorporation by DNA polymerases, damage to the nucleotide precursors in the cellular nucleotide pool, or by damage to DNA. The MMR system that will be discussed here has been extensively studied in bacterial systems, termed the MutHLS system, and until a few years ago, the human or mammalian homologues to the bacterial system were unknown. However, a number of events allowed for the cloning and initial characterization of the human homologues, and currently research into mammalian MMR is a very vigorous area of study. Broadly defined, mismatch repair is the recognition and correction of incorrectly paired nucleotides in DNA resulting in a large fragment of DNA from the mismatched strand being excised and new DNA synthesized.

Initially the genes, mutHLS, were discovered in bacterial cells where their absence was responsible for increasing the rate of spontaneous mutations or a mutator phenotype (Fishel and Wilson, 1997and references within). The major members of the pathway inEscherichia coli are MutS, MutL, and MutH along with DNA polymerase, single-stranded binding proteins, and DNA ligase. Homologues of these enzymes have been identified in mammalian systems. For example, genetic instability of simple repeat (microsatellite) sequences in somatic and hereditary colorectal cancer led to the discovery that the mismatch DNA repair system was defective in hereditary nonpolyposis colon cancer (HNPCC). Currently, there are six homologues of the E. coli mutS gene, termed MSH genes in humans (Fishel and Wilson, 1997) (Fig. 1B). MSH2, MSH3, and MSH6 are found in the nucleus, MSH1 is located in the mitochondria, whereas MSH4 and MSH5 are part of the recombination system. The human MSH2 protein forms a complex with MSH6 or MSH3 (Fishel and Wilson, 1997). When paired with MSH6, the heterodimer is termed MutSα which repairs base to base and insertion/deletion mismatches. MSH2 and MSH3, form a complex called MutSβ that binds to insertion/deletion mispairings. MSH2 can bind to mismatches by itself, but the binding is an order of magnitude lower than when interacting with MSH3 or MSH6 (Fishel and Wilson, 1997). There are at least 16 genes that code for homologues of the E. coli mutL gene, termed MLH or PMS, in humans (Peltomaki, 1997). MSH2 and MSH6 heterodimers have also been shown to bind O⁶-methylguanine and O⁴-methylthymine residues, whereas MSH2 has been shown to bind thymine-thymine dimers (Fishel and Wilson, 1997). These data implicate a relationship between the MMR and direct reversal repair pathways.

Defects in the MMR pathway lead to replication errors particularly in microsatellite regions, in both the coding and noncoding strands, of the genome. These microsatellite regions contain multiple repeating DNA sequences in tandem that can cause polymerases to “slip” and produce mismatched nucleotides that are not repaired in MMR deficient cells. The vast majority of HNPCC cases are due to a defect in one of the known MMR genes: particularly MSH2, MSH6, MLH1, and PMS2 (Liu et al., 1996). This has led to the mutator hypothesis, which proposes that cells that are deficient in MMR accumulate somatic mutations in proto-oncogenes and tumor suppressor genes (Vogelstein and Kinzler, 1993). In other words, defective MMR leads to defective proofing of DNA following replication leading to an accumulation of multiple mutations, a prerequisite for tumorigenesis. Further analysis has shown that the majority of HNPCC cases and sporadic tumors with microsatellite instability are mutated in MLH1 or MSH2, and only a few mutations are seen in MSH3, MSH6, PMS1, and PMS2 (Nicolaides et al., 1994). PMS1 and PMS2 are members of the MLH1 family. In fact, recent reanalysis demonstrates over 90% of the HNPCC cases have mutations in either MSH2 or MLH1 (Fishel and Wilson, 1997).

Other sporadic cancers have also been demonstrated to contain defects in MMR genes including ovarian, endometrial, small and nonsmall cell lung, pancreatic, gastric, cervical, and breast carcinomas (Wooster et al., 1994). Recently, cells with defects in the MMR repair pathway have shown an increase in resistance to DNA damage from chemotherapeutic agents like procarbazine, temozolomide, busulfan, cisplatin, and carboplatin (Fink et al., 1998). Finally, it has been shown in cell lines, and more recently very elegantly in mouse knockout models, that cells deficient in O⁶-methylguanine-DNA methyltransferase (MGMT) and MMR are more resistant to alkylating agents than cells deficient in MGMT, but having a competent MMR system (Kawate et al., 1998). These results indicate that cells that are MMR deficient cannot attempt to repair the mismatch that would occur following the replication of DNA with an O⁶-methylguanine (O⁶-meG) lesion, allowing the cells to complete replication. However, the mutation rate would be elevated in these cells. Cells with a competent MMR system would attempt to repair the O⁶-methylguanine-thymine lesion, but without MGMT present, would become caught in a vicious futile cycle leading to chromosome alterations, such as double strand breaks and cytotoxicity.

Knockout mice models for MSH2 have clearly shown MMRs role in cancer susceptibility (de Wind et al., 1995). The homozygous MSH2-deficient mice are viable and fertile but approximately 30% develop leukemia/lymphomas within the first 5 months of life and die of the disease by 1 year (Reitmair et al., 1995). The majority of mice that live longer than 6 months may also develop gastrointestinal and skin tumors (Reitmair et al., 1996). Various cells from these homozygous MSH2 knockout mice demonstrate a lack of mispair binding and microsatellite instability (Fishel and Wilson, 1997).

The overexpression of some MMR proteins in mammalian cells usually leads to the induction of apoptosis (MSH2 or MLH1) (Zhang et al., 1999), and the loss of MMR proteins appears to lead to cellular resistance to chemotherapeutic agents. However, it would be highly unlikely that treatment paradigms would be employed that decreased the level of the MMR system in cells to make them resistant to chemotherapeutic agents. Such an affect would lead to other mutations and would be unacceptable. The overexpression model leading to cell killing could be exploited for tumor killing if restricted to the tumor cells by some sort of targeting approach. This could be exploited to use lower doses of agents such as cisplatin or carboplatin thereby preventing unwanted peripheral toxicities.

Nucleotide Excision Repair

The NER pathway involves at least 11 proteins or complexes of proteins (comprehensively reviewed in Wood, 1996; Lindahl et al., 1997). Often thought to be coupled with transcription, termed “transcription-coupled repair” (TCR), NER repairs bulky lesions of transcribed genes. However, NER also repairs lesions irrespective of genome location and point in the cell cycle. Large adducts including cyclobutane pyrimidine dimers, polycyclic aromatic hydrocarbons, and cisplatin lesions are thought to be repaired mainly through the NER pathway (Wood, 1996), although recombinational repair is also thought to facilitate cisplatin repair. Finally, NER has also been hypothesized to back up other repair systems and contain some shared components with other DNA repair pathways (Wood, 1996).

In humans, deficits within the NER genes can lead to inherited disorders like xeroderma pigmentosum (XP), Cockayne's syndrome, and trichothiodystrophy. Defects in one of the seven XP complementation genes (XPA-XPG) leads to an increase in skin cancers, primarily after exposure to sunlight or UV irradiation, and neurologic abnormalities (Lindahl et al., 1997). In contrast, Cockayne's syndrome and trichothiodystrophy are NER disorders that show no evidence of increased cancer risk, but these disorders are associated with congenital neurological degeneration and skeletal abnormalities and demonstrate some overlapping symptoms with XP (Lindahl et al., 1997).

Repair initiated by NER can be divided into four steps: recognition/preincision, incision, gap-filling, and structural repair (Fig. 1C) (Wood, 1996; Lindahl et al., 1997). Recognition of damage occurs by a protein complex, XPA and replication protein A (RPA), which appears to have affinity for damaged DNA (He et al., 1995). Following recognition, the XPA-RPA proteins interact with the transcription factor IIH complex [composed of both excision repair cross-complementing 3 and 2 (ERCC3 and ERCC2) or XPB and XPD and other gene products], which is associated with transcription and contains helicase activity. Finally two protein nucleases, ERCC1-XPF and XPG, comprise the remaining subunits of the exinuclease, or major enzyme complex (He et al., 1995).

The next step, incision, occurs when the exinuclease complex cleaves the damaged DNA strand 5′ and 3′ to the adduct and requires an ATP-dependent opening of the double-stranded DNA performed by members of the exinuclease (Wood, 1996; Lindahl et al., 1997). Cleavage occurs when the ERCC1-XPF portion acts on the phosphodiester backbone 5′ to the lesion in an Mg²⁺-dependent manner. The location of the incision is adduct-dependent and occurs between 16 and 25 phosphodiester bonds 5′ to the adduct. The XPG portion incises the DNA 3′ to the lesion and contains three types of nuclease activities: single strand-specific endonuclease, 5′- to 3′-specific endonuclease, and flap endonuclease (FEN) activity (Wood, 1996; Lindahl et al., 1997). The 3′ cleavage is also adduct-specific and takes place two to nine phosphodiester bonds on the 3′ side of the adduct. Regardless of the lesion, the structural characteristics of the exinuclease limit the overall size of the liberated incision product to 26 to 27 nucleotides (Wood, 1996; Lindahl et al., 1997).

The final two steps of gap-filling and structural repair are accomplished by polymerase δ and ε, proliferating cell nuclear antigen (PCNA), replication factor C (RFC), DNA ligase, and other associated proteins (Wood, 1996; Lindahl et al., 1997). RFC and PCNA associate with the DNA, load, and clamp the polymerase onto the DNA. Polymerase δ and ε add the appropriate complementary nucleotides to the free 3′-OH group and proceed to fill the gap. To finish the repair process, DNA ligase seals the nick, restoring the intact nucleotide helix.

Two proteins not discussed above, that appear to have NER associations, are XPC (with HHR23B) and XPE binding factor. Although the actual role of XPC is not currently known, it is theorized to bind to the damaged DNA strand in the exinuclease complex and help to target the nucleases, XPF-ERCC1 and XPG, to the proper incision site while protecting the rest of the DNA in the preincision complex (Sancar, 1996). HHR23B tightly binds and copurifies with XPC and might act to stabilize XPC (Reardon et al., 1996). XPE is not required for excision but might act as an accessory factor, binding damaged DNA (Sancar, 1996).

Currently, the complex nature of the NER pathway precludes its use in gene therapy strategies adding back individual repair genes to offer protection from damaging agents. However, a dominant-negative approach could be undertaken with a catalytically inactive component of the NER system added to tumor cells to enhance cell kill. This approach could be coupled with agents that are known to impact on NER, such as cisplatin and oxidative DNA damaging agents. Further research might demonstrate a severely rate-limiting step of NER for a specific adduct that could be added back to specific cell types. Recently, using cell-targeting techniques, there appears to be the potential of inhibiting XPA, thereby sensitizing tumor cells to cisplatin (Koberle et al., 1999).

Direct Repair: O⁶-Methylguanine-DNA Methyltransferase

DNA damage corrected through direct reversal of the modified base is probably the most efficient mechanism of repair, because it removes the alteration to the base without removing the damaged base. The main component of the direct repair family in mammalian cells are the alkyltransferase proteins, which remove alkyl groups from the O⁶ position of guanine and to a lesser extent from the O⁴ position from thymine. O⁶-meG adducts cause damage by mispairing with thymine during replication leading to G:C → A:T transitions (Kelley and Erickson, 2000M. R.L. C.).

Initially two proteins were described with alkyltransferase activity in bacterial cells, however, only one human alkyltransferase (AGT) gene, also known as O⁶-methylguanine-DNA methyltransferase (MGMT), has been identified (Tano et al., 1990). Human MGMT complements E. coli deficient in alkyltransferase activity when cells are challenged with methylating agents. Furthermore, the mechanism of action of human MGMT appears to be similar to the bacterial proteins. Cysteine 145 of MGMT lies in a conserved region—PCHRV—of the protein and accepts the alkyl groups from the modified nucleotide (Tano et al., 1990). Once bound to the cysteine, the alkyl group permanently inactivates MGMT. Therefore, the removal of alkyl groups by MGMT is a stoichiometric process; one protein removes one lesion (Fig. 2A) (Kelley and Erickson, 2000).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

A, mechanism of action of MGMT with O⁶-alkylguanine. The MGMT protein irreversibly transfers the alkyl group from the O⁶ position of guanine to a cysteine residue within the conserved active site PCHRV of the MGMT protein. Only one alkyl group per protein molecule can be transferred through a stoichiometric reaction. There is no cleavage of the DNA and the guanine is restored to its original unaltered state. B, the DNA base excision repair pathway in mammalian cells using class I or class II AP endonucleases/lyases. Class I enzymes, such as FPG, Ogg1, and NTH1 recognize and remove the damaged base by cutting 3′ of the damaged base/AP site. A 3′-phosphatase or 3′-phosphodiesterase removes the sugar-phosphate backbone leaving a 3′-OH and 5′-phosphate, which are termini for DNA polymerase β. DNA ligase I or DNA ligase III/XRCCI complete the repair. Class II AP endonucleases are involved in the repair of alkylated bases such as alkylated N³-adenine, a lesion formed using alkylating agents, which is cytotoxic to cells. A glycosylase removes the damaged base leaving the sugar-phosphate backbone intact, followed by incision on the 5′-side of the baseless site through a hydrolytic mechanism. The resulting termini are “polished” by a deoxyribose phosphate hydrolase (activity found in DNA β-polymerase) removing the 5′-phosphoribosyl followed by insertion of a new base by DNA β-polymerase and ligation by DNA ligase I or DNA ligase III/XRCCI. C, short- versus long-patch repair in BER. Short-patch repair involves a “regular” AP site produced by an AP endonuclease followed by DNA polymerase β and ligase I or DNA ligase III/XRCC I. Long-patch repair may be more involved in the repair of oxidized or reduced AP sites. Repair is completed by polymerase δ, ε, or β and PCNA/RFC, followed by FEN I, PCNA, and DNA ligase I.

MGMT levels in cancer cells often correlate to their sensitivity to chemotherapeutic agents. Elevated levels of MGMT have been noted in ovarian cancer, rhabdomyosarcoma, melanoma, breast cancer, lung cancer, pancreatic cancer, colon cancer, and brain tumors. The increased quantity of MGMT generally, but not always, correlates with the tumor's decreased response to alkylating agents like temozolomide, cyclophosphamide, 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU), and dacarbazine (Preuss et al., 1996; Yu et al., 1999). Mammalian cell lines often contain little or no MGMT activity, termed methyl excision repair minus (mer− or mex−), even when the primary tumor that the cell lines were derived from contained activity (Harris et al., 1996). Given the relationship of MGMT activity to O⁶meG repair, MGMT activity might be an important prognosticator for a chemotherapeutic regimens' success, particularly with those selective chemotherapeutic agents that cause O⁶-meG adducts and/or cross-links.

Human hematopoietic cells contain a low or absent level of endogenous MGMT (Moritz et al., 1995; Kelley and Erickson, 2000). Because the primary dose-limiting toxicity of the majority of alkylating agents is bone marrow suppression and secondary hematopoietic tumors occasionally result from the initial treatment of cancer, many laboratories are investigating the use of overexpressed human MGMT in hematopoietic progenitor cells to protect the bone marrow compartment from the chemotherapeutic agent's toxicity (Moritz et al., 1995). In mice, MGMT-transduced bone marrow has provided dramatic protection from chloronitrosourea exposure after lethal irradiation (Moritz et al., 1995). Currently, clinical trials in humans are underway by numerous groups using MGMT to protect the bone marrow compartment from antitumor agents.

Base Excision Repair

BER involves the removal of relatively short stretches of DNA, between 1 and 13 nucleotides. The general scheme of the BER pathway entails removal of the damaged base by a glycosylase, cleavage of the phosphodiester backbone by an apyrimidinic/apurinic (AP) endonuclease or glycosylase/lyase, insertion of the complementary nucleotides by a polymerase with the removal of the phosphodeoxyribose group, and ligation of the DNA backbone restoring the native structure and sequence (Lindahl et al., 1997). However, if a glycosylase/lyase is involved in the removal of the damaged base, the resulting 3′-phosphodeoxyribose group must be removed to leave a 3′-OH for polymerase activity. Representative members of each step in the pathway will be discussed (Fig. 2, B and C).

Removal of the incorrect or damaged base by a DNA glycosylase comprises the first step of the BER pathway (Fig. 2B). 3-methyladenine DNA glycosylase (MPG/AAG) repairs not only 3-methyladenine (3-meA), the major cytotoxic lesion resulting from alkylating agents, but also functions to cleave the major product of all alkylated DNA N⁷-methylguanine (Friedberg et al., 1995). Although N⁷-methylguanine has been proposed to be relatively innocuous, data suggests this lesion can rearrange to form both cytotoxic and mutagenic lesions (Friedberg et al., 1995). Following the characterization of two E. coli genes containing 3-meA activity, tag and alkA, the eukaryotic counterpart MPG was isolated containing activity toward multiple adducts (3-meA, 3-meG, 7-meG, and others) similar toalkA (Friedberg et al., 1995). Although one might expect the overexpression of MPG to enhance repair, particularly if it is rate limiting, the overexpression of MPG actually leads to increased sensitivity to alkylating agents and chromosomal aberrations suggesting that this first step in BER is not rate limiting and that the accumulation of unrepaired AP sites is cytotoxic to cells (Coquerelle et al., 1995). Furthermore, MPG knockout mice are viable and are sensitive to some, but not all, alkylating drugs (Wilson and Thompson, 1997).

Another class of glycosylases was demonstrated to contain both glycosylase and AP lyase activity including formamidopyrimidine-DNA glycosylase (Fpg; Fig. 2B) (Friedberg et al., 1995). The Fpg protein repairs 8-oxo purines, FaPy purines, and ring-opened aminoethyl purines (Karakaya et al., 1997) and has the ability to remove 5′-terminal deoxyribose phosphate (dRPase) groups from DNA, which are formed during the cleavage of DNA by AP endonucleases (Kelley and Erickson, 2000 and references within). However, a direct link between the dRPase activity of Fpg and downstream BER steps has not been experimentally confirmed. During replication, if Fpg does not repair the altered guanine, then an adenine can be inserted into the daughter strand. Expression of Fpg in mammalian cells decreases the mutagenic effects of aziridine (ethyleneimine), thiotepa, sulfur mustard, and gamma rays (Gill et al., 1996).

Recently, the yeast and mammalian functional homologues for Fpg were cloned and designated Ogg1, for 8-oxoguanine DNA glycosylase (Rosenquist et al., 1997; Yu et al., 1999). Functional assays on purified Fpg and human Ogg1 proteins demonstrated equal amounts of 7,8-dihydro-8-oxoguanine (8-oxoG) removal for a given concentration; although controversy over the amount of removal of imidazole ring-opened guanines and adenines by Ogg1 compared to Fpg exists in the literature (Roldan-Arjona et al., 1997). Futhermore, only partial suppression of the mutator phenotype occurs after expression of Ogg1 in the fpg⁻ bacterial strain (Roldan-Arjona et al., 1997). Nevertheless, Ogg1 has demonstrated activity against ring-opened purines. More recent data demonstrates alternative spliced transcripts and products for Ogg1. Although as many as seven alternative spliced products exist, only two, Ogg1-type 1a and Ogg1-type 2 are the most predominant (Nishioka et al., 1999). Ogg1-type 2 appears to have a mitochondrial location, whereas Ogg1 is primarily nuclear (Nishioka et al., 1999). More recently, an Ogg2 protein has been isolated that removes 8-oxoG that is incorporated opposite A in DNA from reactive oxygen species-induced 8-oxo-dGTP (Hazra et al., 1998). Other human DNA glycosylases (Table1) exist that will, most likely, be shown to be of similar importance as Ogg1.

Removal of a damaged nucleotide by a glycosylase forms an AP site. AP sites are also formed through ionizing radiation, oxidative agents, and even spontaneous hydrolysis of the N-glycosylic bond (approximately 10,000 bases are lost per cell per day) (Friedberg et al., 1995). Repair of AP sites inhibits both the mutational and cytotoxic effects of nucleic acid loss. As DNA polymerase encounters an AP site, progression is usually halted; however, occasionally DNA polymerase will bypass the AP site often incorporating an A opposite the baseless site potentially leading to mutations. AP endonucleases recognize abasic sites and cleave the phosphodiester DNA backbone (Fig.2C).

Two enzymes with AP activity were characterized in E. coli: endonuclease IV (nfo; endo IV) and exonuclease III (xth; exo III). In E. coli, endo IV acts as the minor bacterial endonuclease cleaving the DNA backbone on the 5′ side of an AP site similar to all AP endonucleases but lacks 3′- to 5′-exonuclease activity found in the exo III family (Friedberg et al., 1995; Yu et al., 1999). Endo IV also contains 3′-diesterase activity and is capable of removing 3′-phosphate and 3′-phosphoglycolate adducts, which occur following single strand breaks and oxidative damage-induced AP sites (Friedberg et al., 1995; Yu et al., 1999). No mammalian endo IV family members have been discovered to date, although endo IV family members have been characterized in both unicellular and simple multicellular eukaryotic cells: APN1 in Saccharomyces cerevisiae, and CeApn1 in Caenorhabditis elegans, respectively, where APN1 acts as the major AP endonuclease activity in the cell (Friedberg et al., 1995; Yu et al., 1999). Häring et al. (1994) demonstrated that 4′ oxidized lesions, a common product of bleomycin damage, are repaired by endo IV with a 4-fold greater efficiency than native AP sites, whereas exo III requires a 400-fold higher concentration for the same repair rate compared with unmodified AP sites. These data suggest endo IV family members play a role in protecting cells from oxidative damage, strand breaks, and specific AP sites

A second group of AP endonucleases and the major AP endonuclease inE. coli and mammalian cells, the E. coli exo III family, has also been characterized. Again the representative member, exonuclease III (xth; exo III), was initially described inE. coli and represents approximately 90% ofE.coli's endonuclease activity (Friedberg et al., 1995; Yu et al., 1999). Similar to endo IV enzymes, the exo III family cleaves on the 5′ side of the AP site leaving a 3′-OH necessary for DNA polymerase catalyzed nucleotide addition. Another activity found in all tested AP endonucleases is the 3′-phosphodiesterase activity allowing removal of blocking groups found after oxidative damage or free radical attack (Johnson and Demple, 1988; Levin et al., 1988). Cells lackingxth have increased sensitivity to methyl methanesulfonate (MMS), near UV wavelength light, and exhibit extreme sensitivity to H₂O₂ (Friedberg et al., 1995). However, bleomycin and gamma rays have little effect on anxth⁻ mutant strain (Friedberg et al., 1995).

Exo III-like activity has been isolated, identified, and characterized from other bacteria, Drosophila, and mammals including humans. The major AP endonuclease in humans (APEX/hAPE/HAP1/APE1) was identified and characterized in 1991 by three different groups (Demple et al., 1991; Friedberg et al., 1995; Kelley and Erickson, 2000). The following year, a protein containing reduction-oxidation (redox) activity, redox effector factor 1 (ref-1), was identified and determined to have the same sequence as APE (Xanthoudakis et al., 1992). So, three distinct activities are found in APE/ref-1: AP endonuclease, 3′-phosphodiesterase, and redox function. Human APE/ref-1 recombinant protein contains similar AP endonuclease activity asexoIII; however, the 3′-phosphodiesterase activity is relatively low compared with other class members and might not be biologically significant (Winters et al., 1994). Complementation ofE. coli deficient in AP endonuclease activity by APE provides significant protection from MMS and little resistance to H₂O₂ providing further evidence that 3′ blocking group removal activity is limited in APE (Robson and Hickson, 1991).

Due to the lack of an APE-deficient cell line or animal model [homozygous knockout ape mice are embryonic lethal at day 5.5, whereas heterozygous mice develop without any apparent abnormalities (Xanthoudakis et al., 1996)], antisense technology has been applied to determine the effects of lowering APE levels in cells and their response to various damaging agents. Antisense APE containing HeLa S3 cells were more sensitive to the cytotoxic effects of MMS, H₂O₂, menadione, bleomycin, paraquat, hyperoxia (100%), and hypoxia (1%). However, cells depleted in APE did not demonstrate a decrease in survival after challenge with UV irradiation (Walker et al., 1994). Conversely, when APE levels were induced by exposing cells to sublethal doses of ionizing radiation and HOCl, cellular survival to reactive oxygen species and MMS was enhanced (Ramana et al., 1998). Finally, overexpression of APE in cells as a transgene has resulted in increased transcript and protein levels with increased activation of bioreductive drugs leading to enhanced cell kill. This activation of the drug appears to be due to the redox function of APE (Prieto-Alamo and Laval, 1999). These data suggest that the level of APE in the cells has an effect on survival and response to damaging agents and might be a target or strategy for gene therapy protocols.

After cleavage of the phosphodiester backbone by APE, two proposed BER subpathways exist for completion of repair in mammalian systems (Fig.2C). One subpathway, short-patch repair, consists of DNA polymerase β, which removes the deoxyribose phosphate and adds a nucleotide followed by DNA ligase I or DNA ligase III/XRCC1 sealing the nick (Srivastava et al., 1998). XRCC1 can associate with polymerase β and DNA ligase III enabling DNA ligase III to replace DNA ligase I's function. The second subpathway, called long-patch repair, is a proposed minor pathway and involves the displacement of 2 to 13 bases surrounding the AP site (Fig. 2C). Resynthesis of the corresponding nucleotides is accomplished by various polymerases whose function is dependent on PCNA, RFC, and possibly other undetermined factors (Klungland and Lindahl, 1997). After nucleotide addition, the FEN 1 enzyme acts by removing the 5′-dRP groups through the excision of the displaced 5′-flap structure containing the 5′-dRP moieties. Complete repair occurs when DNA ligase I and possibly DNA ligase III/XRCC I, in patches between two and six nucleotides, restores the phosphodiester backbone. In vitro reconstitution of both the short- and long-patch BER repair pathways supplies further evidence for the multiple mechanisms of BER (Klungland and Lindahl, 1997). The selection of which pathway seems to be dependent on the state of the AP site: normal AP sites are repaired through the short-patch BER pathway and modified sites (oxidized, reduced, etc.) are corrected by the long-patch BER pathway (Fig. 2) (Klungland and Lindahl, 1997).

Polymerase β and DNA ligase I knockout mice are both embryonic lethal, whereas polymerase β knockout mouse cells are able to propagate (Wilson and Thompson, 1997 and references within). Cells with no functional polymerase β were able to perform both short- and long-patch repair. However, repair kinetics of the short-patch repair were much slower in the null cells than polymerase β-containing extracts, whereas no difference was noted in the kinetics of the long-patch BER pathway in extracts with or without polymerase β (Wilson and Thompson, 1997). These data suggest a greater dependence of the short-patch to polymerase β than the long patch. Cells deficient in polymerase β showed a hypersensitivity to monofunctional alkylating agents, like MMS, whereas no phenotype was noted for cells treated with H₂O₂ or ionizing radiation, suggesting that oxidized AP sites might undergo repair using the long-patch BER pathway with alternative polymerases (δ or ε) (Wilson and Thompson, 1997; Yu et al., 1999; Kelley and Erickson, 2000). Other investigators have observed that polymerase β or δ is capable of repairing oxidized or reduced AP sites in a PCNA-dependent manner through long-patch BER repair, whereas only regular AP sites could be repaired by short-patch BER (Fortini et al., 1998). Therefore, the use of different polymerases and cofactors in the BER pathway seems to be dependent on the type of AP site created by the damaging agent.

The use of BER genes (cDNAs) in gene therapy, similar in approach to what is being done with MGMT, is currently being undertaken to protect normal cells, such as hematopoietic cells, against alkylating agents. This has recently been reviewed (Limp-Foster and Kelley, 2000). Alternatively, use of selected BER enzymes, such as MPG/AAG, when overexpressed could imbalance the BER pathway leading to cell killing (Coquerelle et al., 1995). Alternatively, altered BER enzymes that are catalytically inactive and dominant-negative could be used in a therapeutic gene therapy setting to sensitize cancer cells to ionizing radiation or chemotherapy. This has also been extensively discussed in a recent review (Limp-Foster and Kelley, 2000).

Conclusions

The various DNA repair pathways in cells offer protection from an extremely large variety of endogenous and exogenous agents. By gaining a greater understanding of the mechanism of action, regulation, and repair profiles of the DNA repair proteins, specific strategies may be designed for protection of normal cells or, alternatively, in some cases sensitization of cancer cells to chemotherapeutic agents. The designing of clinical cancer protocols would be further enhanced by a greater understanding of the specific actions of the chemotherapeutic agents and their various damage profiles within cells, both normal and cancer. In combination, this data would allow for combination of various DNA repair proteins with themselves or other detoxifying cellular proteins to aid in the treatment of cancer.