Review

Base excision repair (BER) pathway executed by a complex network of proteins is the major system responsible for the removal of damaged DNA bases and repair of DNA single strand breaks (SSBs) generated by environmental agents, such as certain cancer therapies, or arising spontaneously during cellular metabolism. Both modified DNA bases and SSBs with ends other than 3'-OH and 5'-P are repaired either by replacement of a single or of more nucleotides in the processes called short-patch BER (SP-BER) or long-patch BER (LP-BER), respectively. In contrast to Escherichia coli cells, in human ones, the two BER sub-pathways are operated by different sets of proteins. In this review the selection between SP- and LP-BER and mutations in BER and end-processors genes and their contribution to bacterial mutagenesis and human diseases are considered.


INtRODuctION
Living organisms are continuously exposed to damaging agents both from the environment and from endogenous metabolic processes, whose action results in modification of proteins, lipids, carbohydrates and nucleic acids.Events that lead to DNA modification include radiation, hydrolysis, exposure to reactive oxygen or nitrogen species and other reactive agents, like alkylating agents and lipid peroxidation products (Lindahl, 1993;Tudek et al., 2006;Olinski et al., 2007;Tudek, 2007).To counteract these threats organisms are equipped with multiple damage prevention and repair systems to ensure the stability of DNA and to protect the genome from potential mutagenic modification and allow accurate transmission of genetic information.
The knowledge of DNA repair processes is critical to our understanding of how and why the genome is affected during the lifespan of the organism, and how the DNA, RNA and nucleotide repair systems efficiently work via several different pathways, such as: (1) sanitization of the nucleotide pool (for more details see accompanying review by

2007
J. Krwawicz and others Arczewska & Kusmierek, this issue), (2) direct reversion of base modifications by (i) demethylation processes (for more details see accompanying review by Nieminuszczy & Grzesiuk, this issue) and (ii) by 6-4 photolyase or CPD photolyase (Kim et al., 1994;Sancar, 1994), or (3) excision of (i) misincorporated bases in the newly replicated DNA strand by mismatch repair (MMR) (for more details see accompanying review by Arczewska & Kusmierek, this issue), (ii) excision of bulky damage from both DNA strands or from the transcribed strand by nucleotide excision repair (NER) (for more details see accompanying review by Maddukuri et al., this issue), and (iii) excision of oxidized, methylated or misincorporated bases from DNA by base excision repair (BER) which is described in more details in this review.Damaged bases are also a source of the single strand breaks (SSBs) and double strands breaks (DSBs).SSBs repair (SSBR) is also discussed in this review.The strand breaks are subject to recombinational repair (for more details see accompanying review by Nowosielska, this issue).Despite the protection provided by these mechanisms some of the damage escapes repair.Unrepaired DNA damage may block replication and engage alternative DNA polymerases in the process of so-called translesion synthesis (TLS) to by-pass the lesion in an error-free or error-prone fashion (reviewed by: Bebenek & Kunkel, 2004;Shcherbakova & Fijalkowska, 2006).To sum up, the unrepaired DNA damage leads to replication and transcription errors and in consequence to mutagenesis, ageing and various diseases, including carcinogenesis and neurodegeneration (reviewed by: Krokan et al., 2004;Bartsch & Nair, 2006).

PROcESSES LEADINg tO DNA BASE MODIfIcAtIONS
DNA base modifications are formed by both exogenous (i.e.environmental) and endogenous factors.Endogenous DNA damage occurs at a high frequency compared with exogenous damage and the types of damage produced by normal cellular processes are identical or very similar to those caused by some environmental agents (Jackson & Loeb, 2001).It has been proposed that the DNA damage from endogenous sources gives rise to 20 000 lesions per mammalian cell per day, most of the lesions being deaminations, spontaneous hydrolysis of the N-glycosidic bond, alkylations, and damage by reactive oxygen or nitrogen species and lipid peroxidation products (Lindahl, 1993;Drablos et al., 2004;Tudek et al., 2006;Olinski et al., 2007).Lesions are also caused by errors in DNA metabolism, including the formation of SSBs and DSBs from the collapse of replication forks and the introduction of modified nucleic acid bases during DNA replication.Examples of base modifications discussed below and repaired by BER are shown in Fig. 1.

Deamination of DNA bases
DNA bases containing an exocyclic amino group, namely adenine (A), guanine (G), cytosine (C), and 5-methylcytosine (5-meC) are susceptible to spontaneous hydrolytic deamination to hypoxanthine (Hyx), xanthine (X), uracil (U), and thymine (T), respectively.Deamination occurs more frequently in single-stranded than in double-stranded DNA, where the amino groups are protected by participating in hydrogen bonds (Lindahl, 1993).Spontaneous deamination is rather slow, but it can be significantly accelerated in vivo by nitrogen dioxide (NO 2 ) and dinitrogen trioxide (N 2 O 3 ) formed during inflammation or by UV-as well as by γ-irradiation (reviewed by: Kavli et al., 2007).

Loss of DNA bases via N-glycosidic bond hydrolysis
The N-glycosidic bond between base and deoxyribose in DNA can be hydrolyzed spontaneously or by DNA N-glycosylases during removal of damaged or incorrect bases from DNA by BER.This process leads to formation of an apurinic/apyrimidinic (AP) site (AP-site).Additionally, reactive oxygen species (ROS) and alkylating agents promote the release of bases, often by introducing lesions that destabilize the N-glycosidic bond (Lindahl, 1993;Guillet & Boiteux, 2003).AP-sites are among the most frequent endogenous lesions found in DNA and about 10 000 lesions are formed per human cell per day (Lindahl, 1993).Purines are lost at a rate 500 times higher than pyrimidines, and the depurination rates of A and G are comparable (Loeb & Preston, 1986).AP-sites are highly damaging lesions, can block replication and are both cytotoxic and mutagenic (Loeb & Preston, 1986;Guillet & Boiteux, 2003).Additionally, unrepaired AP-sites may rearrange to generate SSBs (Lindahl, 1993).

Alkylation of DNA bases
Alkylating agents can react with 12 different positions of DNA bases, including all exocyclic oxygens and most of ring nitrogens, and can also modify oxygen atoms in the phosphates groups of the sugar-phosphate backbone.Depending on the mode of action, alkylating agents are divided into two types: S N 1-type agents (e.g.N-methyl-N-nitrosourea, MNU) alkylate both oxygens and nitrogens in nucleic acids, and S N 2-type agents (e.g.methyl methanesulfonate, MMS) alkylate mainly nitrogens.
Mutations in genes involved in BER and DNA-end processors In general, ring nitrogen atoms engaged in hydrogen bonding are almost non-reactive in double-stranded DNA (dsDNA), but can be more readily alkylated in single stranded DNA (ssDNA) or RNA (e.g.N3 of cytosine and N1 of adenine) (reviewed by: Drablos et al., 2004;Sedgwick, 2004;Nieminuszczy & Grzesiuk, this issue).The major product of DNA base methylation is N7-methylguanine, a rather non-muta-  genic and non replication-blocking lesion.However, destabilization of the N-glycosidic bond due to the N7-substitution of guanine results in the formation of AP-sites or imidazole ring opening to yield very mutagenic lesion 7me-FapyG (Tudek et al., 1992).The second most common DNA base methylation is N3-methyladenine (3-meA), which is a potent replication-blocking lesion and is perhaps the most toxic adduct produced by alkylating agents, resulting in TP53 induction, S-phase arrest, chromosomal aberrations and apoptosis (Engelward et al., 1998).In contrast to the limited miscoding potential of N-purines, O 6 -methylguanine (O 6 -meG) and, to a lesser extent, O 4 -methylthymine (O 4 -meT) are major contributors to mutagenicity induced by alkylating agents.Endogenous agents may alkylate DNA bases, and among them the best known is S-adenosyl-l-methionine (SAM) (Rydberg & Lindahl, 1982).Physiologically, SAM is a methyl donor in many biochemical reactions, and among others it participates in enzymatic methylation of DNA cytosines at C5 position, which, in Eukaryotes, regulates gene expression.Furthermore, adenine methylation in the GATC sequences is used by the MMR system to distinguish between the newly synthesized and template DNA strands in E. coli cells (reviewed by Arczewska & Kusmierek, this issue).

Oxidation-induced DNA damage
Reactive oxygen species (ROS), together with reactive nitrogen species (RNS) are known to induce both deleterious and beneficial effects.They can be induced by exogenous or environmental factors such as UV light, X-rays or γ-rays (which produce hydroxyl radical ( • OH) by radiolysis of water), xenobiotics, cigarette smoke are present as pollutants in the atmosphere.Endogenously, they are formed as by-products of the respiratory electron transport chain, cytochrome P450 and xanthine oxidase metabolism, by micorsomes and peroxisomes, and are also produced by neutrophils, eosinophils and macrophages during inflammation and in various metalcatalyzed reactions (reviewed by : Valko et al., 2006).Aerobically growing cells depend on energy formed by reduction of atmospheric oxygen (O 2 ) to H 2 O by the respiratory electron transport chain (Babcock & Wikstrom, 1992).The main product of mitochondrial respiration is superoxide anion radical (O 2 •− ), which shows limited reactivity, but upon escape from the respiratory electron transport chain induces side effects by further conversion to H 2 O 2 by superoxide dismutase (SOD), and then to hydroxyl radical ( • OH).O 2 •− induces • OH formation in the Haber-Weiss reaction (O 2 •− + H 2 O 2 → O 2 + • OH + OH − ) and also liberates Fe 3+ from ferritin and reduces it to a Fenton reaction constituent, Fe 2+ , or liberates Fe 2+ from iron-sulfur cluster-containing enzymes (Kruszewski & Iwanenko, 2003).H 2 O 2 can be reduced to water by catalase and glutathione peroxidase.However, in the presence of transition metal ions, such as iron or copper, H 2 O 2 is reduced to • OH by the Fenton-type reaction (H 2 O 2 + Fe 2+ → • OH + OH − + Fe 3+ ).The reactivity of • OH is so high that it can diffuse no further than one or two molecular diameters before reacting with a cellular component, so it must be generated close to the DNA molecule to be able to oxidize it (Michiels et al., 1994).
Hydroxyl radicals and peroxynitrite can cause damage to DNA both by direct attack on the bases or sugar moieties or indirect, via cell membranes lipids peroxidation (LPO).LPO products interact with Mutations in genes involved in BER and DNA-end processors DNA resulting in generation of adducts to bases, abasic sites, single or double strand breaks and subsequently chromosomal alterations.Polyunsaturated fatty acids (PUFAs) are constituents of phospholipid membranes, with the most abundant linoleic and arachidonic acids.Attack of ROS and RNS on polyunsaturated fatty acids causes formation of radicals and breaking of double bonds, which leads to lipid molecules fragmentation with generation of aldehydes, epoxides and other reactive keto-compounds, such as malondialdehyde, acrolein, crotonaldehyde, trans-4-hydroxy-2-nonenal, and 2,3-epoxy-4-hydoxynonenal.These lipid peroxidation compounds react with exocyclic nitrogen atoms of DNA bases and form exocyclic DNA adducts, in this indirect pathway contributing to oxidative DNA damage (Burcham, 1998).

DNA damage induced by other exogenous and endogenous factors
Apart from the above-described agents also numerous other endogenous and exogenous factors contribute to DNA damage and human diseases.These endogenous factors which can induce DNA modification include: (i) base propenols, formed by oxidative DNA damage; (ii) estrogens, which can induce DNA damage directly and indirectly, through redox-cycling processes that generate reactive radical species; (iii) reactive carbonyl species (RCS) (e.g.glyoxal and methylglyoxal), originating from lipid peroxidation and glycation; (iv) chlorinating agents; (v) heme precursors; (vi) and also amino acids (reviewed by : Burcham, 1999).Another important environmental source of DNA damage is UV light, which induces formation mainly of cyclobutane pyrimidine dimers (CPDs) and 6-4 photoproducts ([6-4]PPs), which have proved to be involved in skin carcinogenesis (reviewed by : Pfeifer et al., 2005).

DNA damage induced by anti-cancer treatments
Agents that induce DNA damage in cells and inhibit DNA repair have successfully been used for decades to treat patients with tumors (Bentle et al., 2006).The DNA affected by anti-cancer treatments is detected by DNA damage sensors, which leads to the activation of TP53.Activation of TP53 can lead to the death of cancer cells (Roos & Kaina, 2006).The efficacy of genotoxins used in humans as anticancer agents is, however, limited by their toxicity to normal tissues.Specific sensitization of tumor cells to the action of anti-cancer treatments would reduce the efficacious doses of genotoxins to be used in patients, diminishing the detrimental sideeffects of the drugs on normal tissues.Some drugs, namely bleomycin and neocarzinostatin act by in-duction of DNA strand breaks (Dedon & Goldberg, 1992;Dedon et al., 1992).DNA cleavage by bleomycin depends on oxygen and metal ions.It is believed that bleomycin chelates metal ions (primarily iron) producing a pseudoenzyme that reacts with oxygen to produce superoxide and hydroxide free radicals that cleave DNA.In addition, these complexes also mediate lipid peroxidation and oxidation of other cellular molecules.The drug is used in the treatment of Hodgkin lymphoma, squamous cell carcinomas, and testicular cancer, pleurodesis as well as plantar warts (Katsara et al., 2006;Kopp et al., 2006;Proctor & Wilkinson, 2006).Temozolomide is an alkylating agent which forms O 6 -meG, 7-meG, 3-meA base lesions in DNA.The drug is used for the treatment of refractory anaplastic astrocytoma, a type of cancerous brain tumor (Rabbani et al., 2007).

BER AS tHE MOSt fREquENtLy uSED DNA REPAIR
The base excision repair pathway is responsible for removal of more than 10 000 DNA lesions daily in each human cell.In addition, lesions targeted by the BER pathway are relatively small, causing little helix distortion.Many of these lesions have been shown not to inhibit elongation by some DNA and RNA polymerases both in vivo and in vitro (Doetsch, 2002).However, BER is the major repair pathway involved in the removal of DNA damage involving structurally non-distorting and non-bulky lesions, e.g.oxidized or ring-saturated bases, alkylated and deaminated bases, as well as apurinic/apyrimidinic sites, and also some type of mismatches (Lindahl et al., 1997).Proteins engaged in BER are conserved from bacteria to eukaryotes.BER is initiated (i) by a damage-specific DNA N-glycosylase that recognizes and removes the modified or mismatched base by hydrolysis of the N-glycosidic bond between a 2'-deoxyribose and the base, or (ii) by non-enzymatic hydrolytic depurination leading to base loss (described above), as well as (iii) by SSBs with ends other than 3'-OH and 5'-P.

DNA damage recognition by DNA N-glycosylases
At least 12 genes (plus their splicing variants) and eight ones encoding various glycosylases have been found in mammalian and Escherichia coli cells, respectively, with different substrate specificities and modes of action (summarized in Table 1).Glycosylases effectively ensure repair of the majority of endo-and exogenous DNA base lesions.They often contain a conserved motif of helix-harpin-helix (HhH) in the active site, which enables them to bind DNA.These HhH motifs bind a metal ion but only UDG superfamily-1 is based on structural similarity to uracil DNA glycosylase UDG.Enzymes belonging to this family are active against uracil in ssDNA and dsDNA, and recognize uracil explicitly in an extrahelical conformation via a combination of protein and bound-water interactions.Some of these enzymes are mismatch specific and explicitly recognize the widowed guanine on the complementary strand rather than the extrahelical scissile pyrimidine.AAG superfamily-2 is based on structural similarity to human alkyladenine DNA glycosylase AAG.Members of the UDG and AAG superfamilies are compact single-domain enzymes with relatively small DNA-interaction surface.MutM/Fpg superfamily-3 is based on structural similarity to bacterial 8-oxoguanine DNA glycosylase Fpg.All known members have the unique feature of using their N-terminal proline residue as the key catalytic nucleophile.HhH-GPG superfamily-4 is named for the characteristic active site borne by family members comprising a helix-harpinhelix followed by a Gly/Pro-rich loop and catalytic aspartate residue (reviewed by : Pearl, 2000;Fromme et al., 2004).

Role of end-processors in DNA repair
The AP-sites or DNA ends generated after lesion excision or excision and incision by mono-or bifunctional glycosylases, respectively, are not suitable for the next repair steps, and as the repair intermediates are very mutagenic when unrepaired by BER (Simonelli et al., 2005).DNA ends containing modified 3' and/or 5' ends may also arise as a result of direct chemical modification during SSB formation through the action of ROS (Demple & DeMott, 2002).However, ionizing radiation is a major contributor to the formation of damaged 3' ends, and anti-tumor drugs, such as bleomycin and neocarzinostatin, can also generate SSBs containing 3' PUA and 3'-P, respectively (Dedon & Goldberg, 1992).Moreover, blocked 3' ends in human cells may arise as a result of abortive DNA topoisomerase I (TOP1) activity (reviewed in (Leppard & Champoux, 2005).
At least a few enzymes occur in E. coli and human cells that can restore conventional 3'-OH and 5'-P moieties to allow gap filling and DNA ligation (summarized in Table 2).DNA-end processing is probably the most diverse enzymatic step due to the variety of termini that can arise.
Additionally, 3'-P can be removed by mammalian polynucleotide kinase (PNK), which, together with NEIL1/NEIL2 glycosylases, forms the APEindependent BER subpathway (Wiederhold et al., 2004).Human PNK is the major DNA 5'-kinase and 3'-phosphatase that is able to phosphorylate the 5' end of SSBs and removes blocking phosphate lesions from the 3' end (reviewed by : Dianov & Parsons, 2007).Moreover, also E. coli nucleoside diphosphate kinase (NDK) has been shown to have AP-lyase, 3'phosphodiesterase, 3'-phosphatase and 3'→5' exonuclease activities (Postel & Abramczyk, 2003;Goswami et al., 2006) but its role in BER is still unclear.An E. coli strain lacking NDK shows elevated levels of MMR-dependent base substitutions and frameshifts, induced probably by an altered dNTP pool.All these NDK activities provide a good explanation of the mutator phenotypes induced by NDK depletion (Postel & Abramczyk, 2003;Goswami et al., 2006).Furthermore, human NM23/NDP kinase has been identified as tumor suppressor, and is associated with tumor metastasis.Its reduced expression seems to be related to an increased metastatic potential in most cancer cell types.Moreover, NM23/NDK kinase was also shown to activate transcription and to have a nuclease activity (Postel et al., 2000;2002).Although, tyrosyl-DNA phosphodiesterase 1 (TDP1) is not involved in the major BER pathway, it may be a DNA-end processor during SSBs repair.The primary substrates for TDP1 are the TOP1-linked 3' ends (TOP1-SSBs) that arise through abortive topoisomerase I (TOP1) activity, which can be induced by the drug camptothecin or by nearby unusual DNA secondary structures or other types of DNA lesions (Plo et al., 2003;Caldecott, 2007;El-Khamisy et al., 2007).TDP1 converts the 3'-TOP1-SSB peptide complex into a 3'-P end further processed by PNK (Connelly & Leach, 2004).TDP1 is mutated in the neurodegenerative disease spinocerebellar ataxia with axonal neuropathy-1 (SCAN1).In addition to being defective in the removal of stalled 3'-TOP1-SSB intermediates, SCAN1 cells also exhibit a reduced capacity to excise the 3'-phosphoglycolate end, a common oxidative damage (Straussberg et al., 2005;Zhou et al., 2005).
Aprataxin (APTX), the protein defective in the neurological disorder ataxia oculomotor apraxia (AOA1), is a member of the HIT domain superfamily of nucleotide hydrolases/transferases.Cells deficient in APTX are defective in SSBR.APTX has been found to be involved in resolution of abortive DNA ligation intermediates by catalysing the nucleophilic release of adenylate groups covalently linked to 5'P-ends at single-strand nicks and gaps (Ahel et al., 2006).APTX is also responsible for the repair of typically endogenous damage produced by reactive oxygen species on 3' DNA ends, including 3'-PUA and 3'-P (Takahashi et al., 2007).APTX acts preferentially on adenylated nicks and DSBs rather than on SSBs.Moreover, APTX has been found to act in BER, specifically in the removal of adenylates that arise from abortive ligation reactions that take place at incised AP-sites in DNA, and may have a general proofreading function in DNA repair, removing DNA adenylates as they arise during SSBR, DSBR, and in BER (Rass et al., 2007).
Recently, it has been found that the high-mobility group box 1 protein (HMGB1) specifically interacts with a BER intermediate.HMGB1 possesses weak dRP-lyase activity and stimulates AP-endonu-clease and FEN1 activities on BER substrates (Prasad et al., 2007).

SP-BER
Oxidized and ring-saturated bases are recognized and removed from DNA by the bifunctional DNA N-glycosylases/AP-lyases.Next, APEs remove 3'-PUA and 3'-P (for more details see above) leaving 3'-OH and 5'-P ends suitable for filling by mammalian Pol β or E. coli Pol I and for end-sealing by mammalian LIG3α or bacterial ligase I.In contrast, alkylated and deaminated bases as well as some types of mismatches, are recognized and removed from DNA by the monofunctional DNA N-glycosylases.In this process the N-glycosydic bond connecting the aberrant base to the sugar-phosphate backbone is cleaved and an AP-site is created (Krokan et al., 1997).The AP-site is recognized and processed by the APEs (Taylor & Weiss, 1982) that hydrolyze the phosphodiester DNA backbone at the 5' side of the AP-site, leaving 3'-OH and 5'-dRP ends flanking the gap.From this point, the choice of the pathway depends on the ability of the enzymes to remove the 5'-sugar phosphate.In mammalian cells both pathways are initiated by Pol β, which inserts one nucleotide into the repair gap.In SP-BER Pol β also removes 5'-dRP by its 5'-dRPase activity, and finally DNA LIG3α-XRCC1 complex seals the ends.Additionally, Pol λ may partially backup Pol β, since it has a 5'-dRPase activity.XRCC1 is a platform protein and was shown to interact with Pol β, LIG3α, PNK, APE1 and PARP-1 (Fig. 3).The lesions removed by bifunctional DNA glycosylases are processed mainly by SP-BER, since the 3'-OH and 5'-phosphate ends may be readily filled in by Pol β.In E. coli, 5'-dRP is removed by Fpg, Nei, or RecJ or by the 5'→3' exonuclease activity of Pol I, and the resulting gap is filled in by Pol I and sealed by DNA ligase I.

LP-BER
In human cells modification of the 5'-dRP moiety by oxidation or reduction prevents its excision by Pol β and the lesion is further processed by LP-BER.First, Pol β falls off and PCNA (replication sliding clamp) is recruited together with Pol δ or Pol ε.The polymerase adds a few nucleotides to figure 2. Model for the BER and SSBR subpathways.P, phosphate; OH, hydroxyl group; 3'PUA, 3'-unsaturated aldehyde; 5'dRp, 5'-deoxyribose phosphate; AMP, adenylate group; TOP1, topoisomerase I-linked 3'-end.The types of DNA lesions repaired by common subpathways of single strand breaks repair and base excision repair are marked in purpure.Escherichia coli enzymes are on left, and are in blue, human enzymes shown on right, are in red.
the 3'-OH end and generates a flap containing the 5'-dRP end, which is then removed by FEN-1 and finally the ends are sealed by DNA ligase I (LIG1).PCNA interacts not only with the polymerase, but also with FEN-1 and LIG1 (Fig. 3).Furthermore, replication protein A (RPA), which interacts with the MUTYH and UNG2 glycosylases, is required by Pol δ and Pol ε for DNA synthesis, and may stimulate LP-BER.Pol δ requires also replication factor C (RF-C; which loads the PCNA sliding clamp on the double helix) and PCNA for efficient synthesis, while Pol ε is highly processive in the absence of PCNA (reviewed by: Krokan et al., 1997;Nilsen & Krokan, 2001;Dianov et al., 2003;Slupphaug et al., 2003;Sung & Demple, 2006).If in E. coli cells the 5'-dRp residue is not removed prior to repair synthesis, Pol I displaces the dRP-containing strand via a strand displacement reaction (Mosbaugh & Linn, 1982) during filling of the gap.The displaced strand is cleaved by the 5'→3' exonuclease activity of Pol I (Xu et al., 1997;Xu et al., 2001), and 2 to 8 or even more nucleotides are removed and replaced, leading to so-called long-patch BER (LP-BER) (Radicella et al., 1993;Sung & Mosbaugh, 2003).Furthermore, the length of repair synthesis may also be determined by the availability of DNA ligase I, its lack leading to longer repair patches in vitro (Sung & Mosbaugh, 2003).
Mutations in genes involved in BER and DNA-end processors increase hOGG1 turnover on damaged DNA and stimulate its excision activity (Hill et al., 2001).BER is further complicated by other proteins which interact with its components (Fig. 3).PARP-1 binds to SSB immediately after its formation and dissociates after self-poly(ADP-ribosyl)ation.PARP-1 has been proposed to prevent cleavage of the strand break ends by nucleases, and was also shown to stimulate LP-BER strand displacement synthesis by Pol β (Dianov et al., 2003;Parsons et al., 2005).Furthermore, Werner syndrome protein (WRN) stimulates Pol β strand displacement DNA synthesis via its helicase activity, and provides proofreading of 3'-mismatches via its 3'→5' exonuclease activity (which is absent in Pol β) (Harrigan et al., 2006).Additionally, Cockayne syndrome group B (CSB) functions in the catalysis of 8-oxoG excision by BER and in the maintenance of efficient hOGG1 expression (Tuo et al., 2002).

MutAgENESIS IN E. coli cELLS cAuSED By DySfuNctION Of BER PROtEINS
DNA guanine is frequently oxidized to 8-oxoG, which, if unrepaired, can be bypassed by DNA polymerases and pair with its cognate C as well as noncognate A, leading to GC→TA transversions.E. coli has evolved a complicated strategy to avoid mutations from this commonly oxidized base.Fpg (also called MutM), removes 8-oxoG paired with C in DNA, while the MutY protein removes A mispaired with 8-oxoG.Finally, the MutT protein, an 8-oxodGTPase, removes oxidized dGTPs from the nucleotide pool, preventing their misincorporation opposite adenine (reviewed by Arczewska & Kusmierek, this issue).E. coli mutants defective in Fpg or MutY, and double mutants lacking both proteins, exhibit higher than wild type spontaneous mutation frequencies (Au et al., 1988;Cabrera et al., 1988;Boiteux & Huisman, 1989;Michaels et al., 1991;Fowler et al., 2003;Speina et al., 2005b;Hamm et al., 2007).Mutants lacking the MutT protein also exhibit high spontaneous mutation frequencies (Akiyama et al., 1989).Oxidized pyrimidines are repaired in E. coli by Nth (endo III) and Nei (endo VIII).The nth mutants exhibit a small mutator phenotype, while nei mutants exhibit no mutator phenotype.Double mutants, lacking both proteins, exhibit spontaneous mutation frequency higher than the wild type (Saito et al., 1997).Surprisingly, triple mutants lacking Fpg, MutY, and Nei and quadruple mutants lacking all four DNA glycosylases, Fpg, MutY, Nei, and Nth, exhibit significant synergistic effects, suggesting an overlap in the substrate specificities of the "pyrimidine-specific" and "purine-specific" enzymes (Blaisdell et al., 1999).Moreover, the nth nei double mutants are hypersensitive to ionizing radiation and hydrogen peroxide but not as sensitive as APEs mutants xth nfo.Single nth mutants exhibit wild-type sensitivity to X rays, while nei mutants are consistently slightly more sensitive than the wild type.Additionally, ung cells are not able to initiate base excision repair of uracil-containing DNA.These mutants have a high GC→AT mutation rate because they are not able to repair deaminated cytosine residues.Uracil residues also accumulate in the DNA of ung mutants as a consequence of the occasional biosynthetic incorporation of uracil into DNA in place of thymine (Duncan & Weiss, 1982).The mug mutant does not show a mutator phenotype in dividing E. coli, and is only a modest mutator in stationary phase cells (Jurado et al., 2004).It is possible that this lack of a strong phenotype is caused by the presence of alternative enzymes in E. coli that process the promutagenic lesion U and the T : G mispair.However, E. coli mug mutant is sensitive for agents causing etheno-adducts (Maciejewska, unpublished;Borys-Brzywczy et al., 2005).Moreover, E. coli possesses two different DNA repair glycosylases, Tag and AlkA, which have similar ability to remove the alkylation product 3-meA from dsDNA.These enzymes have quite different activities for the excision of 3-meA from ssDNA, AlkA being 10-20 times more efficient than Tag.AlkA may have an important role in the excision of base damage from single-stranded regions transiently formed in DNA during transcription and replication (Bjelland & Seeberg, 1996).All bacterial DNA N-glycosylases are summarized in Table 1.The double xth nfo E. coli strain devoid of BER is very sensitive to H 2 O 2 and MMS.Moreover, the triple mutant for the DNA repair genes xth nth nfo, chronically induces the SOS response (Janion et al., 2003).Bacterial nucleases are summarized in Table 2.

PROcESSES cONtROLLED By DNA gLycOSyLASES IN HuMAN cELLS
The sources of uracil in DNA are spontaneous or enzymatic deamination of cytosine (U:G mispairs) and incorporation of dUTP (U : A pairs), inducing CG→TA transitions during DNA replication (Duncan & Weiss, 1982).Uracil is usually an inappropriate base in DNA, but it is also a normal intermediate during somatic hypermutation (SHM) and class switch recombination (CSR) in adaptive immunity.In B-cells cytosine is actively deaminated to uracil by activation-induced cytosine deaminase (AID), which leads to numerous CG→TA transitions in the immunoglobulin (Ig) Ig locus.This process increases immunoglobulin diversity.Paradoxically, proteins involved normally in error-free base excision repair and mismatch repair, seem to be co-opted to facilitate SHM and CSR, by recruiting error-prone transle-sion polymerases to the DNA temple containing dU created by AID (Neuberger et al., 2003;Samaranayake et al., 2006).
Mammalian cells posses at least four enzymes with UDG activity, namely UNG, TDG, SMUG1 and MBD4.The major ones are nuclear UNG2 and mitochondrial UNG1 encoded by the UNG gene (Nilsen et al., 1997) and SMUG1 that also removes oxidized pyrimidines.The other ones are TDG that removes U and T from mismatches arising after deamination of C and 5-meC, respectively, and methyl binding domain IV (MBD4) that removes U from CpG contexts.UNG2 is found in replication foci during the S-phase and has a distinct role in the repair of U : A pairs, but it is also important in U : G repair, a function shared with SMUG1.Humans lacking UNG2 suffer from recurrent infections and lymphoid hyperplasia, and have skewed SHM and defective CSR, resulting in elevated IgM and strongly reduced IgG, IgA and IgE.UNG-defective mice also develop B-cell lymphoma late in life.The Phe251Ser UNG2 variant protein has been found to be mistargeted to mitochondria, resulting in deficient nuclear activity and increased uracil genomic content (Akbari et al., 2007;Kavli et al., 2007).
5-meC is normally present in DNA and constitutes up to 30% of total number of cytosines in the mammalian cell and at CpG sequences 5-meC is involved in silencing of gene expression (Li et al., 1992;Yoder et al., 1997).In humans, G : T mispairs arise from replication errors, which are handled by the mismatch repair pathway, or from the deamination of 5-meC to T. Because cytosine methylation occurs at CpG dinucleotides, G : T mispairs caused by 5-meC deamination are found at CpG sites.It has been shown that TDG is active for G : T mispairs with a 5' C : G pair, suggesting that a predominant biological role of the enzyme is to initiate the repair of CpG : T lesions.However, the U : G mispair is the most efficiently processed physiological substrate for TDG (Gallinari & Jiricny, 1996;Hardeland et al., 2001;Cortazar et al., 2007).Epigenetic silencing through methyl-CpG (mCpG) is implicated in many biological patterns such as genomic imprinting, X chromosome inactivation, and cancer development as well as the silencing of repetitive genetic elements.According to the facts described above, TDG could contribute to tumor suppression in a number of different ways: it may (i) help maintain genomic stability through the repair of mutagenic DNA base damage (e.g.deamination of C or 5-meC); (ii) provide epigenetic stability through the excision of erroneously methylated Cs in gene regulatory sequences; (iii) and/or it may assure proper cell differentiation and control the number of stem cells and/or tumor progenitor cells in certain tissues by its ability to cooperate with nuclear receptors and other tran-scription factors that integrate differentiation signals (Cortazar et al., 2007).
Another human DNA glycosylase, mentioned earlier, MBD4, exhibits specificity for G : T mispairs at CpG sites and also plays a role in the integrity of CpG sites (Hendrich et al., 1999).It has been shown that MBD4 binds to hypermethylated promoter of the MLH1 gene (MLH1 is a MMR protein).These results suggest that also MBD4 is one of the essential components involved in epigenetic silencing and its repair activity is necessary for the maintenance of hypermethylated promoters (Kondo et al., 2005).
Taking it all together, in human cells a few mechanisms exist that regulate the level and activity of DNA glycosylases by post-translational modifications (reviewed by : Tudek, 2007).Also UNG2 expression is up-regulated during S-phase of the cell cycle where the protein associates with PCNA and RPA at replication foci, implicating a role for this glycosylase in the removal of misincorporated U during DNA replication.Conjugation of SUMO to TDG induces glycosylase dissociation from DNA (Baba et al., 2006).Cells entering S-phase eliminate TDG through the ubiquitin-proteasome pathway.Degradation of TDG is critical for S-phase progression and cell proliferation.Strikingly, TDG levels decline just when UNG2 expression goes up and vice versa, suggesting that uracil repair is handled by distinct pathways throughout the cell cycle that are coordinated by the ubiquitin-proteasome system.The inability of TDG to discriminate between the parental and newly synthesized DNA strands would fix C→T transition mutations in cases where the T is in the parental strand.In addition, TDG-induced postreplicative G : T repair in the parental DNA strand, particularly in the parental lagging strand, could destabilize the replication fork and thereby impede the replication process.Thus, G : T correction during DNA synthesis should be left to the postreplicative mismatch repair system, which is designed to correct the error in the newly synthesized DNA strand (Hardeland et al., 2007).
Expansion of CAG trinucleotide repeats encoding polyglutamine has been identified as the pathogenic mutation in at least nine different genes associated with hereditary neurodegenerative disorders, including Huntington's disease (HD), dentatorubral pallidoluysian atrophy (DRPLA), spinal and bulbar muscular atrophy (SBMA), and the spinocerebellar ataxias: SCA1, SCA2, SCA3 (also known as Machado-Joseph disease), SCA6, SCA7, and SCA17 (Adachi et al., 2007;Underwood & Rubinsztein, 2007;Walker, 2007).Also, the two most common triplet expansion human diseases, myotonic dystrophy 1 and fragile X syndrome, are caused by expanded CTG/ CAG and CGG/CCG repeats, respectively (Wang, 2007).Moreover, oxidative lesions are known to be Mutations in genes involved in BER and DNA-end processors associated with ageing and neurological diseases (Olinski et al., 2007).Recently it has been found that the age-dependent somatic mutation associated with Huntington's disease occurs in the process of removing oxidized base lesions (8-oxoG) and is remarkably dependent on OGG1.OGG1 was shown to initiate an escalating oxidation-excision cycle that leads to progressive CAG expansion.Age-dependent CAG expansion provides a direct molecular link between oxidative damage and toxicity in post-mitotic neurons through a DNA damage response and error-prone repair of SSBs (Kovtun et al., 2007).
cOuLD APc BE A fActOR DEtERMININg tHE PAtcH SIzE DuRINg REPAIR SyNtHESIS?
The adenomatosis polyposis coli (APC) tumor suppressor is a multifunctional protein that is mutated in a majority of colon cancers.Close examination of the function of APC has shown that this multifunctional protein is involved in a wide variety of processes, including regulation of cell proliferation, cell migration, cell adhesion, cytoskeletal reorganization, and chromosomal stability.Clues to the different functions of APC have been provided by the identification of proteins interacting with several discrete motifs within APC.Each of these putative functions could link APC inactivation to cancerogenesis (reviewed by : Fodde et al., 2001;van Es et al., 2001).
Familial adenomatous polyposis (FAP) is caused by mutations in the APC gene.More than 800 mutations in the APC gene have been identified in families with classic and attenuated types of familial adenomatous polyposis.Most of these mutations cause the production of an APC protein that is abnormally short and nonfunctional.This short protein cannot suppress the cellular overgrowth that leads to the formation of polyps, which can become cancerous.The most common mutation in FAP is a deletion of five bases in the APC gene.This mutation changes the sequence of amino acids in the resulting APC protein beginning at positions 1309.Additionally, Ile1307Lys, Glu1317Gln, Asp1822Val, and other polymorphisms have been found.However, these kinds of polymorphism in APC gene are regional and population specific and are responsible together with environmental factors for the risk of colorectal cancer (Friedl et al., 2001;Locker et al., 2006;Guerreiro et al., 2007).Also, the Ile1307Lys mutation has been found to be clearly associated with a somatic additional adenine insertion in the region of codons 1306-1309, but other mutations in the region of codons 1277-1348 were found to be no more prevalent in carriers than in noncarriers (Zauber et al., 2005).
Recently it has been shown that human APC protein can interact with the human DNA Pol β-mediated one-nucleotide as well as strand-displacement synthesis of reduced abasic, nicked-, or 1-nt gapped-DNA substrates.APC also blocks strand-displacement synthesis of LP-BER and 5'-flap endonuclease as well as the 5'→3' exonuclease activity of FEN-1, resulting in the blockage of LP-BER.These studies will have important implications for understanding APC role in DNA damage-induced carcinogenesis and chemoprevention, especially critical APC role in several cellular processes (Narayan et al., 2005;Jaiswal et al., 2006).
Moreover, the fidelity of BER is dependent on the polymerization step, where the major BER Pol β incorporates nucleotide into the gap.Recent studies have indicated that expression of some Pol β variants or changes in expression of wild type Pol β protein, frequently found in cancer cells, can lead to DNA repair synthesis errors and confers to cells a mutator phenotype (reviewed by: Chan et al., 2006).In this case, it can not be excluded that APC could act not only as a factor determining the patch size during repair synthesis in BER by Pol β but also as a factor limiting the spearing of incorrect incorporation by Pol β.

BER ENzyMES AS BIOMARKERS IN MOLEcuLAR EPIDEMIOLOgy
Oxidative DNA damage and DNA repair mediate the development of several human pathologies, including cancer.The major pathway for oxidative DNA damage repair is base excision repair.Functional assays performed in blood leukocytes of cancer patients and matched controls show that specific BER pathways are decreased in cancer patients, and may be risk factors (Olinski et al., 1998).These include 8-oxoguanine (8-oxoG) repair in lung and head and neck cancer patients and repair of lipid-peroxidation-induced εA in lung cancer patients.Decrease of excision of lipid peroxidation-induced DNA damage εA and εC was observed in blood leukocytes of patients developing lung adenocarcinoma, a specific histological type of cancer related to inflammation and healing of scars (Gackowski et al., 2003;Speina et al., 2003;Speina et al., 2005a).The activity of BER proteins depends on gene polymorphism, interactions among BER system partners, and post-translational modifications.Polymorphisms of DNA glycosylases may change their enzymatic activities, and some polymorphisms increase the risk of inflammation-related cancers, colorectal, lung and other types.
Alternative splicing of the human OGG1 gene produces two major protein isoforms, α-OGG1 and β-OGG1.β-OGG1 is transferred to mitochondria, while α-OGG1 is targeted to the nucleus.Both isoforms of human OGG1 exhibit the same catalytic activity.Several single nucleotide polymorphisms (SNP) are present in OGG1 sequence, with Ser326Cys (C→ G in exon 6) being the most frequent.The Cys326 variant has lower activity than Ser326 and is not stimulated by APE1.Since APE1 stimulates excision of 8-oxoG from DNA by OGG1, it seems reasonable that the activity of 8-oxoguanine DNA glycosylase is significantly affected by the Ser326Cys polymorphism (Collins & Gaivao, 2007).Therefore, the human OGG1-Cys326 variant has been proposed to increase the risk of lung cancer, prostate cancer, and nasopharyngeal carcinoma.Also the rare Arg154His (G→T) OGG1 polymorphism has been identified in sporadic colorectal cancers but did not segregate with cancer phenotypes.Two other less frequent OGG1 polymorphisms, Arg46Gln and Arg154His, also influence OGG1 activity and were detected in human lung and gastric cancers (reviwed by : Nohmi et al., 2005).
Alternative splicing of the human MUTYH gene produces two major protein isoforms, type 1 (535 amino acids), which localizes in mitochondria, and type 2 (521 amino acids), which is transferred to the nucleus.Interestingly, type 2 protein has a higher glycosylase activity than type 1 protein.MUTYH interacts with a number of proteins, such as RPA, APE1, PCNA and MSH6 (Fig. 3), and its expression is increased during S phase.Thirty various mutations that are predicted to turncate the protein product have been reported in MUTYH gene, comprising 11 nonsense, 9 small insertion/deletion and 10 splice site variants.In addition, 52 missense variants and 3 small inframe insertion/deletions have been reported (reviewed by: Cheadle & Sampson, 2007).Gly382Asp and Tyr165Cys substitutions cover more than 70% mutations reported in MUTYH gene and are linked with higher cancer incidence.The Gly382Asp and Tyr165Cys MUTYH variant proteins that are devoid of glycosylase activity towards the 8-oxoG : A pair, have been found in familial polyposis.These mutations are associated with GC→TA transversions in the APC gene.Other MUTYH alterations which have been found in patients with colorectal tumors are missense Tyr90 or Glu466 to stop codon mutations (reviewed by : Nohmi et al., 2005).
For other components of BER, the association of mutations in genes encoding proteins engaged in base excision repair with cancerogenesis appears less consistent.However, Pol β has been found to be overexpressed at both mRNA and protein level in about 30% of all tumors studied, with the overexpression being particularly frequent in uterus, ovary, prostate and stomach.Pol λ and Pol ι were also found to be overexpressed to a significant extent in a range of tumor types, albeit less frequently than Pol β (Albertella et al., 2005).Additionally, approx.30% of human tumors examined for mutations in POLB gene appear to express Pol β variant proteins (Starcevic et al., 2004).Many of these variants result from a single amino-acid substitution.The Lys289Met and Ile260Met variants exhibit reduced polymerase fidelity and are observed in colon and prostate cancer, respectively (Lang et al., 2004;2007;Sweasy et al., 2005).Moreover, the Glu295Lys gastric carcinoma Pol β variant acts in a dominant-negative manner by interfering with BER, which leads to an increase in sister chromatid exchanges and genomic instability indicating that BER is critical for maintaining genome stability and could therefore be a tumor suppressor mechanism (Lang et al., 2007).The Pro242Arg and Lys289Met polymorphism of Pol β can be an auxiliary marker for breast cancer risk and cancer progression (Sliwinski et al., 2007).Apart from the single substitution, several Pol β cancer-related variants were found, e.g.truncation and deletion mutants.The wild type and truncated forms of Pol β proteins are expressed in primary colorectal and breast adenocarcinomas and in a primary culture of renal cell carcinoma.Three types of deletion variants were detected in squamous, non-small, or large cell carcinomas.The most common variant was a deletion of 87 bp from POLB cDNA at a site corresponding to exon 11.In addition, a variant exhibiting deletions of 87 and 140 bp together with an insertion of 105 bp was identified in lung tumors (Bhattacharyya & Banerjee, 1997;Bhattacharyya et al., 1999;Chen et al., 2000;Bhattacharyya et al., 2001;Bhattacharyya & Banerjee, 2003).Additionally, the 208-236 deletion variant found in many human tumors has been shown ex vivo to reduce BER capacity.Pol β -/-knockout mice are not viable and Pol β +/-haploinsufficient mice demonstrate higher level of SSBs and increased chromosomal aberrations (Cabelof et al., 2003).Also a few SNPs in POLL and POLI genes were found, resulting in amino-acid substitutions within the Pol λ and Pol ι variant proteins, respectively.Mutation in POLI has been shown to be associated with NSCLC (Sakiyama et al., 2005).
Several amino-acid substitution variants were identified in the repair domain of human APE1.Functional characterization revealed that the variants, Leu104Arg, Glu126Asp and Arg237Ala, exhibited approx.40-60% reductions in specific incision Mutations in genes involved in BER and DNA-end processors activity.Moreover, the Asp283Gly and Asp283Ala variants were found to exhibit approx.10% repair capacity.The most common substitution Asp148Glu had no impact on endonuclease and DNA binding activities, nor did the Gly306Ala substitution.The Gly241Arg variant showed a slightly enhanced endonuclease activity relative to the wild type.All reduced function variants may be associated with increased disease susceptibility (Hadi et al., 2000).However, a significant association between the Asp148Glu (T→G in exon 5) polymorphism in APE1 gene and lung cancer risk was found (De Ruyck et al., 2007).Ape1 -/-knockout mice are not viable (Friedberg & Meira, 2004).
Poly(ADP-ribose) polymerase 1 (PARP-1) modifies a variety of nuclear proteins by poly(ADPribosyl)ation, and plays diverse roles in molecular and cellular processes.PARP-1 is also a platform protein associated with SSBR and interacts with several DNA repair proteins (Fig. 3).The common Val762Ala polymorphism of PARP-1 in the catalytic domain is implicated in susceptibility to cancer.The PARP-1 Val762Ala polymorphism reduces the enzymatic activity (Wang et al., 2007).PARP-1 deficiency causes mammary tumorigenesis in mice underlying the role of PARP-1 in suppressing mammary tumorigenesis in vivo and suggesting that dysfunction of PARP-1 may be a risk factor for breast cancer in humans (Tong et al., 2007).Recently, all PARP-1 exons, intronexon junctions and promoter sequences have been sequenced.Rare genetic variants of PARP-1, including Ser383Tyr (C→A), Arg452Arg (C→A), Lys940Arg (C→G) were detected in about 11% breast cancers.Interestingly, the Ala284Ala (T→C) PARP-1 variant was likely associated with loss of estrogen-and progesterone-receptor expression.This implies that genetic variants of PARP-1 may contribute to breast cancerogenesis and that the PARP-1 Ala284Ala variant protein may influence hormonal therapy of breast cancer (Cao et al., 2007).Parp-1 -/-knockout mice are viable.In contrast, double Parp-1 -/-Parp-2 -/-mice are not (Friedberg & Meira, 2004).
BER is also changed in tumors in comparison to unaffected surrounding tissues, and this change may be due to transcription stimulation, post-translational modification of BER enzymes as well as protein-protein interactions.Modulation of BER enzymes' activities may be, then, an important factor determining the risk of cancer and also may participate in cancer development (Tudek et al., 2006;De Ruyck et al., 2007;Tudek, 2007).

SuMMARy
As outlined above, the development of mutator phenotype is proposed to be an early step in car-cinogenesis.The best known examples of such a situation are defects in the MUTYH gene which increase the G→T transversions in the APC gene in human colorectal cancers.Importantly, these mutations are frequently formed in hot spots of tumor suppressor genes or oncogenes, and thus further influence carcinogenesis.DNA-damaging agents may also preferentially modify hot spots of tumor suppressor genes or oncogenes.For instance, etheno-adduct-forming chemicals, such as vinyl chloride and urethane, have been shown to induce specific hot spot mutational patterns in TP53 (Kowalczyk et al., 2006) and Ha-ras in liver, lung or skin cancers in humans (occupationally exposed to vinyl chloride), rats, and mice (reviewed by: Barbin, 2000).Individual susceptibility is an important factor in cancer development that depends on carcinogen uptake, balance between metabolic activation and detoxification, DNA repair activity, and varying effects of genes involved in DNA repair, signal transduction pathways and regulation of the cell cycle.
Polymorphisms in genes encoding DNA repair functions can lead to varying capacities of defense against endogenous and environmental DNA damaging agents.Since variations in DNA repair genes may influence and modulate an individual's cancer susceptibility, screening for polymorphisms has become recently a promising area of research in molecular epidemiology.Moreover, this knowledge is necessary to allow a number of DNA repair inhibitors as potential anticarcinogenic compounds.However, the ability of the DNA repair inhibitors to prevent cancer development is difficult to interpret and is sure to depend upon the system used and the type of genotoxic stress.

Figure 1 .
Figure 1.Major endogenous DNA base modifications.Modified positions are shown in red.