DNA repair of oxidative DNA damage in human carcinogenesis

Abstract

Efficient DNA repair mechanisms comprise a critical component in the protection against human cancer, as indicated by the high predisposition to cancer of individuals with germ-line mutations in DNA repair genes. This includes biallelic germ-line mutations in the MUYH gene, encoding a DNA glycosylase that is involved in the repair of oxidative DNA damage, which strongly predispose humans to a rare hereditary form of colorectal cancer. Extensive research efforts including biochemical, enzymological and genetic studies in model organisms established that the oxidative DNA lesion 8-oxoguanine is mutagenic, and that several DNA repair mechanisms operate to prevent its potentially mutagenic and carcinogenic outcome. Epidemiological studies on the association with sporadic cancers of single nucleotide polymorphisms in genes such as OGG1, involved in the repair of 8-oxoguanine yielded conflicting results, and suggest a minor effect at best. A new approach based on the functional analysis of DNA repair enzymatic activity showed that reduced activity of 8-oxoguanine DNA glycosylase (OGG) is a risk factor in lung and head and neck cancer. Moreover, the combination of smoking and low OGG activity was associated with a higher risk, suggesting a potential strategy for risk assessment and prevention of lung cancer, as well as other types of cancer.

1. Introduction

The multiple biochemical reactions in which oxygen is involved lead to the formation of reactive toxic intermediates that may cause DNA damage. Indeed, while life depends on oxygen, it also depends on the activity of sophisticated mechanisms that have evolved to detoxify potentially dangerous reactive oxygen species (ROS), and repair oxidative DNA damage once it was formed in DNA. Because oxidative damage to DNA can cause mutations, and mutations are known to cause cancer, much effort has been devoted to study the role in carcinogenesis of oxidative DNA damage, its prevention, and repair [1,2]. The pivotal role that DNA repair plays in defending humans against cancer became evident with the accumulation of studies showing that many of the human hereditary cancer predisposition diseases are caused by germ-line mutations in DNA repair genes [2-9]. Interestingly, even in these cases, although the DNA repair defect is present in all cell types, cancer is limited to certain tissues, underscoring the notion that a deficiency in a particular repair gene in itself is not sufficient to cause cancer. Additional major factors are involved, including the level of DNA damage generated in a particular tissue, and the activity of other protection mechanisms at both the molecular and cellular levels, such as alternative DNA repair pathways, apoptosis, and the immune system.

In contrast to hereditary cancer, our molecular understanding of the role of DNA repair in sporadic cancer, which accounts for most cancer cases in humans, is limited, as the translation of the wealth of molecular understanding of basic DNA repair mechanisms into human cancer has been slow. A major cause is that in contrast to germ-line mutations in DNA repair genes, which cause a strong deficiency in DNA repair activity in all cell types, in sporadic cancer, where there is no germ-line mutation, deficiencies in DNA repair, if any, are expected to be much milder. This makes research on the role of DNA repair in sporadic cancer much more difficult. Further slowing down progress are the paucity of accurate and reproducible assays suitable for epidemiological studies, poor epidemiological design of many studies, and a rather low activity in translational research of investigators in the field of basic research in DNA repair. Recently significant progress was made on the role of oxidative damage repair in human cancer, as more reliable DNA repair assays are being developed, and studies conformed to more careful epidemiological criteria. This review will focus on the role in human cancer of the repair of a specific type of oxidative DNA damage, 8-oxoguanine. This damage, which is mutagenic, is one of the major oxidative DNA lesions, and frequently taken as a measure of oxidative stress [1].

2. Mechanisms of repair of DNA containing 8-oxoguanine

8-Oxoguanine is formed in DNA by one of two main mechanisms: (a) oxidation of a guanine base in DNA [10]; (b) incorporation of an oxidized dGTP (8-oxo-dGTP) during DNA synthesis [11]. The mutagenicity of 8-oxoguanine stems from its propensity to form a mispair with A, which can lead to a GC→TA transversion if a dAMP moiety is inserted opposite a template 8-oxoguanine, or an AT→CG transversion if 8-oxo-dGMP is inserted opposite a template A [11-14]. The deleterious effects that 8-oxoguanine may have on life can be clearly deduced from the existence of multiple mechanisms that have evolved to neutralize it. Thus, organisms from bacteria to humans have no less than three dedicated mechanisms, as well as at least two backups, which protect them from 8-oxoguanine. The main protective mechanisms operate at three levels (Fig. 1): (a) Prevention — to avoid the incorporation of 8-oxo-dGMP in DNA by hydrolysis of 8-oxo-dGTP. This reaction is catalyzed by the enzyme MTH1. (b) Repair — to excise 8-oxoguanine from DNA and restore the original DNA sequence. This is carried out primarily by OGG1-initiated base excision repair (BER). (c) Proofreading — to remove an adenine misincorporated opposite a template 8-oxoG in order to enable conversion of the premutagenic 8-oxoG:A mispair into a 8-oxoG:C base pair. This reaction is carried-out by MUTYH-initiated BER. The 8-oxoG:C base pair can subsequently be repaired by OGG1-initiated BER to restore the G:C base pair.

Fig. 1.

The major protective mechanisms against the oxidative DNA lesion 8-oxoguanine. Prevention, involves MTH1-catalyzed hydrolysis of 8-oxo-dGTP, to prevent its incorporation into DNA by DNA polymerases during DNA synthesis. Repair, involves OGG1-initiated base excision repair (BER), which repairs the 8-oxoG:C base pair to the original G:C base pair. Proofreading operates on the 8-oxoG:A mispair. When A is in the newly synthesized DNA strand, it is eliminated by the MUTYH adenine DNA glycosylase, thereby enabling the formation of a 8-oxoG:C base pair that can be repaired by OGG1-initiated BER. When the A is in the template strand, the 8-oxoG might be removed by mismatch repair (MMR). See text for details.

2.1. Prevention

8-Oxo-dGTP can be used as a substrate by DNA polymerases during DNA synthesis, leading to the incorporation of 8-oxo-dGMP. This occurs opposite a template C, but also opposite template A, thereby forming a pre-mutagenic intermediate A:8-oxoG mispair (Fig. 1; [11,14]). The MTH1 protein acts to limit this reaction by hydrolyzing 8-oxo-dGTP to 8-oxodGMP, thereby eliminating it from the pool of DNA synthesis precursors. The MTH1 gene is located on chromosome 7p22. It consists of 5 exons, and can produce up to seven types of mRNA as a result of multiple initiation sites and alternative splicing. Type 1 mRNA and the 18kDa protein variant are the major products of MTH1 [15]. The enzyme hydrolyzes also two oxidation products of dATP, namely 2-hydroxy-dATP and 8-oxo-dATP, and to a lesser degree also the oxidized rGTP [16,17]. Interestingly, humans contain an enzyme related to MTH1, termed NUDT5. This enzyme, originally known as ADP-sugar pyrophosphatase, was found to effectively hydrolyze 8-oxo-dGDP. 8-Oxo-dGDP is formed by enzymatic cleavage of 8-oxodGTP and by oxidation of dGDP, and although it is not a substrate for DNA polymerases, it can be easily converted into the triphosphate form by nucleoside diphosphate kinase [18].

2.2. Repair

8-Oxoguanine present in DNA as part of an 8-oxoG:C base pair can be produced by oxidation of a guanine in DNA, or when residual 8-oxo-dGTP, which has escaped the MTH1-catalyzed hydrolysis, incorporates opposite a template C. In that case, the 8-oxoguanine is removed from DNA by BER (Fig. 1), a key DNA repair mechanism that operates usually on bases with small modification, including many oxidative DNA lesions. BER is initiated by a DNA glycosylase that recognizes the modified base, and eliminates it from DNA by cleaving the glycosylic bond linking the base to the deoxyribose sugar in the DNA backbone, leaving behind an abasic site. Approximately 10 different DNA glycosylases are known in mammalian cells, each operating on a small group of chemically related base modifications. Some DNA glycosylases have an associated AP lyase activity, which cleaves the abasic site, whereas others lack it [1,19-21]. 8-Oxoguanine is removed primarily by 8-oxoguanine DNA glycosylase 1 (OGG1), which possesses an associated AP lyase activity [22,23]. Subsequent DNA repair steps are generally common to all BER sub-pathways, and involve the activity of AP endonuclease 1, DNA polymerase β, and DNA ligase III or I, although the precise mechanism slightly differs for DNA glycosylases without an associated AP lyase activity [1,24]. Interestingly, AP endonuclease I stimulate the enzymatic activity of OGG by actively increasing its turnover [25,26]. Another DNA glycosylase that can excise 8-oxoguanine from DNA is NEIL1, but its activity is rather weak [27,28]. Remarkably, while OGG1 acts on 8-oxoguanine in dsDNA, NEIL1 acts on 8-oxoguanine also in ssDNA segments of bubble DNA substrates, suggesting that it may function during BER linked to transcription or replication [29]. The NEIL1-dependent pathway does not require APE1, but requires polynucleotide kinase instead, for processing the intermediate formed by the DNA glycosylase and AP lyase activities of NEIL1 [30]. There is evidence to suggest that nucleotide excision repair (NER) might act as a backup for the repair of 8-oxoguanine [31].

2.3. Proofreading

An A inserted opposite 8-oxoG gives rise to an 8-oxoG:A mispair, which is not a target for OGG1 [32], most likely to avoid a situation in which the erroneous A is used as a template, thereby establishing a GC→TA mutation. Instead, a special adenine DNA glycosylase, MUTYH (MutY homolog; previously termed MYH), identifies the 8-oxoG:A base pair, and removes the A rather than the oxidized G (Fig. 1; [33-35]). This might seem strange at first glance, since the action of MUTYH must be followed by DNA synthesis across 8-oxoG in the template strand, which is potentially mutagenic. However, synthesis across 8-oxoG has a chance of at least 80% to insert a C opposite 8-oxoG [36-38]. If this happens, OGG1 can make an additional attempt of error-free removal of the 8-oxoG. In 20% of the cases when 8-oxoG miscodes and an A is inserted again, MUTYH will act again [33,38]. Interestingly, MUTYH was found to excise from DNA also 2-hydroxyadenine, another type of oxidative lesion [39], suggesting that its function may be broader than the removal of A from 8-oxoG:A mispairs. The MUTYH gene contains 16 exons, and is localized on the short arm of chromosome 1, between p32.1 and p.34.3. It produces up to 10 alternatively spliced transcripts, and encodes multiple forms of the enzyme, which are targeted to both the mitochondria and the nucleus [40]. Interestingly, MUTYH activity is replication-coupled [41], and the enzyme directly interacts with a number of DNA replication and DNA repair proteins, including the DNA sliding clamp PCNA (proliferating cell nuclear antigen), the AP endonuclease APE1 (HAP1), the ssDNA-binding protein RPA (replication protein A), and the mismatch repair protein MSH6 [40]. When 8-oxoG is inserted opposite a template A, the action of MUTYH might cause a fixation of an AT→CG mutation. Although it is not clear yet whether this indeed happens in humans, it is possible that under such circumstances mismatch repair removes the 8-oxoguanine from DNA [42-44].

3. Repair of 8-oxoG and carcinogenesis in mice

Each of the three main genes involved in the protection against 8-oxoguanine was knocked-out in mice, providing a powerful tool for molecular analysis of their in vivo activity, as well as their involvement in carcinogenesis. Somewhat surprisingly, no drastic phenotypes were observed in either of the MTH1^-/-, OGG1^-/-, or MUTYH^-/- knockout mice (recently reviewed in [45,46]). Thus, OGG1^-/- mice showed an abnormal twofold increase in the amount of chromosomal 8-oxoguanine, and increased spontaneous mutagenesis. However, the effect on tumorigenesis was marginal: Two studies did not find an effect on spontaneous carcinogenesis, whereas one study reported spontaneous development of lung cancer 1.5 years after birth [47-49]. MUTYH^-/- mice were reported to have no effect on spontaneous tumor incidence when examined 15-17 months after birth in one study [50], while in another study they were reported to have an overall 1.7-fold in spontaneous tumors, and a more pronounced 5-fold increase in the incidence of intestinal tumors 18 months after birth [51]. MTH1^-/- mice showed a marginal increase in general spontaneous mutagenesis, and a significant 5.7-fold increase in single nucleotide frameshifts in mononucleotide runs. When examines 18 months after birth, they were found to have more tumors in the lung, liver and stomach compared to MTH1^+/+ mice [52].

The overall picture that emerges from these studies in mice is that while defects in the main enzymes involved in the repair of 8-oxoguanine do facilitate spontaneous carcinogenesis, the effects are rather mild and occur late in life. This is somewhat surprising given the abundance of 8-oxoguanine and its known mutagenic effects. A possible obvious explanation is that there exist backup mechanisms that act to minimize the effect of 8-oxoguanine in these knockout mice. Indeed, a DNA repair pathway that depends on the Cockayne syndrome gene CSB but not on OGG1 was described [53,54]. In addition, the mismatch repair system might be involved in the repair of 8-oxoguanine, as described above [42,43]. These results highlight the importance of mice model systems as extremely powerful tools in the elucidation of molecular mechanisms for the repair of 8-oxoguanine (and, of course, other types of DNA damage). On the other hand, as far as carcinogenesis is involved, there are major differences between mice and humans in the effects of specific DNA repair enzymes. These differences might reflect the difference in the life span of mice and humans: whereas mice evolved to live approximately 2 years, human live 40-50 fold longer. Therefore, deficiencies in the repair of DNA lesions, which may not have a strong effect in mice during their short life span, may in fact have a dramatic effect on human cancer susceptibility as will be discussed below.

4. Repair of 8-oxoguanine and hereditary cancer in humans

Germ-line mutations in DNA repair genes operating in several pathways were found to cause very strong predisposition to hereditary forms of cancer [1,55]. These include genes involved in nucleotide excision repair (predisposition to skin cancer), mismatch repair (predisposition to hereditary colorectal cancer), non-homologous end joining (e.g., predisposition to lymphoma), and homologous recombination repair (e.g., breast cancer) [2-9]. As far as the main pathways involved in the repair of 8-oxoguanine, germ-line biallelic mutations in the MUTYH gene were found to cause MUTYH-associated polyposis (MAP), a disease that causes strong predisposition to a hereditary form of colorectal cancer [5,56,57]. This was discovered in a family with several members showing clinical symptoms of familial adenomatous polypolsis (FAP), usually caused by germ-line in the APC gene. In this family however, APC mutations were found in the tumor, but not in the germ-line. An unusually high proportion of the somatic APC mutations were GC→TA transversion, characteristic to deficiencies in the repair of 8-oxoguanine, as described above. DNA sequence analysis has revealed the presence of bi-allelic mutations in MUTYH [5]. Based on this and subsequent studies MAP is responsible for approximately 1% of all colorectal cancer cases. MAP is the first, and so far only case of cancer caused by a germ-line mutation in a gene belonging to the BER pathway. So far no germ-line mutations in OGG1 or MTH1 were identified in human cancer. While additional studies may reveal germ-line mutations in OGG1 and/or MHT1, the existence of backup mechanisms for the repair of 8-oxoguanine which is part of a 8-oxoG:C base pair may ameliorate repair deficiencies caused by mutations in these genes. Alternatively, the OGG1 and/or MTH1 genes may be essential in humans.

5. Repair of 8-oxoguanine and sporadic cancer in humans

In sporadic cancers, which account for a large majority of human cancer, there is no single germ-line mutation that causes a strong predisposition. Rather, a multitude of genetic and environmental components cooperate to form cancer. This together with the obvious limitations of conducting experiments in humans makes research of the genetic origins of sporadic human cancer much more difficult. Moreover, the enormous inter-individual variation in essentially each parameter, from genetic, via biochemical, up to environment and lifestyle parameters, necessitates extra precaution in the design and interpretation of studies. Most experiments are correlative in nature, and therefore it is not easy to unambiguously identify causality when dealing with attempts to identify the involvement of DNA repair in sporadic cancer. Two main issues need to be considered in the investigation of the role of DNA repair in cancer: (a) Are inter-individual variations in DNA repair capacity manifested in the risk to develop a sporadic cancer, and (b) How do variations in DNA repair affect the growth of a transformed cells on its path to become a cancer cell, and how are they manifested in disease prognosis and response to therapy. The two major approaches that have been utilized to study these issues are the genetic approach based on single nucleotide polymorphisms in DNA repair genes, and the functional assay approach.

5.1. Genetic polymorphisms in DNA repair genes

The existence of millions of polymorphisms in the human genome sparked the hope that these DNA sequence variations are at the basis of inter-individual variations, and will be useful for risk assessment of diseases, including cancer [58]. Translating this approach to the role in cancer etiology of DNA repair of oxidative damage such as 8-oxoguanine, many studies have undertaken to identify SNPs in DNA repair genes, and examine whether they are associated with cancer risk. The most common SNP studied in the OGG1 gene is OGG1Ser326Cys. Analysis of the biochemical properties of the purified OGG1Ser326Cys variant enzyme led to conflicting results. Dherin et al reported that the wild-type and variant OGG proteins have similar enzymatic propertied [59], while Hill and Evans [60] reported that the variant OGG1 is deficient in it catalytic activity, appeared in an atypical dimeric form, and unlike the native enzyme, was not significantly stimulated by the presence of AP endonuclease. To add to the confusion, Janssen et al reported that the excision activity of OGG1 in protein extracts prepared from human lymphocytes was not affected by the Ser326Cys polymorphism [61]. Finally, Luna et al reported that unlike the wild-type enzyme, the variant OGG1 did not translocate to the nucleoli during the S phase of the cell cycle, most likely due to a change in its phosphorylation status [62]. Taken together these reports underscore the difficulty that may arise in the apparently straightforward task of evaluating the functional significance of a variant protein.

Epidemiological studies addressing the association between the OGG1Ser326Cys polymorphism and cancer similarly led to conflicting results. A meta-analysis of studies published until 2005 on the association of the OGGSer326Cys SNP and lung cancer, which combined data from 7 studies, yielded an odds ratio 1.24 (95% CI 1.01 to 1.53). Combining 3 studies on upper aerodigestive tract cancers, they found an odds ratio of 1.15 (95% CI 0.90 to 1.46) [63]. A later large study found no association between this polymorphism and lung cancer in Denmark [64]. Studies on polymorphism in other genes involved in BER of oxidative DNA damage, such as APE1 or XRCC1, similarly yielded either weak or no associations [63,65]. These results suggest that SNPs are too simple to be reliable predictors of complex traits such as cancer risk. Moreover, in most cases they may not even be significant determinant of activity in vivo, because they are only one of many factors that affect gene expression and protein activity. A better approach may be to evaluate the role of SNPs in multiple DNA repair genes [66,67]. However, this approach might be limited by the small number of individuals who are affected by multiple SNPs in DNA repair genes.

5.2. Functional DNA repair assays

Determining of DNA repair activity has an obvious advantage over SNP analysis, since it measures the actual capacity of a particular repair activity. There are three main obstacles to using such an approach: (1) The assays are usually more complex that genotyping; (2) The needs to use a surrogate tissue (usually blood), necessitates to demonstrate the relevance of the results obtained to the actual target tissue. (3) In retrospective studies, which are a critical research methodology in these studies, it is difficult to discriminate between cause and result [68]. In other words, when groups of cancer patients and healthy individuals are compared, a lower DNA repair capacity in the cancer patients can be interpreted either by predisposition, or an effect of the cancer. This last limitation does not exist in prospective studies. All three difficulties, namely complexity, location, and timing do not exist in genotyping: The assay is very simple and high throughput, and because SNPs are in the germ-line, it is clear that they have preceded the tumor, and are present in the target tissue. This explains why genotyping is so popular, however, it is often ignored that in general, SNPs are too simple to be significant predictors of in vivo activity, and even less so, of cancer risk. Recently several studies on the role of the repair of oxidative DNA damage in cancer etiology took the functional approach, yielding exciting results [69-73].

5.2.1. Low OGG activity is associated with the risk of lung and head and neck cancer

The activity of the enzyme OGG1 can be conveniently assayed in protein extracts prepared from peripheral blood mononuclear cells (PBMC), which are mostly lymphocytes [73]. We and others have used a synthetic duplex oligonucleotide carrying a site-specific 8-oxoguanine as a substrate [61,69,70,74]. When incubated with the protein extract, OGG1 activity excised the 8-oxoguanine, and the abasic site thus formed was subsequently cleaved by either the AP endonuclease activity present in the extract, or the AP lyase activity of OGG1. This OGG1-dependent nicking activity was quantified by measuring the amounts of substrate and cleaved product. Although OGG1 was shown to be responsible for most of glycosylase activity against 8-oxoguanine [75], there exist in the extract minor activities that remove this lesion (e.g., NEIL1, as discussed above), and therefore we termed this assay the OGG assay (for the general 8-oxoguanine DNA glycosylase activity) rather than OGG1 assay. A thorough optimization, and detailed reproducibility study established that the OGG assay is reproducible, with an inter-investigator variability of 10%. Analysis of 120 healthy individuals revealed a mean OGG activity of 7.2 units/μg protein, and a distribution range of 2.8-fold from 3.6 to 10.1 units/μg protein, suggesting that individuals at the lower end of this distribution might be predisposed to cancer. Interestingly there was no difference in OGG activity between males and females, and no difference between smokers and non-smokers [73].

A case-control study with 68 non-small cell lung cancer patients and 68 healthy control subjects, frequency matched for age and sex, showed that OGG activity was reduced in non-small cell lung cancer (NSCLC) patients compared to healthy subjects [69]. Gackowski et al, using a similar assay in a study with 51 NSCLC patients and 64 healthy subjects, also found lower OGG activity in the cancer patients [70]. We went further to analyze our results by three logistic regression models, after adjusting for age and smoking status. When groups were divided by medians, the odds ratio was 5.2 (95% CI 1.9 to 14.1), with P=0.001; when the groups were divided by tertiles, the odds ratio was 4.8 (95% CI 1.5-15.9), with P=0.01. Using a continuous model, the odds ratio associated with a unit decrease of OGG activity was 1.9 (95% CI 1.3 to 2.8), with P<0.001. These results suggest that low OGG activity is associated with increased risk of NSCLC, and is independent on smoking [69].

To further examine the biological and epidemiological significance of these results we addressed the two main issues raised above, namely the validity of measurements in a surrogate tissue, and whether reduced OGG activity preceded the onset of the lung cancer. Since the main therapy of NSCLC is surgery, we had an opportunity to compare OGG activity in PBMC and non-cancer tissue from the same individuals. We found a linear correlation between OGG in blood and lung, suggesting that at least for OGG enzymatic activity, PBMC are a valid surrogate for the lung [69]. The question of whether the low OGG activity in NSCLC preceded the onset of cancer cannot be definitively answered in a case-control study; for that, a prospective study is needed. However, we could obtain supporting evidence, by assaying OGG activity in patients from 10 days before surgery, up to over a year after surgery. The prediction was that if OGG activity is an inherent characteristic of an individual regardless of the presence of the tumor, then there should be no difference in OGG activity before or after surgery. However, if low OGG activity was caused by the tumor, as time after surgery passes, the effect of the tumor may weaken, or even disappear, leading to an increase in OGG activity. We found no association between the time of blood specimen withdrawal and time of surgery, suggesting that OGG activity was not affected by the tumor [69,76].This, of course, is not a definitive proof, since the effect of the tumor (if any) could have been irreversible. However, the fact that OGG values did not change even over a year after surgery, when at least part of the lymphocyte population turned over, is consistent with lack of an effect of the tumor on OGG activity [76]. Definitive proof however must await prospective studies. Taken together these results suggest that low OGG activity measured in PBMC is associated with the risk of NSCLC, with the estimated relative risk increasing as OGG activity decreases.

Most interestingly, when reduced OGG activity was combined with smoking, there was a much greater estimated relative risk for NSCLC (Fig. 2), suggesting that smokers with low OGG activity are highly predisposed to lung cancer [69,77].

Fig. 2.

Estimated relative risk to develop non-small cell lung cancer of smokers with reduced OGG activity. The reference group is of non-smokers, with an OGG activity of 7 units/μg protein. The graph is based on logistic regression analysis with OGG activity as a continuous variable, as presented in ref 68.

We have conducted also a case-control study on the role of OGG activity in head and neck cancer with 37 cases with squamous cell carcinoma of the head and neck (SCCHN), and 93 control subjects matched for age and gender [72]. We found that reduced OGG activity was associated with SCCHN, with an adjusted odds ratio associated with a unit decrease in OGG activity of 2.3 (95% CI 1.5 to 3.4; P<0.0001). Individuals in the lowest tertile of OGG activity were at increased risk with an odds ratio of 7.0 (95% CI 2.0-24.5; P=0.002). Like in the case of lung cancer, the combination of low OGG activity and smoking caused a much greater risk for SCCHN. To address the question of whether low OGG activity is a risk factor in SCCHN or caused by the cancer, we re-assayed OGG activity in 18 patients who fully recovered from the disease after successful therapy. In all 18 patients, OGG activity values measured 3-4 years after successful treatment were within the 95% CI of the OGG activity values measured at diagnosis [72]. This suggests that the low OGG activity measured at diagnosis was not caused by the presence of the cancer. Rather it represents inherent characteristics in these individuals, and caused increased risk of SCCHN.

In an attempt to examine the origins of the inter-individual variations in OGG activity, we assayed in parallel OGG activity and OGG1 mRNA expression in the same individuals. Although there was a group trend of increased OGG activity with increased mRNA expression, correlation at the individual level was poor [73], indicating that mRNA expression cannot be used to predict OGG enzymatic activity. Evidently, factors other than mRNA expression affect OGG activity, such as translation, as well as inhibitors and stimulators, such as AP endonuclease, as described above. Notably, also the protein complex XPC-HR23B, which is usually involved in NER, was found to stimulate the activity of OGG1 [78]. This suggests that inter-individual variability in OGG1 activity might be influenced by inter-individual variations in many other cellular components, including AP endonuclease and XPC-HR23B.

5.3. Other methods for measuring repair of oxidative DNA damage

5.3.1. Measuring 8-oxoguanine in genomic DNA

Measuring the steady state of DNA damage reflects the sum of the rates of its formation and repair. The advantage of this method is that it integrates exposure and repair, and thereby offers a ‘bottom-line’ of the amount of damage in DNA [79-83]. There are, however, two main limitations to this approach. (a) The baseline amount of most types of DNA damage is low, and may be very close to the limit of detection of the analytical method used for its measurement. This causes a rather low signal/noise ratio and potentially introduces large measurement errors. (b) Most studies measure lesions in genomic DNA. Since most DNA is non-coding, this means that lesions are practically measured in non-coding DNA. Since there are heterogeneities in both induction of DNA damage and its repair, measurement in non-coding DNA might yield either irrelevant or misleading results. Notable, DNA repair can preferentially occur in expressed sequences [1], which are likely to be more important in carcinogenesis than non-coding DNA. This can be overcome by using gene-specific measurement of DNA damage [84]. An additional complication is that the surrogate tissue used (usually blood) is not necessarily subjected to the same level of exposure to DNA damaging agents as the target tissue. In the case of 8-oxoguanine an additional complication is the ease by which routine sample processing causes oxidation of guanine in DNA, thereby further increasing background noise [85]. Still, it was reported that in lung cancer patients who had lower enzymatic activity of OGG than control healthy subjects, had consistently also higher amounts of 8-oxoguanine in their DNA [70].

5.3.2. Expression arrays and proteomics

The availability of high throughput platforms for the simultaneous measurement of expression patterns of most (if not all) genes in the human genome has led to many attempts to use individual expression profiles primarily for detection of cancer and prognosis of treatment. Research is focusing on mRNA expression arrays, proteomics and antibody arrays [86-91]. Obviously, such studies, which address a large number of genes and/or proteins, can provide information also on the role of DNA repair in cancer.

5.3.3. The comet assay

The comet assay measures breaks in chromosomal DNA. Cells embedded in agarose on a microscope slide are lysed with detergent and high salt to form nucleoids. Electrophoresis at high pH results in structures resembling comets, the intensity of the comet tail relative to the head reflects the number of DNA breaks. Specific oxidative DNA damage can be identified by including a step in which the nucleoid DNA is incubated with a lesion-specific endonuclease (e.g., OGG1 for 8-oxoguanine), which increases the number of breaks. This increases the intensity of the comet tail, thereby improving the quantification of the results, which is a weakness of this assay. An advantage of the comet assay is its high sensitivity, and it can be conducted with single cells.

The comet assay was used to measure DNA repair capacity by monitoring the speed by which cells remove lesions, e.g., after exposure of freshly isolated lymphocytes to an oxidative damaging agent such as H₂O₂. However, it should be remembered that the breaks measured represent, in addition to DNA repair, other mechanisms like detoxification that protect the DNA from oxidation. The use of the comet assay for measuring repair of oxidative DNA damage was recently reviewed [92,93].

5.3.4. Host-cell reactivation

The host-cell reactivation (HCR) assay is based on the fact that expression of a damaged gene carried on a plasmid is lower than an intact plasmid. A plasmid carrying a reporter gene is treated with a DNA damaging agent, after which it is used to transfect growing cells in culture. Expressions levels are measured, and taken to represent the efficiency of DNA repair in the cells [94]. Using this assay with stimulated PBMC it was shown, for example, that the repair of benzo[a]pyrene-induced DNA damage was associated with increased risk of lung cancer [95]. The repair of oxidative damage was studies using the HCR assay in cultured cells [96,97], but not in cancer risk studies. Its limitations are that gene expression rather than DNA repair is measured, and being based on transfection of growing cells it is rather complex for large scale epidemiology studies. In addition, complications may arise when assaying lesions that do not strongly affect transcription.

5.3.5 Mutagen sensitivity

The mutagen sensitivity assay involves measuring the number of chromatid breaks in cultured peripheral blood lymphocytes that were exposed to a mutagen, typically bleomycin. It is based on the assumption that the number of breaks is correlated to the DNA repair capacity of the cells. Perhaps a better correlation of mutagen sensitivity would be to the ability to cope with DNA damage, of which DNA repair is only one component. Other processes that may affect mutagen sensitivity include detoxification, cell cycle and checkpoints, and apoptosis. Mutagen sensitivity was recently comprehensively reviewed [98].

6. Application to cancer prevention

The importance of elucidating the role of oxidative DNA-damage repair in cancer is in its potential to provide biomarkers for cancer risk assessment, early detection, and prognosis. Since cancer risk, development, and treatment are all affected by a multiplicity of factors, a critical issue is whether successful clinical implementation will have to rely on a collection of a large numbers of biomarkers, or whether there might exist cases of ‘bottleneck’ processes, for which a small numbers of biomarkers may suffice. The usage of enzymatic activity as a biomarker has the advantage that it integrates many (dozens, or even more) factors that are known to affect protein expression and activity, including genetic polymorphism, epigenetic effects, transcription factors, splicing, RNA stability, translation factors, protein stability, post-translational modification, and perhaps environmental and life style factors [68,77]. The repair of 8-oxoguanine might in fact be a bottleneck pathway in the development of lung and head and neck cancers, and perhaps other types of cancer [69,72]. This is indicated by the significant estimated relative risk to develop lung cancer of smokers that have low OGG activity (Fig. 2). Such large odds ratios suggest that screening smokers for OGG activity may be used as a powerful tool for cancer prevention. Smokers who are diagnosed with low OGG activity can be advised to enter smoking cessation programs, as a way to reduce cancer risk. Such an approach based on personal susceptibility is expected to be more effective than a general warning on the hazards of smoking, as has been seen in the case of personal risk factors for cardiovascular diseases (e.g., personal cholesterol levels) [99]. Lung cancer prevention programs based on biomarkers such as OGG activity, in combination with other risk factors, may provide an effective mechanism for large-scale efforts to reduce the incidence of lung cancer. In addition, it would be interesting to examine whether supplemental dietary anti-oxidants may have a beneficial anti-cancer effect in individuals with a low ability to repair oxidative DNA damage [100-102]. Anti-oxidants did not protect individuals from lung cancer in a large prospective study [103]. However, most individuals appear to repair very well oxidative DNA damage, and therefore extra protection has apparently no added value. The situation may be different for individuals with a low ability to repair oxidative DNA damage, where anti-oxidants may reduce the amount of DNA damage, thereby easing up the repair load of the weak DNA repair system, and perhaps leading to a reduced cancer risk.

Abbreviations

BER

base excision repair

95% CI

95% confidence interval

HCR

host cell reactivation

NER

nucleotide excision repair

NSCLC

non-small cell lung cancer

PBMC

peripheral blood mononuclear cells

ROS

reactive oxygen species

SNP

single nucleotide polymorphism