8-oxoguanine DNA glycosylase (OGG1) is a bifunctional DNA glycosylase that removes oxidized bases such as 8-oxo-G, 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FaPyG) and 7,8-dihydro-8-oxoadenine (8-oxo-A) from the DNA ([13–22], and reviewed in [2]). Importantly, while OGG1 removes 8-oxo-G and FaPyG when base pared to a natural cytosine (C), 8-oxo-A will not be removed when paired to native thymine (T). OGG1 is expressed in at least four different splice forms in mammalian cells, of which at least two contribute differentially to BER pathways removing 8-oxo-G from nuclear (nDNA) and mitochondrial DNA (mtDNA) [23]. By catalyzing the excision of an oxidatively damaged base, OGG1 initiates a canonical SP-BER pathway that involves the action of APE1, Pol β and XRCC1/DNA ligase III to reconstitute the original intact base pair.

It is known that oxidative DNA damage plays a role in the process of ageing. OGG1 as one of the main regulators of 8-oxo-G levels in the genome, was shown to be widely expressed and active in human as well as rodent brains [24]. Initial transient decrease in OGG1 expression directly upon birth of mice, was followed by an increase after 8 weeks of age. Along this line, using a Comet-assay analysis of DNA damage in isolated neurons and astrocytes from the cortex of young (7 days), adult (6 months) and old (2 years) rats, Swain et al. revealed an age-dependent increase in the number of OGG1-sensitive sites, accompanied by a decrease in the OGG1 activity [25]. Further, by testing the activity of neuronal extracts from rat cerebral cortices in an in vitro assay with synthetic oligonucleotide duplexes, the authors observe a marked decline of 8-oxo-G repair capacity with age [26]. The decline could be attributed to a decrease in the expression levels of OGG1 and other BER enzymes including APE1 and Pol β. Supplementation of the neuronal extracts with the reduced components individually did not result in rescuing of the BER activity, suggesting that the age-dependent decline was not a result of an overall deficiency in the single DNA repair factors. However, addition of OGG1 together with Pol β and T4 DNA ligase markedly improved the BER activity and thus suggested that several BER proteins are limiting factors in adult and old neurons.

Acetylation of OGG1 has been shown to promote its enzymatic activity up to 10-fold in vitro[27]. Analysis of the OGG1 acetylation status in brain neurons of young rats revealed an increase in the acetylated form of OGG1 associated with either exercise or insulin-like growth factor-1 (IGF-1) treatment, a factor known to enhance neurogenesis [28]. In contrast, total OGG1 levels, as well as the amount of the acetylated OGG1 form, decreased with age in rats and correspondingly 8-oxo-G levels increased. The age-associated decrease in neurogenesis was possible to attenuate with exercise and IGF-1 treatment; at the same time exercise also improved the spatial memory, while IGF-1 treatment inhibited this process. These findings could potentially underline a role of oxidative DNA damage in age-related neuropathologies. Ogonovszky et al. further showed that neither were the levels of 8-oxo-G nor the OGG1 activity altered by exercise training in rats, suggesting that over-training does not induce oxidative stress in the brain and does not cause loss of memory [29].

Besides investigating the impact of the age-related decline on cellular level, studies of the mtDNA repair in particular revealed an association between DNA damage levels in the mitochondrial genome and different brain regions. By determining the mtDNA repair status in the central auditory system using a rat model of D-galactose-induced aging, Chen et al. observed a significant age-associated increase in mtDNA 4834 base pairs (bp) deletions and the number of terminal deoxynucleotidyl transferase–mediated uridine 5′-triphosphate-biotin nick end-labeling (TUNEL)-positive cells, a marker for apoptosis [30]. Interestingly, expression of Pol γ, the major mitochondrial Pol, and OGG1 were remarkably down-regulated in the auditory cortex. Thus, potentially indicating that during aging, increased mtDNA damage likely resulted from a decrease in its DNA repair capacity. These findings are supported by the work of Gredilla et al. addressing the efficiency of BER throughout the murine lifespan in mitochondria from cortex and hippocampus, both of which are regions severely affected during aging and in neurodegenerative diseases [31]. OGG1 activity peaked at middle-age in cortical mitochondria, followed by a significant drop at old age. However, only minor changes were observed in hippocampal mitochondria during the whole lifespan of the animals. Furthermore, OGG1 activity was lower in hippocampal than in cortical mitochondria. Taken together, these data suggest an important region-specific regulation of mitochondrial BER during aging.

The expression of OGG1 can also be modulated by many exogenous compounds, as shown by several studies discussed in the following. Cigarette smoke was found to induce DNA damage, as well as to alter OGG1 activity and distribution in several regions of the brain in neonatal mice; underlining the importance of cigarette smoke as risk factor for neurodevelopmental, as well as neurodegenerative disorders [32]. Fenvalerate is a synthetic pyrethroid widely used as pesticide in agriculture in developing countries and acts as neurotoxic compound in adults. To investigate the potential toxicity of fenvalerate to developing organisms, Gu et al. treated zebrafish larvae with this pesticide and found that OGG1 expression was down-regulated in a concentration-dependent manner. Fenvalerate also caused brain impairment during zebrafish development, further underlining the toxic nature of the compound especially during development [33]. Another pesticide, the organochlorine dieldrin, is a known neurotoxicant ubiquitously distributed in the environment and is toxic for dopaminergic neurons in vitro. Dieldrin slightly up-regulated OGG1 activity in proliferating PC12 cells, while the 8-oxo-G levels remained unchanged [34]. Differentiated PC12 cells on the other hand showed a longer lasting decline in OGG1 activity and a concomitant increase in 8-oxo-G levels. The differences between proliferating and differentiated cells might explain at least in part the vulnerability of post-mitotic neurons to oxidative stress and neurotoxins. A study that analyzed the impact of developmental exposure to lead (Pb), a known inducer of oxidative stress in the brain, found that cerebral 8-oxo-G levels were only transiently modulated early in life, at postnatal day 5, but were markedly elevated 20 months after the exposure had ceased [35]. OGG1 activity itself was not altered by developmental Pb exposure, resulting in loss of the age-dependent inverse correlation between OGG1 activity and 8-oxo-G accumulation. Exposure to Pb in old age did not have an impact on 8-oxo-G levels, suggesting that age-related oxidative damage accumulation and neurodegeneration could be markedly influenced by developmental disturbances.

Along the line of OGG1 involvement in development, OGG1 was found to be important for the determination of neural stem cell (NSC) differentiation by repairing mtDNA damage in differentiating neural cells [36]. NSCs derived from OGG1 knockout (ko) mice spontaneously accumulated mtDNA damage and shifted their differentiation direction toward an astrocytic lineage. A similar phenotype was observed when wild-type (wt) NSCs were subjected to mtDNA damaging insults, thus suggesting that mtDNA damage might be one of the primary signals for elevated astrogliosis and the lack of neurogenesis, a phenomenon observed after neuronal injury. Another study also demonstrated that small interfering RNA (siRNA) mediated knockdown (kd) of the DNA glycosylases OGG1 and endonuclease VIII (Nei)-like protein (NEIL) 3 decreased the differentiation ability of NSC, resulting in a decline of both neuronal and astrocytic gene expression after mitogen withdrawal, as well as a decrement in the stem cell marker Musashi-1 [37]. This suggests that OGG1 plays a role in governing essential NSC characteristics.

Besides the above-described impacts on development and ageing, alterations in OGG1 have been associated with numerous neurodegenerative disorders as elaborated bellow.

MUTYH (also sometimes called MUTY or MYH) is, like OGG1, a DNA glycosylase of the helix-hairpin-helix (HhH) family. It mediates the removal of adenine (A) paired with an 8-oxo-G [76,77], a situation that arises when replicative Pols bypass 8-oxo-G in an inaccurate manner by inserting a wrong A instead of a correct C. With this action MUTYH gives rise to a novel BER pathway involving Pol λ that reconstitutes the correct C:8-oxo-G base pair, which is then a substrate for OGG1 ([78–80], and reviewed in [2]); Several nuclear as well as mitochondrial isoforms of MUTYH are present in mammalian cells [81]. Biallelic mutations in MUTYH predispose to a familial adenomatous polyposis variant called MUTYH-associated polyposis (MAP) [82]. However, no evidence of increased risk for cancers of the brain tissue has been found in MAP patients [83].

Up to date, a limited number of insights have been obtained regarding the potential roles of MUTYH in the brain. A study using MUTYH ko mice showed that there was no time-dependent accumulation of 8-oxo-G in brain tissue [84]. Interestingly, Lee et al. showed that levels of one embryonic isoform of MUTYH could be detected in rat brains at the E14 embryonic stage, after which it decreased during embryonic and neonatal development, while new isoforms appeared and gradually increased in the neonate and adult brain [85]. It seemed that during embryonic development expression levels of MUTYH followed the expression profile of proliferating cell nuclear antigen (PCNA). In addition, these proteins also colocalized in the nucleus. At later time points, when the levels of PCNA declined, MUTYH was detected primarily outside the nucleus. An activity for excision of A opposite 8-oxo-G was detected in all the extracts. Even though the authors suggested that MUTYH might be primarily involved in post-replicative repair of nDNA, it is possible that MUTYH might rather be involved in repair of mtDNA in post-mitotic neurons.

Though not much is known about the detailed role of MUTYH in the brain DDR, alterations in the MUTYH homeostasis have been associated with various neurodegenerative diseases, such as PD.

The methyl-CpG binding domain protein MBD4 (also known as MED1) is a DNA glycosylase that belongs to the MBD protein family within the HhH domain superfamily. It processes a wide substrate range of DNA base lesions mispaired with guanine (G), such as uracil (U), 5-fluorouracil (5-FU), 3,N⁴-ethenocytosine (ɛC) and T ([93–95], and reviewed in [7]), Mbd4 ko mice are viable and show no developmental defects [96,97]. Though lack of MBD4 does not lead to defects in mice, it has been found that MBD4 mRNA levels are significantly up-regulated in the hippocampus of both schizophrenia and bipolar disorders suggesting a potential involvement of this glycosylase in human neurodegenerative diseases [98]. To understand the exact mechanism how MBD4 contributes to these disorders, future studies are needed.

Endonuclease III-like 1 (NTHL1, also known as NTH1) is a DNA glycosylase that belongs to the family of endonuclease III-like 1 proteins, a subfamily of HhH DNA glycosylases. It catalyzes excision of ring fragmented purines or oxidized pyrimidines like thymine glycol (Tg), 4,6-diamino-5-formamidopyrimidine (FaPyA), FaPyG, 5-hydroxycytosine (5-OHC) and 5-hydroxyuracil (5-OHU) when paired to G in double stranded DNA ([99–106], and reviewed in [7]). It is assumed that loss of NTHL1 function can be compensated for by NEIL glycosylases, because NTHL1 ko mice show no abnormalities [107,108] As was the case for NEIL1 and NEIL2, also no association for NTHL1 with the risk of developing multiple sclerosis was found [109]. So far, nothing more regarding a possible role of NTHL1 in neurodegenerative diseases is known.

NEIL1 is a DNA glycosylase that belongs to the family of endonuclease VIII (Nei)-like proteins. Its preferred substrates are damaged pyrimidines and purines, such as Tg, FaPyA, FaPyG and others, but also 8-oxo-G and 5-OHU in double stranded DNA and bubble structures ([102,103,105,108,110–119], and reviewed in [7,120]), NEIL1 ko mice display a phenotype very close to the metabolic syndrome, and harbor increased levels of DNA damage in their mtDNA [121]. NEIL1 mRNA has been detected in different mammalian tissues including the brain and both its mRNA and protein levels were shown to increase during S-phase [102]. Further, widespread NEIL1 expression was reported at all ages in mice and it even increased with age, as did FaPyG lesion (induced by treatment of DNA with N-[³H]methyl-N′-nitrosourea) excision activity in all brain regions tested [24]. By investigating the efficiency of mitochondrial BER during the murine lifespan in the cortex and hippocampus, Gredilla et al. found that, similarly to OGG1; NEIL1 activity reached its maximum at middle-age in cortical mitochondria followed by a significant drop at old age, while only minor changes were observed in hippocampal mitochondria [31]. In addition, NEIL1 DNA glycosylase activity was lower in hippocampal than in cortical mitochondria. These findings indicate that regulation of mitochondrial NEIL1 activity in the brain is region and age specific.

Among other BER enzymes, Rolseth et al. investigated the impact of OGD on NEIL1 activity and gene expression levels in organotypic rat hippocampal slice cultures (particularly in the regions CA1 and CA3/fascia dentata) [62]. While base removal of U did not differ between the two hippocampal regions, removal of 5-OHU was slightly less efficient in CA3/FD than in CA1. After 30 min of OGD an increase in the activity on ɛA by approximately 25% could be detected in CA1, whereas activities for 8-oxo-G, 5-OHU and U remained unchanged. Later, 8 h after OGD, none of the enzyme activities differed from control values. As for OGG1, transcription of NEIL1 was not changed in response to OGD treatment at time point 0 h.

Englander et al. measured the expression and activities of BER enzymes during brain development where the physiological transition of neuronal cells from the proliferative to the post-mitotic differentiated state takes place [122]. Expression of NEIL1 increased during brain development concomitant with maintenance of the capacity for excision of 5-OHU from bubble structured DNA in the mature rat brain, suggesting a potential role of NEIL1 in the maintenance of the integrity of transcribed DNA in the post-mitotic brain. A recent study by Canugovi et al. demonstrates that NEIL1 ko mice exhibit an impairment in memory retention, as assessed by a water maze test [123]. However, these mice did not display abnormalities in motor performance, anxiety or fear conditioning. Furthermore, the deficiency in NEIL1 results in an increase in brain damage after ischemia/reperfusion due to apoptosis. Also, in these mice the incision activity of 5-OHU in a bubble structure was lower in the ipsilateral sides of ischemic brains as well as in mitochondrial lysates of unstressed old ko mice, suggesting that NEIL1 is a central player in learning, memory and neuronal protection against ischemia.

Like NEIL1, NEIL2 belongs to the family of endonuclease VIII (Nei)-like proteins. Its preferred substrates are strongly overlapping with the ones of NEIL1 and include oxidized pyrimidines, such as Tg, 5-OHU, 5-OHC, 5,6 dihydrothymine and 5,6 dihydrouracil in double stranded DNA and bubble structures ([111,119,125], and reviewed in [7,120]). NEIL2 mRNA has been detected in the brain, but unlike NEIL1, the expression of NEIL2 was independent of the cell cycle stage [111]. Analysis of the distribution patterns in mouse brains showed widespread expression of NEIL2 at all ages, and the excision activity of chemically induced FaPyG lesions increased with age in all brain regions tested [24].

Rolseth et al. found that transcription of NEIL2 in two different regions of the hippocampus was not changed in response to OGD treatment at time point 0 h [62]. As was the case for NEIL1, expression of NEIL2 levels increased during the physiological transition of neuronal cells from the proliferative to the post-mitotic differentiated state in brain development [122]. This was concomitant with the maintenance of the capacity for excision of 5-OHU from bubble structured DNA in the mature rat brain, suggesting a role for NEIL1 and NEIL2 in the maintenance of the integrity of transcribed DNA in the post-mitotic brain. Similarly to NEIL1, no association between NEIL2 and the risk of developing multiple sclerosis was found [109]. Future studies are needed to completely understand if and how NEIL2 could be associated with different neurodegenerative diseases.

Similarly to NEIL1 and NEIL2, NEIL3 also belongs to the family of endonuclease VIII (Nei)-like proteins. In contrast to the two former glycosylases, NEIL3 excises FaPyG and FaPyA lesions but is inactive on 8-oxo-G ([126,127] and reviewed in [7]). Additionally, the mouse ortholog was shown to remove a broad spectrum of DNA base lesions on single-stranded DNA substrates, including secondary oxidation products of 8-oxo-G, such as spiroiminodihydantoin and guanidinohydantoin, suggesting that NEIL3 prevents the accumulation of these cytotoxic and mutagenic lesions in mammalian cells [127]. Though NEIL3 ko mice are viable and fertile, NEIL3 has been implicated to play a role in hematopoiesis or the immune system, since it is preferentially expressed in hematopoietic tissues [128]. In brains of newborn mice, NEIL3 revealed a discrete expression pattern in the subventricular zone, the rostral migratory stream, and the hilar region of the hippocampal formation, all of which are brain regions known to harbor stem cell populations [24]. Expression of NEIL3 decreased with age, and in brains of old mice it could be only detected in layer V of the neocortex. The distribution of NEIL3 thus indicates a potentially specific role of this enzyme in stem cell differentiation. Along with this study, expression pattern analysis of NEIL3 in the brain during mouse embryonic development revealed a tight regulation at both temporal and spatial levels. High expression of NEIL3 was observed at embryonic days 12–13, which coincides with the start of neurogenesis [129]. Subsequently, the expression of NEIL3 decreased gradually, and it could not be detected anymore in adult brains by RT-qPCR. Interestingly, expression during embryogenesis and in newborn mice was observed in areas with neural stem and progenitor cells, such as the subventricular zone and the dentate gyrus, suggesting that brain areas with neurogenesis and a high proliferative potential specifically express NEIL3. Subsequently, Sejersted et al. demonstrated a profound neuropathology in NEIL3 ko mice, which was characterized by a reduced number of microglia and a loss of proliferating neuronal progenitors in the striatum after hypoxia-ischemia [130]. Furthermore, NEIL3 ko neural stem/progenitor cells displayed an inability to increase neurogenesis and a reduced capacity to repair oxidized base lesions in single stranded DNA, indicating that NEIL3 could occupy a highly specialized role to accurately repair DNA in rapidly proliferating cells. Another study also demonstrated that, similarly to OGG1, siRNA-mediated kd of NEIL3 decreased NSC differentiation ability, resulting in a decrease of both neuronal and astrocytic gene expression after mitogen withdrawal as well as a decrease in the stem cell marker Musashi-1 [37]. Furthermore, a deficiency in NEIL3 led to a decrease in cell proliferation along with an increase in heterochromatin protein 1γ immunoreactivity, a sign of premature senescence, while cell survival remained unaffected. This potentially suggests that OGG1 and NEIL3 play a role in governing essential neural stem cell characteristics.

AAG mouse models, such as AAG ko and AAG transgenic (AAG-Tg) mice, provided valuable tools to study the influence of AAG-initiated BER on brain development and neurodegeneration [138–141]. Treatment with the alkylating agents methylazoxymethanol (MAM) and methyl methanesulfonate (MMS) induced extreme cerebellar toxicity and dramatically impaired motor function in AAG-Tg mice, while these effects were suppressed in AAG ko animals [131,142]. These findings support the idea that AAG activity, induced by alkylation treatment, promotes accumulation of toxic BER intermediates, while loss of AAG prevents their formation, thus ensuring resistance. Though several lines of evidence strongly indicate that a lack of AAG-initiated BER prevents induction of alkylation induced cell death in different tissues [131,138,142,143], it is important to note that the impact of BER absence on cellular survival largely depends on the type of DNA lesions induced by alkylating treatment, as well as the affected cell type. One such example is the treatment of neuronal and astrocyte cell cultures, obtained from the cerebellum of wt or Aag ko mice, with either chloroacetaldehyde (CAA) or the alkylating agent 3-methyllexitropsin (Me-Lex). Treatment with both CAA and Me-Lex resulted in increased sensitivity of AAG ko neurons, while the sensitivity of AAG ko astrocytes did not differ from the wt cells [144]. Present studies clearly imply an essential and specific role of AAG-mediated BER in alkylation-mediated neurodegeneration.

The UDG family in eukaryotes can be divided into three subfamilies: the uracil N-glycosylase (UNG), the single-strand-specific monofunctional uracil DNA glycosylases (SMUGs) and the mismatch-specific uracil DNA glycosylases (MUGs). Although members of the UDG family have very diverse amino-acid sequences, they share a common alpha-beta fold present in the catalytic active site. Humans and mice have two different UNG isoforms, UNG1 and UNG2 localized in the mitochondria and the nucleus, respectively [145]. UNG protein levels vary in different tissues and cell types. In the human brain UNG levels are (i) extremely high in cerebellar Purkinje cells and neuronal cells of the cerebral cortex; (ii) moderate expression is observed in neuronal cells of the lateral ventricle, and in cerebellar cells of the granular and molecular layer as well as in glial cells; while (iii) very low levels or no UNG is detected in glial cells of both the hippocampus and the lateral ventricle [137]. Several studies clearly demonstrated variations in the activity and levels of UNG with age [25,26,31,146,147]. Neuronal extracts prepared from the cerebral cortex of young (7 days), adult (180 days) and old (720 days) rats showed a dramatic decrease with age in the ability to remove U from the DNA [26]. Supplementation of these extracts with recombinant purified UNG, Pol β and T₄ DNA ligase significantly restored the loss of BER in aging neurons [26]. Single cell gel electrophoresis experiments of neurons and astrocytes from the cortex of young, adult and old rats revealed a marked increase in the number of UNG sensitive sites with age; further indicating age-dependent decrease in UNG activity [25]. Additionally, analysis of nuclear and mitochondrial UNG activities in different brain regions (the caudate nucleus, frontal cortex, hippocampus, cerebellum and brain stem) of young and adult mice revealed an age-dependent decrease in mitochondrial UNG-initiated BER [146]. In contrast to mitochondria, no region- or age-specific differences were detectable in the UNG nuclear activity, with exception of the cerebellum where uracil incision capacity was reduced with age [146]. Gredilla et al. similarly reported reduced UNG1 action in cortical mitochondria, however they did not detect any age-dependent change in the U removal ability in hippocampal mitochondria, while in cerebellar mitochondria UNG1 activity reached its maximum at old age [31]. Taken together, present findings clearly indicate an important role of UNG in different brain regions and suggest that an age-dependent increase in damage to mtDNA might contribute to the normal aging process.

The thymine DNA glycosylase (TDG) is a member of the MUG subfamily. Besides T, TDG recognizes and excises U, 5-FU, ɛC, 5-hydroxymethyluracil, 5-formylcytosine and 5-carboxylcytosine when base paired with G [155–163]. TDG ko in mice is embryonically lethal, suggesting an essential function of TDG during development [164,165]. While the expression of TDG in human brain is documented in detail, studies in rat brains indicated that similar to UNG, TDG levels were inversely correlated with age [166,167]. Future studies are needed to reveal a role of this unique glycosylase in brain integrity and consequentially in neurodegeneration.

So far not much is known about the role of the remaining two members of the UDG family in brain development and homeostasis.

A study of the APE1 hippocampal expression in human AD brains revealed elevated APE1 levels both in senile plaques, a histopathological hallmark of AD, and in injured neurons [182]. This increase was further found to be localized to the nuclear fractions of AD brains [183], which was confirmed in a immunohistochemical analysis of the cerebral cortex, where an intensive nuclear APE1 signal in all cortical layers was detected [184]. Aβ, a major contributor to AD development, is known to induce oxidative stress in neurons [185]. Tan et al. investigated the impact of various Aβ concentrations on APE1 levels and cell survival in isolated rat hippocampal neurons [186]. Interestingly, treatment with high concentrations of Aβ (5 μM) caused a reduction in cellular APE1 levels and activity, which correlated with extensive neuronal degeneration of the cultured hippocampal neurons. In contrast, lower concentrations of Aβ (1 μM) induced APE1 expression and activity, resulting in no substantial loss of the neurons [186]. The cyclin-dependent kinase 5 (Cdk5) was shown to regulate APE1 through phosphorylation, leading to a reduction of its endonuclease activity [187]. Subsequent accumulation of DNA damage, together with the finding that levels of phosphorylated APE1 were increased in brain tissue from AD and PD patients, might implicate Cdk5-mediated APE1 phosphorylation in the development of these neurodegenerative disorders. Independently of the endonuclease activity, but through its redox function, APE1 was found to mediate neuroprotection against Aβ and H₂O₂ via induction of the glial cell-derived neurotropic factor (GDNF) receptor α1 transcription, thereby increasing the GDNF responsiveness [188]. In a very recent proteomic study, where neuronal cells were challenged with an Aβ peptide fragment (25–35), novel interaction partners of APE1 were identified [189]. Among them, (i) tropomodulin 3, involved in the synaptic activity; (ii) heterogeneous nuclear ribonucleoprotein-H1, a regulator of alternative splicing and (iii) the pyruvate kinase 3 isoform 2, a key enzyme in the glycolysis; all of these factors might have a functional relevance for neuronal cell survival and Aβ resistance. In addition, a potential association between the APE1-D148E SNP and the onset of AD was investigated, however no significant correlation was found [190].

Decreased APE1 levels were found in patient cells affected by Ataxia with Oculomotor Apraxia Type 1 (AOA1) [191], a neurodegenerative disorder originating in mutations of the APTX gene [192,193], which results in a cellular aprataxin deficiency [194]. Comparable findings were also obtained in ALS patients, where frontal cortical APE1 levels, as well as activity were significantly reduced [195], and in some cases missense mutations within the APE gene were identified [196]. ALS manifests in the progressive loss of motor neurons [197] and appears in a sporadic as well as a familial form [198]. For the sporadic form, a significant association with the D148E APE1 polymorphism was shown [199]. In contrast to the analysis of frontal cortical levels, a study by Shaikh et al. indicated increased APE1 levels in the spinal cord and motor cortex of ALS patients and showed that protein extracts from this tissue samples were more proficient in in vitro processing of AP sites [200]. Hyperactivity of APE1 potentially also contributes to the genomic instability by resulting in an increased number of extremely harmful DNA breaks.

In a rat model where epileptic-like seizures were induced by the application of kainic acid (KA), a subsequent induction of APE1 expression was observed in KA-vulnerable brain regions (CA1, CA3 and hilar subregions of hippocampus, pyriform cortex, amygdala and thalamus) [201]. Furthermore, APE1 colocalized with the BER protein XRCC1, the oxidative DNA damage marker 8-oxo-G, the tumor suppressor p53 and also with fragmented DNA, as assessed by TUNEL staining [201]. These findings thus indicate that BER is activated but not sufficient to counteract excitotoxicity-mediated neuronal cell death.

A cold injury-induced brain trauma (CIBT) mouse model revealed an early post-traumatic decrease of APE1 levels within the lesion, which preceded later DNA fragmentation [202]. Similar observations were made after severe traumatic cortical brain injury [203]. However, the outer boundary area that survived CIBT showed a significant increase in APE1 immunoreactivity [202]. Transient focal cerebral ischemia (FCI) [204] or a defined hypoxic-ischemic insult [205] resulted in decreased APE1 protein levels, a reduction exclusively detected in the hippocampus. In addition, APE1 levels selectively decreased in the hippocampal CA1 neurons 2 days after transient global cerebral ischemia (GCI), which was followed by DNA fragmentation after 3 days [206]. Intra cerebral application of the pituitary adenylate cyclase-activating polypeptide (PACAP) in the context of transient GCI reversed the effect, by inducing APE1 expression in hippocampal CA1 neurons, which correlated with improved cell survival [207]. This neuroprotective effect of PACAP was dependent on the DNA repair activity of APE1, as was shown through a loss-of-function rescue attempt of APE1 deficient cells, by overexpressing DNA repair-incompetent APE1. Upon transient spinal cord ischemia, spinal APE1 levels decreased while oxidative DNA damage increased [208]. Interestingly, an ischemic tolerance could be established by sub-lethal ischemic preconditioning, which resulted in subsequent up-regulation of APE1 levels and other BER proteins and therefore better neuroprotection in the case of severe ischemia [63]. On the other hand, APE1 overexpression was shown to increase cell viability of cultured hippocampal and sensory neurons after ionizing radiation-induced DNA damage [209]. Glutamate-induced oxidative DNA damage was found to cause an increase in APE1 expression in rat cerebral cortical neurons via a pathway involving the cAMP-response element-binding protein, thereby improving the DNA repair activity of oxidized lesions [210].

In summary, the multifunctional enzyme APE1 is implicated in a broad spectrum of neuropathologies via both, its endonuclease and redox activity. However, the exact regulation of APE1 in this context and the underlying mechanisms remain to be investigated.

A recent study by Shen et al. linked four mutations in the PNKP gene to an autosomal recessive disease characterized by microcephaly, early-onset, severe seizures and developmental delay (MCSZ) [216]. The mutations were either frame-shift mutations (T424Gfs48X and exon15Δfs4X) due to small deletions or duplications within the kinase domain, resulting in truncated PNK proteins or point mutations (L176F and E326K) within the phosphatase domain. Assessing the DNA repair capacity by comet assay in patient-derived lymphocytes after H₂O₂ or camptothecin treatment, revealed a significant impairment in the case of MCSZ cells compare to healthy controls and in addition, PNK protein levels were found reduced in these patient cells[216]. Interestingly, a further study showed in vitro that PNK^T424Gfs48X, PNK^exon15Δfs4X and PNK^L176F mutant proteins exhibit a markedly decreased 5′-DNA kinase activity [217]. However, a moderate reduction in the 3′-DNA phosphatase activity was only observed in the case of recombinant PNK^L176F. Analysis of phosphatase and kinase activities in MCSZ patient cells with decreased mutant PNK levels [216], revealed an overall marked reduction [2,17]. Furthermore, alkaline comet assay analysis of cells from affected individuals exhibiting the PNK^T424Gfs48X and PNK^exon15Δfs4X mutations as well as of cells homozygous for PNK^E326K, revealed inefficient repair of DNA strand breaks upon γ-irradiation, whereas camptothecin treatment only led to an accumulation of DNA damage in cells containing both, mutated PNK^T424Gfs48X and PNK^exon15Δfs4X[217]. Even though PNK appears to be crucial for proper neurodevelopment, further mechanistical studies will be needed to unravel the potential link between impaired DNA repair capacity and the development of MCSZ.

AD cell culture models demonstrated that the application of Aβ to cortical neurons induces cell cycle reentrance followed by DNA replication resulting in reflective apoptosis of these cells [244]. Pol β was shown to be a leading Pol mediating de novo DNA synthesis after Aβ-induced cell cycle reentrance and thereby to play a role in the neuronal loss [245]. Consequently, in Aβ-treated cortical neurons, Pol β was also found to co-immunoprecipitate with cell division cycle 45 (Cdc45) and the DNA primase in nucleoprotein fragments, implicating its association with the DNA replication fork [246]. Aβ treatment of neural progenitor cells induced Pol β expression, with the consequence of cell differentiation along the neuronal lineage [247]. However, Weissman et al. showed overall decreased Pol β protein levels in AD brain tissue accompanied by reduced single nucleotide gap-filling activity [154]. Cell cycle re-entry was also observed in cultured cerebellar granule cells treated with the neurotoxin 1-methyl-4-phenylpyridinium (MPP⁺), which is known to mimic PD by selective toxicity against dopaminergic neurons [248]. MPP⁺-mediated cell death was accompanied by increased Pol β expression. Interestingly, reduction of Pol β activity by either inhibition with dideoxycitidine or the expression of a dominant negative Pol β variant in these cells attenuated the neuronal loss [248].

Several studies addressed the role of Pol β in the CAG triplet repeat expansion associated with HD. Kovtun et al. showed in an in vitro OGG1-initiated BER assay, that Pol β tends to perform strand displacement DNA synthesis within CAG repeats instead of single-nucleotide incorporation, resulting in longer DNA products [53]. Thus indicating that the initiation of BER by OGG1 in a CAG sequence contributes to the TNR expansion by Pol β-mediated strand displacement. Supporting this finding, a further study showed that multinucleotide incorporation by Pol β results in strand displacement and the formation of CAG hairpins that become stabilized and promote repeat expansion [249]. Pol β was also shown to accumulate along CAG repeats in the striatum of HD mice, the brain region most susceptible to degeneration in HD patients (for more details see Chapter 3.6.) [57]. Furthermore, Goula et al. recently determined the protein levels of the major BER proteins in the striatum, as well as the HD-spared cerebellum of HD transgenic mice [250]. While it was previously shown that Pol β protein levels were not significantly changed in these tissues, DNA ligase I, FEN1, APE1 and XRCC1 levels were increased by at least 2-fold in cerebellum [57,250]. In addition, in vitro repair of AP-sites with either (i) purified BER proteins in the striatum-specific stoichiometry or (ii) with wt and HD mice striatum extracts, was shown to be less proficient than cerebellar repair regardless of the sequence context [250]. The same study demonstrated that lesions within CAG repeats tend to be repaired via LP-BER and that this process is more efficient under cerebellar conditions compared to the striatal ones. Moreover, a lesion closer to the 3′ end of the repeat sequence was more efficiently repaired than a 5′ lesion, most probably due to the fact that the formation of stable hairpins becomes impaired the closer a lesion is to the 3′ end of the repeat tract [250]. Taken together these findings suggest a potentially important role of Pol β mediated repair in the onset of triplet repeat expansion diseases, such as HD.

Pol β activity was observed to be up-regulated in cerebral cortical neurons of newborn piglets exposed to hypoxia, and this was suggested to be beneficial to reduce hypoxia-induced DNA damage [251]. Upon transient FCI in rats, an increase in markers for oxidative DNA damage, namely 8-oxo-G and AP sites could be detected [252]. These lesions were efficiently repaired during reperfusion in the surviving cortex, which was proposed to be at least partially due to a protective long-lasting up-regulation of Pol β expression as well as its activity. As already described for APE1, ischemic preconditioning also elevated the protein levels of Pol β as well as XRCC1 and DNA ligase III [253]. This up-regulation in particular of SP-BER proteins was implicated to play a pivotal role in the neuroprotection observed in subsequent severe ischemic episodes.

Summarizing, Pol β was shown to be crucial for proper neurodevelopment, whereas later on Pol β levels decline in an age-dependent manner. In the context with AD, spurious de novo DNA synthesis contributes to neuronal loss. However, upon hypoxic insults, Pol β activity mediates neuroprotection.

The discovery of XRCC1 polymorphisms gave raise to studies investigating a potential predisposition to diseases [274]. AD belongs to the most common diagnosed age-related neurodegenerative disorders resulting in dementia and it is often histologically characterized by the presence of Aβ plaques, neurofibrillary tangles and severe neuronal loss (reviewed in [275]). Since the R194W XRCC1 polymorphism lies within a conserved amino acid residue sequence [274], its potential functional relevance for neurodegeneration was addressed in a case-control study focusing on the late-onset AD in a Turkish population [276]. An increased risk for late-onset AD in the presence of the R194W polymorphism was proposed [276]. However, the risk estimates did not reach a statistically significant level and were not confirmed in a later comparable study among Han Chinese people [277]. It therefore remains under debate, whether the R194W XRCC1 polymorphism promotes the onset of AD. The analysis of two additional XRCC1 polymorphisms, namely R280H and R399Q revealed no dependency between allele frequency and the onset of AD [190]. However, an increased risk was found for the development of PD in the presence of XRCC1 R399Q [278].

XRCC1 is not only implicated in chronic neurodegeneration, but also in neuronal loss after acute central nervous system injuries, including the cerebral ischemia-reperfusion phenomenon [279]. It is well established, that brain ischemia/reperfusion, a condition that resembles a stroke, goes along with an excessive production of ROS (reviewed [280,281]). In mice subjected to transient FCI by occlusion of the MCA, it was shown that XRCC1 levels decreased shortly after reperfusion and remained low until 24 h in the total MCA territory [279]. The loss of XRCC1 coincided with a positive TUNEL staining, corresponding to fragmented DNA arising 24 h later. A CIBT mouse model further supported this observation, where a similar early decline in XRCC1 levels within the injured region preceded later DNA fragmentation [282]. Thus, these studies indicated a potential role of the early XRCC1 decrease in the DNA damage-mediated neuronal cell death in traumatic brain regions [279,282].

Taken together, the scaffold protein XRCC1 was shown to be particularly important for DNA repair in post-mitotic neurons. Further association with the Down’s syndrome, seizure episodes and ischemic insults render XRCC1 a critical player in CNS homeostasis.

In the severe neurodegenerative disorder HD, BER initiated by OGG1 has already been implicated to contribute to the CAG trinucleotide expansion in somatic cells, as discussed above [53]. In addition, a study in HD mice showed, that oxidative damage specifically accumulates along CAG repeats in a length-dependent manner [57]. This event is specifically taking place in the striatum, the brain region most prone to degeneration in HD patients, when compared to the disease-spared cerebellum. Even though gap-filling activity is reduced in the striatum several factors, such as: (i) pronounced accumulation of Pol β at CAG repeats; (ii) promoted Pol β-mediated strand displacement activity; and (iii) low 5′-flap endonuclease activity by FEN1, contribute to the somatic instability in the context of LP-BER [57]. This is further in line with model of HD which suggests that the Pol β generated displaced strand, when not efficiently removed by FEN1, forms a hairpin structure that can become stably integrated leading to trinucleotide expansion. The stoichiometry between BER proteins seems to be an important factor for tissue-specific trinucleotide expansion-vulnerability [57]. Elevated levels of FEN1 in the cerebellum were implicated to significantly contribute to the increased BER efficiency on CAG substrates observed by measuring the repair capacity with BER proteins in the cerebellar stoichiometry (discussed in more detail under 3.3.) [57,250]. This effect was not observed when BER efficiency was addressed under the striatum-specific conditions [250]. Liu et al. similarly reported that FEN1 might facilitate TNR expansion, however through a slightly different mechanism by alternate cleavage of hairpin structures arising during Pol β multinucleotide DNA synthesis and strand displacement during LP-BER [249]. As FEN1 is unable to process 3′ ends of stable hairpins, it rather tends to cleave lose 5′-flaps of the hairpin, thereby providing ligatable nicks and potential TNR expansion. A comprehensive overview of the FEN1 involvement in the TNR expansion is presented in a recent review by Liu and Wilson [290].

As already observed for other BER proteins, ischemic preconditioning resulted in an induction of the DNA ligase III in neuronal and glial cells [253]. This up-regulation of the BER pathways was further reflected in the increased BER activity in nuclear brain extracts from preconditioned animals [253].

In summary, even though not much is known so far about DNA ligases in the physiology and pathology of the CNS, initial hints point towards an involvement in distinct neurodegenerative disorders.