It is also important to consider the role of glial cells in relation to oxidative DNA damage, neuroinflammation and neurodegeneration. When activated, microglia can produce several factors that are toxic to neurons, such as pro-inflammatory cytokines TNFα, PGE
2, and INFγ and ROS (NO, H
2O
2, O
2·−, NOO
−), in response to diverse stimuli, including neuronal damage, misfolded proteins, and environmental toxins [
227]. Microglial NOX2 is a major regulator of neurotoxicity by producing excessive ROS [
228]. Through a complex antioxidant response, astrocytes also enhance the decomposition and clearance of free radicals produced by neurons and other cell types in the CNS [
229]. Excessive free radicals can result in reactive astrogliosis, inducing neuroinflammation, which can lead to further oxidative stress [
229]. Senescence is also strongly linked to DNA damage, and it is also implicated as a potential driver of neuroinflammation in neurodegenerative diseases [
20].
AD
AD is the most common neurodegenerative disease [
235]. It is characterised clinically by progressive memory loss with neuropsychiatric symptoms due to the degeneration of cortical neurons in the entorhinal cortex and hippocampus [
236]. The pathological hallmarks of AD include the accumulation of cytoplasmic senile plaques composed of amyloid beta (Aβ) peptides (resulting from cleavage products of amyloid precursor protein) and the formation of neurofibrillary tangles (composed of hyper-phosphorylated tau) [
237,
238]. Most AD cases are sporadic in nature. However, 5%–10% cases are familial with a predominately autosomal dominance inheritance pattern, consistent with polygenic origins and multifactorial pathogenic disease processes [
239]. Several mechanisms have been implicated in the pathogenesis, including both oxidative stress and DNA damage [
236,
240].
There is direct evidence linking redox dysregulation with DNA damage in AD. Increased levels (twofold) of DNA strand breaks were observed in the cerebral cortex of AD brains [
241]. Higher levels of oxidative DNA damage, in the form of 8-OHG adducts and oxidized purine and pyrimidine bases, were detected in peripheral leukocytes of AD and mild cognitive impairment (MCI) patients compared to healthy controls [
242‐
244]. Similar observations were made in AD lymphocytes [
245]. Another study reported an age-dependent increase in the levels of 8-OHG in DNA in the cerebral cortex of AD patients. Consistent with this finding, elevated levels of 8-OHG, 8-hydroxyadenine (8-OHA) and 5-hydroxycytosine were detected in the total DNA of AD parietal lobe regions compared to matched controls [
245]. The same study also observed higher levels of thymine glycol, 5-hydroxyuracil, 4,6-diamino-5-formamido-pyrimidine (FapyAde), and FaPyGua in several AD brain areas. In another study, significantly higher levels of 8-OHG, 8-OHA and 5-OHU were detected in the temporal and parietal lobes of AD compared to control patients [
246]. Expression of adducts 8-OHG in RNA and 8-hydroxy-2’-deoxyguanosine (8-OHdG) in DNA, were also observed in late-stage AD compared to age-matched control patients [
247].
Significantly more aldehydic by-products HNE and acrolein were detected in late-stage AD brains and CSF, including in the most vulnerable areas (hippocampus and superior and middle temporal gyri) of MCI and early-stage AD brains [
248]. Moreover, higher levels of acrolein/guanosine adducts were also observed in the hippocampus of late-stage AD patients compared to controls [
249]. Mutations in the gene encoding OGG1 have been identified in AD patients, resulting in reduced enzymatic activity [
250]. Reduced levels of OGG are present in AD brains, implying that BER is defective in affected neurons [
251]. Consistent with this notion, defective BER, diminished activity of DNA glycosylase and reduced DNA synthesis by DNA polymerase β have been reported in AD tissues [
252].
More excision repair cross-complementing gene products have been also identified in AD brains compared to controls, suggesting that DNA repair pathways are activated to counteract increased oxidative damage [
253]. In addition, higher levels of SSBs and small increases in DSBs were observed in AD brains [
254]. In contrast, reduced DNA repair of SSBs [
255] or DSBs by DNA-PK-mediated NHEJ [
256] were reported in AD brains compared to controls. Similarly, significantly low levels of MRE11 DNA repair complex proteins were identified in the neocortex of AD brains [
257]. This would hamper the recognition of DNA damage and its subsequent repair, contributing to neuronal death in AD [
248].
Impaired SOD1 activity has also been detected in AD animal models and post-mortem AD brains [
258]. The expression of SOD1 and SOD2 is elevated in age-matched AD brain tissues compared to controls [
259]; however, the activity of both enzymes decreases significantly in the same tissues [
259]. Enhanced formation of Aβ plaques, neuroinflammation, tau phosphorylation, and consequent memory decline have also been observed in SOD1-deficient Tg2576 mice [
258,
260].
Oxidative imbalance and mitochondrial dysfunction are observed in AD [
261‐
263], together with oxidative stress-induced mtDNA damage. Significantly higher levels of 8-OHdG and 8-OHG were reported in AD brains compared to age-matched control samples [
264‐
266]. Another study analysing oxidized nucleosides revealed three-fold more oxidative damage in mtDNA in AD patients [
266]. In addition, sporadic mutations were detected in mtDNA of AD brain tissues [
267,
268]. Similarly, mutations in mtDNA in the blood of AD patients and in the lymphoblastoid lines derived from the blood of AD patients have been reported [
269].
Elevated levels of both PAR polymers and PARP-1 were detected in neurons of human AD brain tissues [
270,
271]. In addition, overexpression of PARP-1 is observed in AD brains, largely in the frontal and temporal lobes [
240], and the accumulation of Aβ peptides is preceded by oxidative stress and upregulation of PARP-1 in the hippocampus of adult rats [
272]. Similarly, a study in SHSY-5Y cells revealed that upregulation of PARP-1 induces pathological features of AD such as deposition of Aβ and the formation of tau tangles [
273]. Moreover, co-immunoreactivity of PARP/PAR with Aβ, tau and microtubule-associated protein 2 has been observed in human AD brain tissues [
274,
275]. p53 is increased in the temporal cortex of AD patients [
276,
277]. Expression of Aβ peptides triggers p53-mediated microglial apoptosis and microglial neurotoxicity [
278]. p53 is also prone to aggregate and is a component of misfolded aggregates in a tau mouse model and in human AD brains [
279]. Interestingly, p53-mediated DDR has been found to be impaired in AD [
279]. Taken together, these studies provide evidence that redox imbalance is associated with DNA damage and inefficient DNA repair, which together contribute to neurodegeneration in AD.
PD
PD is the second most common neurodegenerative disorder [
11]. It is characterised by the loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) accompanied by the formation of intraneuronal inclusions called Lewy bodies [
11,
280,
281]. The majority of PD cases (95%) are sporadic while only 5% of cases are linked to mutations in specific genes [
11]. Multiple lines of evidence implicate both oxidative stress and DNA damage as key mechanisms in disease pathogenesis [
282‐
285].
ROS-induced DNA damage in the form of oxidized bases and impaired repair of SSBs has been implicated in PD etiology [
84]. Studies have reported elevated levels of 8-OHdG, resulting from DNA oxidation, in PD brains [
11,
282] and increased levels of 8-OHG in the SNpc of PD patients [
282]. Similarly, more DNA damage, indicated by elevated levels of markers γH2AX and p53-binding protein foci, is present in dopaminergic neurons of two synucleinopathy PD mouse models [
286]. Further in vitro studies with dopaminergic SH-SY5Y cell lines suggested that excessive oxidation is at least partially responsible for DDR activation observed in vivo [
286]. In comparison to age-matched controls, the SNpc of PD patients displays increased SOD levels, whereas the activities of CAT, GPx and GR are similar as controls [
287]. Reduced levels of GSH and altered GSH/GSSG ratio, resulting in more of the oxidized form, have been detected in the SNpc of PD brains [
288]. Similarly, depletion of GSH is observed in patients with a pre-symptomatic form of PD, known as incidental Lewy body disease, compared to control subjects [
289]. Under elevated oxidative stress conditions, reduction in GSH results in dopaminergic neuronal loss [
290]. In addition, depletion of GSH results in increased NO and MPTP/MPP + toxicity in dopaminergic neurons in animal models of PD [
291,
292]. Glutamyl cysteine ethyl ester and GSH ethyl ester, two precursors of GSH, increase GSH levels in neuronal cells both in vitro and in vivo and are protective against oxidative and nitrosative stress [
293,
294]. Similarly, intracellular GSH levels are also rescued by thiol antioxidants such as α-lipoic acid in both in vitro and in vivo PD models [
295,
296]. Depletion of the antioxidant vitamin C has also been detected in PD [
297] and vitamin C levels in lymphocytes may be a potential biomarker of disease progression in PD [
298]. Furthermore, cells with lower levels of uric acid (UA) are more vulnerable to oxidative damage [
299] and individuals with low cellular uric acid levels may be at a greater risk of developing PD [
300]. UA prevents 6-hydroxydopamine (6-OHDA)-induced oxidative damage in neuron-like PC12 cells and increases GSH and SOD1 [
301]. Similarly, GSH levels, SOD1 activity and dopaminergic neuronal damage are rescued in a 6-OHDA rat model of PD following UA treatment [
302]. SOD1 may be a first-line protection against enhanced ROS production in PD patients [
303]. RNS, such as NO and its metabolite PN, may also cause DNA damage in PD [
11] by reacting with superoxide anion radicals. NO can then generate more oxidatively active PN, which in turn may induce DNA fragmentation [
74].
Recent studies have identified mtDNA damage in PD [
304] and abasic sites in mtDNA of dopaminergic neurons in PD post-mortem human and rat brains [
305], which precede the onset of neurodegeneration [
305]. Significant accumulation of abasic sites in dopaminergic neurons, but not in cortical neurons, has been detected [
305‐
307]. Elevated levels of ROS render dopaminergic neurons in the SNpc more prone to DNA damage and contribute to neurodegeneration [
305,
308]. Consistent with this notion, BER activity is increased in the SNpc of PD patients [
309‐
311]. Knockout mouse models lacking DNA repair enzymes (human MutT homolog 1, an oxidized purine nucleoside triphosphate; and OGG1) are more susceptible to dopaminergic toxins and age-related degeneration in the nigrostriatal system [
312,
313]. Moreover, transgenic mouse models expressing a mitochondrial-targeted restriction enzyme causing mtDNA damage in dopaminergic neurons recapitulate many of the key features of PD, including motor phenotype, progressive loss of dopaminergic neurons in the SN and depletion of dopamine in the striatum [
314]. Taken together, these studies imply that redox dysregulation can induce mtDNA damage in PD and may contribute to neurodegeneration. α-Synuclein is co-localized with γH2AX and PAR in human HAP1 cells and in transgenic α-synuclein mouse models [
315]. Reducing α-synuclein levels using bleomycin results in higher DSBs and impaired DNA repair in these cells [
315]. Moreover, α-synuclein knockout mice show increased DSB levels [
315], suggesting that it may play a role in DNA repair. Interestingly, increased DNA damage and dopaminergic neuronal death have been observed in two PD mouse models [
286].
P53-mediated selective cell death is also evident in PD. NO-induced, p53-mediated dopaminergic neuronal death has been observed in a mouse SNpc-derived cell line (SN4741) as well as in vivo models of PD [
316]. The neurotoxin 6-OHDA is widely used to induce selective degeneration of dopaminergic and noradrenergic neurons and therefore, can imitate PD symptoms [
23]. DNA damage induced by 6-OHDA treatment is linked to p53-mediated cell death of primary dopaminergic neurons [
317]. Together, these lines of evidence suggest that DNA damage resulting from redox dysregulation may contribute to neurodegeneration in PD. Several studies also reported parthanatos in PD. MPTP induces neurodegeneration of dopaminergic neurons in SNpc, leading to PD symptoms [
281]. Several studies have linked neurotoxicity of MPTP to parthanatos of dopaminergic neurons. MPTP treatment induces DNA fragmentation both in vivo and in vitro [
318,
319]. Similarly, PARP upregulation-mediated toxicity to dopaminergic neurons is observed following MPTP administration in a mouse model [
320] and inhibition of PARP significantly attenuates these toxic effects [
321,
322]. Activation of PARP-1 and progressive loss of dopaminergic neurons by parthanatos have also been reported in a transgenic mouse model overexpressing aminoacyl tRNA synthase complex-interacting multifunctional protein 2, a parkin (E3 ubiquitin ligase) substrate [
323]. MPTP-induced parthanatos requires neuronal NO synthase [
320], suggesting a link between MPTP-induced PARP activation and subsequent ADP-ribose polymerisation as well as NO-induced DNA damage. Increased NO levels are also observed in nigral cells in PD [
324,
325].
ALS
ALS is a fatal, rapidly progressing neurodegenerative disorder that affects motor neurons in the brain, brainstem, and spinal cord [
326]. It is clinically, genetically and pathologically linked to FTD, which manifests as frontotemporal lobar degeneration [
327]. Variants in more than 40 genes cause ALS, most common of which are those encoding SOD1, chromosome 9 open reading frame 72 (C9orf72), TAR DNA-binding protein-43 (TDP-43) and fused in sarcoma (FUS), which are all linked to both sporadic and familial forms of disease [
328].
DNA damage is now increasingly implicated as an important pathophysiological mechanism in ALS [
329‐
331], particularly with the identification of both TDP-43 and FUS as proteins with normal cellular functions in DNA repair [
332‐
334]. Elevated levels of oxidative DNA damage are consistently observed in both sporadic and familial ALS patients [
335,
336]. Moreover, DNA damage is associated with redox dysregulation in ALS. Increased levels of 8-OHdG have been detected in the motor cortex of sporadic ALS patients, and in the spinal cords of both sporadic and familial ALS patients [
337,
338]. Similarly, analysis of plasma, urine and CSF of ALS patients revealed increased levels of 8-OHdG [
338]. High levels of 8-OHdG have also been reported in the SOD1
G93A transgenic mouse model [
339]. Decreased levels of BER enzymes DNA polymerases α and β have been detected in motor neurons of SOD1
G93A mice [
340]. Furthermore, decreased mitochondrial activity of OGG1 and increased 8-OHdG levels have been detected in spinal motor neurons of sporadic ALS patients, indicating that impairment of redox function, resulting in oxidative stress, disrupts DNA repair in the mitochondria [
341]. In addition, a polymorphism in
OGG1, resulting in the substitution of serine with cysteine (Ser326Cys), reduces DNA activity and is associated with increased risk of sporadic ALS [
342]. The levels of a common mitochondrial DNA deletion mutation (mtDNA4977) encoding a subunit III of the redox enzyme cytochrome oxidase, involved in OXPHOS [
343], are higher in Brodmann area 4 of primary motor cortices in sporadic ALS patients compared to controls [
343]. Moreover, increased abasic sites are also detected in spinal motor neurons of ALS patients compared to controls [
330]. Likewise, the levels of 8-OHdG are increased in cells expressing SOD1-G37R and SOD1-G85R compared to wild-type SOD1 [
344]. A meta-analysis examining the levels of blood oxidative stress markers in ALS patients reported increased levels of 8-OHdG, MDA (the end product of lipid peroxidation) and AOPP (advanced oxidation protein product, a marker of protein oxidation), and reduced levels of GSH, compared to healthy controls, which all reflect both DNA damage and redox dysfunction in ALS [
345]. Hence, these data imply that DNA damage is closely associated with ALS and is linked to redox dysregulation.
Similar to the other neurodegenerative diseases, PARP1 hyperactivation and toxicity are implicated in ALS pathogenesis [
346,
347]. Elevated PAR levels are observed in motor neurons of patients carrying a polyglutamine expansion in the gene encoding ataxin-2 and cases displaying the G4C2-hexanucleotide repeat expansion in C9orf72 [
348]. PARP-1 expression is also increased in astrocytes in sporadic ALS patients [
349] and it is widespread in the cerebellum, motor cortex and parietal cortex, reflecting increased activation [
348,
350]. PARP-1 levels are also elevated in astrocytes in the spinal cord in mutant SOD1
G93A transgenic mice [
351]. Pharmacological inhibition of PARP inhibits the accumulation of stress-induced TDP-43 granules in the cytoplasm and toxicity in rat primary spinal cord cultures [
348]. Furthermore, another study demonstrated that PARP1 knockout or treatment with PARP inhibitor olaparib reduces PAR levels and rescues TDP43-induced death in NSC-34 cells [
352]. Inhibition of PARP-1 inhibits the ROS-induced cell death and suppresses mitochondrial ROS production via ATF4 and MAP kinase phosphatase-1 in human cell lines [
353].
A significant increase in p53 expression has been detected in spinal cord tissues of ALS patients [
354]. Similarly, increased nuclear p53 immunoreactivity was detected in the motor cortex and spinal ventral horn of post-mortem tissues from ALS patients [
355] and in spinal motor neurons in SOD1
G86R mice [
356]. Importantly, p53 knockout or knockdown extends the lifespan of a mouse model expressing poly(PR), and protects against neurodegeneration in
Drosophila models [
357]. Strikingly, p53 knockout also inhibits DNA damage in poly(PR)-transduced cells and C9orf72-ALS iPSC-derived motor neurons. Increased DNA damage was observed following the ectopic expression of poly(GR)
80 or (GR)
80 in iPSC-derived control neurons[
13] and pharmacological or genetical reduction of oxidative stress partially retrieved DNA damage [
13]. The adverse role of poly(GR) on DNA damage has also been confirmed in neuronal cells in
Drosophila [
358]. Together, these studies indicate that p53 mediates the C9orf72-DPR-induced toxicity upstream of DNA damage, rather than downstream, implying that redox homeostasis is crucial for regulation of p53 function, and its modulation may protect against DNA damage [
357].
Expression of the C9orf72 repeat expansion induces DNA damage in familial ALS patient tissues and in cellular models [
13,
347,
359]. Moreover, poly(GR)
80 aggregates induce DNA damage and increase ROS levels in iPSC motor neurons derived from C9orf72-ALS/FTD patients, linking redox dysregulation to DNA damage [
13]. Similarly, induction of oxidative stress and upregulation of DNA damage markers γH2AX, ATR, GADD45, and p53 were observed in an age-dependent manner in iPSC-derived C9orf72 motor neurons [
13]. In the same study, cellular toxicity was rescued following administration of a water-soluble antioxidant and vitamin E analog, Trolox, in C9orf72 iPSCs. Furthermore, in another study, myogenic progenitors derived from C9orf72 ALS patients demonstrated high susceptibility to oxidative stress and dysregulation of mitochondrial and DNA repair genes, leading to cellular toxicity [
360]. Similarly, modifiers of poly(GR)
100 toxicity induce dysregulation of mitochondrial NADPH and DNA damage repair-related pathways in yeast cells [
361]. Another study demonstrated increased mtDNA due to increased ROS specifically in C9orf72 ALS patient-derived fibroblasts, but not in TDP-43 A382T fibroblasts [
362].
R-loops associated with oxidative stress are increased in post-mortem spinal cord tissues of C9orf72 patients and in poly(GA)-transfected MRC5 cells [
359]. Both DNA damage and cell death in poly(GA)-expressing cells can be partly rescued by overexpressing senataxin, which resolves R-loops [
359,
363]. Interestingly, mutations in the senataxin gene cause an autosomal dominant form of ALS, ALS4 [
364]. Interestingly, there is evidence that sentataxin regulates redox homeostasis. An N-terminal truncation mutant of Sen1, the yeast homolog of human senataxin, is critical for cell survival through regulation of redox homeostasis. This mutant also displays severe loss of mtDNA [
365]. Importantly, the N-terminal substrate interaction and C-terminal RNA/DNA helicase domains are conserved in Sen1, implying that the same domains may perform a similar function in human senataxin. Furthermore, sentataxin has 31 cysteine residues involved in disulphide bonding via redox-regulated PDI [
366]. Moreover, residue C1554, which is expected to engage in disulphide linkage with C1509, is mutated in a sporadic case of ALS4 [
367]. These findings together suggest that dysregulated redox signalling, leading to ROS production, is associated with C9orf72 in ALS.
While wild-type SOD1 is protective against DNA damage, ALS-mutant SOD1
G93A displays less nuclear localization and antioxidant activity and is not protective in cellular models [
368]. SOD1
G93A expression in cells deficient in aprataxin, which facilitates SSB and NHEJ DNA repair, sensitises cells to oxidative stress, exacerbates DNA repair deficiencies and increases the levels of heterochromatin [
369]. Another study demonstrated more DNA damage in peripheral blood mononuclear cells (PBMCs) of sporadic ALS patients, which display high levels of aggregated SOD1 compared to controls. However, no DNA damage was observed in PBMCs expressing soluble SOD1 only [
370]. Increased levels of oxidative DNA damage, DNA strand breaks, p53 activity and apoptosis were detected in cells expressing mutant SOD1
G93A compared to wild-type SH-SY5Y cells [
371]. However, in a more recent study, similar levels of DNA damage, assessed by the presence of γH2AX-positive foci, were detected in SOD1
G93A and SOD1
A4V patient-derived iPSC lines compared to isogenic controls [
330].
TDP-43 is redox-regulated because the cellular redox conditions control its solubility and nuclear function [
372]. We and others have demonstrated that TDP-43 normally functions in the repair of DSBs by NHEJ and associates with XRCC4 and ATP-dependent DNA Ligase 4 [
329,
373]. However, this function is impaired by the ALS-associated mutations [
332]. Moreover, GSH depletion by buthionine sulfoximine significantly increases mutant TDP-43
M337V mislocalisation and inclusion formation in Neuro-2a cells [
12]. Redox dysfunction results in oxidation and phosphorylation of TDP-43 by GADD34, which is induced by DNA damage, leading to cytoplasmic mislocalisation in HEK293T cells [
374]. Neuronal cells expressing mutant TDP-43 (Q331K and M337V) exhibit shortened neurites, increased oxidative stress and lower levels of heme oxygenase HO-1, which regulates redox signalling [
375]. Hence, as TDP-43 is important in DNA repair, mutations in TDP-43 could lead to DNA damage and induce redox dysfunction.
FUS is recruited to oxidative DNA damage sites in response to DNA SSB formation, where it facilitates the recruitment of XRCC1 and nuclear ligase III to regulate its ligation activity for optimal BER activity [
376]. Loss of nuclear FUS results in defects in DNA nick ligation in motor neurons due to reduced recruitment of XRCC1/ligase III to DNA strand breaks in cellular models [
376]. Interestingly, PARP-dependent DNA damage and apoptosis have been detected in human iPSCs over-expressing mutant FUS-NLS, which induces FUS mislocalisation to the cytoplasm [
377]. PARP is also involved in the formation of aberrant phase transition of FUS from the liquid compartments to solid-like aggregates, a process which is redox-regulated, at DNA damage sites [
378‐
380]. In addition, FUS
R521C transgenic mice display both oxidative damage and defects in DNA ligation [
381], implying that defects in DNA repair mechanisms and redox dysregulation are associated with FUS in ALS.
APE1 is implicated as a possible ALS-associated gene that is upregulated in motor neurons of ALS patients [
382‐
384]. Furthermore, the motor cortex of ALS patients contains epigenetic hypomethylation of the
APE1 promoter, and this region is vulnerable to DNA lesions induced by free radicals and intermediates [
330]. In pre-symptomatic transgenic SOD1
G93A mice, expression of APE1 is reduced in spinal motor neurons, indicating that a deficiency in DNA repair precedes motor neuron degeneration [
385]. However, whether the APE1 redox and DNA repair activity are dysregulated in ALS is unknown.
HD
HD is a severe, rapidly progressing autosomal-dominant condition caused by expansion of CAG (encoding glutamine) repeats in the gene encoding huntingtin protein [
386]. Translation of the polyglutamine repeat then produces an abnormally long protein. HD involves motor, psychiatric and cognitive symptoms and it results from degeneration of neurons in the striatum and other regions of the cerebral cortex. Unaffected individuals possess less than 35 polyglutamine repeats, whereas HD patients normally possess 36 to 120 repeats. Interestingly, the number of repeats correlates inversely with the age of disease onset, implying that disease is dependent on repeat length [
387]. Similar to other neurodegenerative disorders, redox dysregulation and impaired DNA repair are implicated in the pathogenesis of HD.
DNA damage is strongly implicated in the etiology of HD [
388]. Elevated DNA damage has been detected in human HD fibroblasts and HD mouse models [
389], where it precedes the aggregation of mutant huntingtin [
389]. More DNA damage has also been detected in HD patient PBMCs compared to controls [
390] and another clinical study also reported increased DNA damage in prodromal HD in blood cells [
391]. ATM was also identified as a modifier of HD-relevant phenotypes in a mouse model [
392]. More recently, many genes involved in DNA repair were found to be important regulators of age of disease onset and severity in a large genome-wide association study, including FANCD2/FANCI-associated nuclease 1(
FAN1) and
ERCC3 (ERCC excision repair 3)
. Also, defects in DNA repair pathways, including inactivation of DNA mismatch repair genes such as MutS Homolog 3 (
MSH3), were associated with modification of age of onset in multiple CAG repeat expansion diseases, suggesting that the CAG repeat itself is the cause of modification [
393]. The most significant hit was FAN1, which is associated with ICL repair, and multiple MMR genes were also detected [
388]. Moreover, several of the genes identified are also related to mitochondrial and redox signalling pathways [
388].
There is also evidence for oxidative DNA damage in HD, as revealed by increased expression of 8-OHdG compared to controls, in both nuclear DNA and mtDNA [
394], further linking DNA damage to redox dysregulation. HD patient fibroblasts also display deficient repair following oxidative DNA damage [
389]. Furthermore, somatic expansion of the polyglutamine repeat has also been associated with BER, which is a naturally error-prone process. This involves the BER enzyme OGG1 and the removal of oxidized base lesions, resulting in somatic expansion of repeats by SSBs and strand slippage [
386]. In addition, the extent of oxidative damage correlates positively with the expansion length [
386,
387]. Furthermore, the somatic expansion process induces further oxidative damage and error-prone repair of this damage by the formation of longer repeats [
386], forming a vicious, escalating oxidation–BER cycle. These data therefore imply that accumulation of oxidative DNA lesions due to dysfunction of mutant huntingtin in DNA repair in conditions of ROS may contribute to the onset of HD [
389].
The normal huntingtin protein is also thought to function in DNA repair, where it is detected as part of the transcription-coupled repair (TCR) complex. TCR is a subtype of NER that rapidly removes specific types of DNA damage from transcribed strands of expressed genes, in contrast to non-transcribed strands. The TCR complex detects lesions and mediates repair during transcription. Mutant huntingtin impairs the function of components of the TCR complex, PNKP and ataxin-3, resulting in more DNA damage and ATM hyperactivation [
395].
Huntingtin protein is also thought to repair damaged DNA following oxidative stress, revealing that redox dysregulation and DNA repair are intimately linked in HD [
389]. Huntingtin localises and forms a scaffold at DNA damage sites via an ATM-dependent process in the presence of ROS. Huntingtin protein is a sensor of ROS and it relocates to the nucleus following oxidation of Met8 [
396]. In the presence of ROS, liquid–liquid phase-separated droplets containing huntingtin colocalised with ATM are increased [
396‐
398]. ATM is activated during oxidative stress [
399] and inhibiting its activity delays disease progression in mouse HD models [
392].
Several studies have reported oxidative damage, decreased levels of antioxidants, cysteine and vitamin C, and deposition of iron (Fe) in the cytoplasm and mitochondria, in cells and tissues from HD models and patients. Uptake of vitamin C is compromised in cellular and mouse models of HD, which precedes mitochondrial dysfunction [
400]. The levels of GSH are dysregulated in the plasma and cortex of HD patients compared to controls [
401,
402], although whether there is an increase or decrease in GSH is still under debate [
403]. Nevertheless, both processes would perturb the cellular redox conditions in HD. There is also evidence for impaired cysteine metabolism in HD [
16].
Whilst the HD repeat length is the main factor determining the age at disease onset, genetic modifiers also make a significant contribution to the variation in onset age [
404]. The expanded CAG polyglutamine repeat is somatically unstable, and its length increases progressively over time in neurons, particularly in the striatum and cortex. Due to this somatic instability, larger increases in repeat length are associated with earlier disease onset [
404]. Interestingly, somatic expansion of the HD CAG repeat is mediated by DNA damage, via a mechanism involving the introduction of mutations by MMR [
388]. Similarly, in cellular and animal models, deficiency of MSH3, MSH2, MLH3, MLH1 or PMS2, or increased expression of FAN1, prevents the somatic expansion of CAG repeats [
405,
406]. Similarly, transcriptome-wide association studies have revealed that lower expression of MSH3, and increased levels of FAN1, are associated with CAG repeat stability, later onset, and slower disease progression [
407,
408]. However, it is unclear if this involves oxidative DNA damage. A more recent study showed that an interaction between FAN1 and MLH1, via a highly conserved SPYF motif at the N terminus of FAN1, is protective against somatic expansion by restricting the recruitment of MLHA by MSH3 [
405]. This study suggested that FAN1 normally stabilizes CAG repeats, by both inhibiting formation of the MMR complex that promotes somatic repeat expansion and enhancing correct DNA repair via nuclease activity.