|Home | About | Journals | Submit | Contact Us | Français|
8-Oxoguanine (8-oxoG), a common DNA lesion caused by reactive oxygen species, is associated with carcinogenesis and neurodegeneration. Although the mechanism by which 8-oxoG causes carcinogenesis is well understood, the mechanism by which it causes neurodegeneration is unknown. Here, we report that neurodegeneration is triggered by MUTYH-mediated excision repair of 8-oxoG–paired adenine. Mutant mice lacking 8-oxo–2′-deoxyguanosine triphosphate–depleting (8-oxo–dGTP–depleting) MTH1 and/or 8-oxoG–excising OGG1 exhibited severe striatal neurodegeneration, whereas mutant mice lacking MUTYH or OGG1/MUTYH were resistant to neurodegeneration under conditions of oxidative stress. These results indicate that OGG1 and MTH1 are protective, while MUTYH promotes neurodegeneration. We observed that 8-oxoG accumulated in the mitochondrial DNA of neurons and caused calpain-dependent neuronal loss, while delayed nuclear accumulation of 8-oxoG in microglia resulted in PARP-dependent activation of apoptosis-inducing factor and exacerbated microgliosis. These results revealed that neurodegeneration is a complex process caused by 8-oxoG accumulation in the genomes of neurons and microglia. Different signaling pathways were triggered by the accumulation of single-strand breaks in each type of DNA generated during base excision repair initiated by MUTYH, suggesting that suppression of MUTYH may protect the brain under conditions of oxidative stress.
The DNA and precursor nucleotides in living organisms are always in danger of oxidation by ROS that are inevitably generated as a by-product of oxygen respiration and are products of host defense and signal transduction mechanisms (1, 2). If oxidized lesions accumulate in DNA, mutagenesis or cell death may result (3–5). Among all nucleobases, guanine is known to be the most susceptible to oxidation, and its simple oxidized form, 8-oxoguanine (8-oxoG), which can pair with adenine as well as cytosine, is one of the major oxidation products in DNA and nucleotides (6, 7).
Mammalian cells are equipped with elaborate means of minimizing accumulation of 8-oxoG in DNA. 8-oxo–2′-deoxyguanosine triphosphatase (8-oxo–dGTPase) encoded by MTH1 hydrolyzes 8-oxo–dGTP to 8-oxo–dGMP and pyrophosphate in nucleotide pools, thereby avoiding incorporation of 8-oxo–dGMP into DNA (8, 9). 8-OxoG DNA glycosylase 1 encoded by OGG1 excises 8-oxoG paired with cytosine in DNA (10), while adenine DNA glycosylase encoded by MUTYH removes the adenine inserted opposite 8-oxoG in template DNA during DNA replication (11), thus preventing mutagenesis. Mutant mice lacking these genes exhibit increased spontaneous mutation rate and susceptibility to carcinogenesis with increased accumulation of 8-oxoG in DNA (12–15).
Oxidative stress is considered to be important in the etiology of several neurodegenerative disorders, and it has been shown that 8-oxoG levels are significantly increased in mitochondrial DNA (mtDNA) as well as nuclear DNA (nDNA) in the brains of patients with Parkinson’s disease (PD) (16), Alzheimer’s disease (AD) (17), and Huntington’s disease (HD) (18) in comparison with control brains. Expression levels of MTH1, OGG1, and MUTYH are also significantly altered in the brains of such patients (16, 19–22), suggesting that their altered expression along with accumulation of 8-oxoG in brain cause neurodegeneration; however, how 8-oxoG and these enzymes are associated with the neurodegenerative process is poorly understood.
The striatum plays a key role in motor, cognitive, and motivational processes (23). Abnormal striatal function is involved in several neurodegenerative disorders, such as PD, AD, and HD. The inhibitor of mitochondrial succinate dehydrogenase 3-nitropropionic acid (3-NP), a naturally occurring plant toxin, has been shown to cause striatal degeneration and motor impairments in animals similar to those seen in HD (24, 25). It has been established that administration of 3-NP to rodents and nonhuman primates replicates most of the clinical and pathophysiological hallmarks of HD, including spontaneous choreiform and dystonic movements, frontal-type cognitive deficits, and progressive heterogeneous striatal degeneration, at least partially by apoptosis (26). We have shown that increased expression of human MTH1 in mouse striatum efficiently suppresses such striatal degeneration, accompanied by effective suppression of the 8-oxoG accumulation in the striatum induced by 3-NP (27). However, it is not clear to what extent 8-oxoG accumulated in DNA is responsible for the neurodegeneration, because MTH1 can hydrolyze oxidized forms of ATP, GTP, and dATP as well as dGTP (28). Moreover, it is not known which type of DNA (nDNA and/or mtDNA) harboring 8-oxoG is associated with such neurodegeneration, nor is it known how the neuronal loss occurs.
In the present study, we investigated the mechanism by which 8-oxoG is involved in neurodegeneration using mutant mice lacking Ogg1, Mth1, and/or Mutyh genes under conditions of oxidative stress. We demonstrated that OGG1 cooperatively protects brain with MTH1, reflecting a beneficial role of DNA repair and, intriguingly, that DNA repair by MUTYH is required for neurodegeneration upon 8-oxoG accumulation in brain, thus exhibiting a harmful role of DNA repair.
We first immunohistochemically confirmed the expression of MTH1 and OGG1 proteins in mouse striatum. Specific immunoreactivities (IRs) for both MTH1 and OGG1 were detected only in WT striatum, but not in Mth1-KO or Ogg1-KO striatum, respectively (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI65053DS1). Immunofluorescence microscopy revealed that cytoplasmic IRs for MTH1 and OGG1 were predominantly detected in DARPP32-positive medium spiny neurons (MSNs) in WT striatum (Figure (Figure1A).1A). Immunoelectron microscopy confirmed that MTH1 and OGG1 IRs were distributed in the cytoplasm and nucleus only in the WT striatum (Supplemental Figure 2). It is evident that the OGG1 cytoplasmic IR colocalized with mitochondrial HSP60 IR (Figure (Figure1A).1A). Thus, we hypothesized that MTH1 and OGG1 play important roles in protecting MSNs in the striatum from oxidative stress.
To evaluate this hypothesis, we examined the effects of 3-NP on motor function in WT and mutant mouse strains lacking Ogg1 and/or Mth1 genes, in small cohorts (Supplemental Figure 3). Chronic exposure of mice to 3-NP resulted in motor impairments. Individual pairwise comparisons to WT mice revealed that Mth1/Ogg1–double-KO (Mth1/Ogg1-DKO) and to a lesser extent Ogg1-KO mice exhibited a statistically significant decline of travel distance in an open-field test, but no significant difference was found between Mth1-KO and WT mice (P < 0.05, P < 0.02, respectively). We then examined the effects of 3-NP on WT and Mth1/Ogg1-DKO mice in large cohorts. First, we evaluated the 3-NP–induced motor symptoms in WT and Mth1/Ogg1-DKO mice based on neurological scores (refs. 29, 30, and Supplemental Figure 4) and found that Mth1/Ogg1-DKO mice exhibited more severe motor symptoms. A quantitative analysis of spontaneous locomotor behavior in their home cages during day and night revealed that Mth1/Ogg1-DKO mice show more hypolocomotor activity in the home cages during the night on the fifth and sixth days of 3-NP exposure (Supplemental Figure 5). An open-field test showed that the distance travelled by DKO mice in the open field was significantly reduced, to 52% of the level seen in WT mice after exposure to 3-NP for 7 days, and vertical rearing was also significantly reduced in DKO mice to 59% of the level seen in WT mice (Figure (Figure1B1B and Supplemental Videos 1 and 2). Mth1/Ogg1-DKO and WT mice both exhibited modest reductions in performance on a rotarod test after exposure to 3-NP for 7 days; however, there was no significant difference between the 2 groups of mice (data not shown), indicating that the significantly decreased motor activity in Mth1/Ogg1-DKO mice was not due to increased muscle disorder.
More severe striatal degeneration was observed in DKO mice than in WT mice after 7 days of exposure to 3-NP (Figure (Figure1C).1C). Laser scanning confocal immunofluorescence microscopy (LSCIM) revealed that most DARPP32-positive MSNs were lost and increased numbers of F4/80-positive microglia were detected in the DKO striatum, especially in the dorsolateral part of the striatum, after exposure to 3-NP for 7 days (Figure (Figure1D),1D), indicating that the 3-NP induces striatal degeneration accompanied by microgliosis. Thus, we concluded that the double deficiency in MTH1 and OGG1 renders the striatum highly vulnerable to 3-NP–induced neurodegeneration.
We previously established that 8-oxoG in mtDNA can be visualized in tissue sections treated only with RNase before incubation with anti–8-oxo–dG, while 8-oxoG in nDNA can be detected in sections treated with RNase and HCl before incubation with the antibody (31, 32). Most neurons in the DKO striatum were lost after exposure to 3-NP for 7 days (Figure (Figure1,1, C and D); therefore, it is likely that 8-oxoG in nDNA may not be detected in the striatum after 7 days of exposure to 3-NP. We found that MSNs expressing DARPP32 were still detectable in the striata of WT and DKO mice after exposure to 3-NP for 4 days (Figure (Figure2A);2A); therefore, we examined nuclear 8-oxoG levels by LC-MS/MS analysis of striatal nDNA prepared after exposure to 3-NP for 1 or 3 days using stable isotope-labeled [15N5] 8-oxo–dG as an internal standard. Nuclear 8-oxo–dG levels were not significantly increased in either group after exposure to 3-NP, although DKO mice had higher basal levels of nuclear 8-oxo–dG in the striatum than did WT mice (Figure (Figure2B).2B). LSCIM following pretreatment of sections with HCl revealed that exposure to 3-NP for 4 days causes a marginal increase in nuclear 8-oxoG IR in the DKO striatum in agreement with the results of LC-MS/MS analysis (data not shown). These results indicate that 8-oxoG is not highly accumulated in nDNA prior to loss of MSNs in the striatum.
Exposure to 3-NP for 7 days caused significant increases in cytoplasmic 8-oxoG IR in the Mth1/Ogg1-DKO striatum in comparison with WT or untreated samples without HCl pretreatment, suggesting 8-oxoG accumulation in mtDNA (Figure (Figure3,3, A and B). LSCIM revealed that DARPP32-positive MSNs in the DKO striatum exhibited strong cytoplasmic 8-oxoG IR without HCl pretreatment after exposure to 3-NP for 4 days (Figure (Figure3C)3C) and that this IR was colocalized with that for mitochondrial transcription factor A (TFAM) with which mtDNA is packaged (ref. 33 and Figure Figure3D),3D), indicating that 8-oxoG accumulates mostly in the mtDNA of MSNs in the DKO striatum. The basal level of mitochondrial 8-oxoG in DKO MSNs was 8-fold higher than that in WT MSNs without exposure to 3-NP, indicating an increased spontaneous accumulation of 8-oxoG in the mtDNA. After 4 days of 3-NP exposure, the levels were significantly increased in both WT and DKO MSNs; the level in DKO MSNs was 3.5 times higher than that in WT MSNs (Figure (Figure3,3, E and F).
On the seventh day of exposure to 3-NP, we observed significantly increased IR for single-stranded DNA (ssDNA) in the cytoplasm of neurons in the DKO striatum, in agreement with selective damage such as single-strand breaks (SSBs) in their mtDNA. Cells with small nuclei exhibited nuclear ssDNA IRs (Figure (Figure3G),3G), indicating that the nuclei of microglia also accumulate DNA damage. On the seventh day of exposure to 3-NP, we noticed that microglia proliferating in the region of striatum, where MSNs were largely lost, exhibit nuclear, and to a lesser extent cytoplasmic, 8-oxoG IR (Figure (Figure33H).
We previously showed that Ogg1-KO cells expressing a nuclear form of human OGG1 accumulate 8-oxoG selectively in mtDNA under oxidative stress and demonstrated that the accumulation of 8-oxoG in mtDNA causes mitochondrial dysfunction and Ca2+ release, thereby activating calpain to execute caspase-independent cell death (34). Because 8-oxoG was selectively accumulated in the mtDNA of MSNs in Mth1/Ogg1-DKO striatum after 3-NP exposure (Figure (Figure3,3, A–F), we examined whether accumulation of 8-oxoG in mtDNA in the striatum activates calpain. Western blotting analysis of α-spectrin, a substrate for calpain (35), revealed that 3-NP exposure for 4 or 7 days caused markedly increased processing of α-spectrin in the Mth1/Ogg1-DKO striatum in comparison with WT (Figure (Figure4A).4A). The relative level of a 145-kDa band corresponding to a major calpain-cleaved product (35) was significantly increased in the DKO striatum after exposure to 3-NP, while such increased cleavage of α-spectrin was barely seen in WT striatum (Figure (Figure4A).4A). It is likely that the dose of 3-NP (120 mg/kg/d) is not enough to activate calpain in WT striatum, thereby causing little neurodegeneration (Figure (Figure1),1), consistent with a previous suggestion that the activation of calpain observed in 3-NP–treated animals is probably instrumental in striatal degeneration (36). In PBS-treated striatum, α-spectrin IR was detected in the nuclei of neurons in both WT and DKO striatum; however, we noticed that perinuclear IR for α-spectrin was evident in the DKO striatum after exposure to 3-NP for 4 days (Figure (Figure4B),4B), indicating that the calpain-dependent processing of α-spectrin results in its altered intracellular localization (37). Because such perinuclear IR of α-spectrin was exclusively detected in DARPP32-positive neurons in the DKO striatum after exposure to 3-NP, calpain must be selectively activated in MSNs (Figure (Figure44C).
Using Ogg1-KO cells, which express a mitochondrial form of human OGG1, we previously demonstrated that the selective accumulation of 8-oxoG in nDNA causes PARP-dependent nuclear translocation of apoptosis-inducing factor (AIF) and that the cells subsequently underwent PARP/AIF-dependent cell death (34). We therefore examined whether PARP and AIF are involved in 3-NP–induced striatal degeneration. LSCIM with anti-poly(ADP-ribose) (anti-PAR) or anti-AIF antibodies revealed a significant accumulation of PAR-polymer and AIF in the striatal region of DKO mice from which MSNs were largely lost after exposure to 3-NP for 7 days (Figure (Figure4D).4D). We observed increased numbers of F4/80-positive microglia in this region (Figure (Figure1D),1D), and IRs for PAR and AIF were mostly detected in the nuclei of microglia (Figure (Figure5A).5A). No AIF or PAR IR was seen in MSNs (Figure (Figure5B),5B), and no AIF IR was seen in GFAP-positive astrocytes after exposure to 3-NP (Figure (Figure5C).5C). Western blotting analysis of AIF revealed that exposure to 3-NP for 7 days caused significant accumulation of a 57-kDa form of AIF, a processed nuclear form of AIF (38), only in the DKO striatum, but not in the WT striatum (Figure (Figure5D),5D), indicating that delayed activation of AIF accompanied by PAR-polymer accumulation occurs selectively in microglia in the DKO striatum exposed to 3-NP. These data indicate that activation of calpain is restricted to MSNs and that activation of AIF is restricted to microglia.
We next tested whether inhibition of calpain and PARP could afford neuroprotection against the neurodegeneration following 3-NP administration, to confirm the involvement of the 2 distinct pathways in 3-NP–induced neurodegeneration.
We examined the effects of administration of a calpain inhibitor (MDL28170) and an inhibitor of PARP (3-aminobenzamide [3-AB]) on striatal degeneration induced by 3-NP. Both inhibitors significantly improved 3-NP–induced motor impairments (Figure (Figure6A)6A) and suppressed MSN loss (Figure (Figure6,6, B and C) as well as microgliosis (Figure (Figure6D)6D) in Mth1/Ogg1-DKO mice. The perinuclear distribution of α-spectrin was suppressed by the calpain inhibitor (Figure (Figure6E),6E), while accumulation of both PAR polymer and nuclear AIF was efficiently suppressed by the PARP inhibitor (Figure (Figure6F).6F). Thus, both calpain activation in MSNs and poly(ADP-ribosyl)ation causing nuclear translocation of AIF in microglia play important roles in 3-NP–induced striatal degeneration. Previous research has shown that downregulation of PARP in microglia protects neurons from secondary damage as well as preventing microgliosis following N-methyl-D-aspartate-induced hippocampal injury (39), as we observed for 3-NP–induced striatal degeneration (Figure (Figure66C).
Buildup of SSBs in either mtDNA or nDNA was found in Mth1/Ogg1-DKO striatum after exposure to 3-NP (Figure (Figure3G).3G). SSBs in nDNA are recognized by PARP, thus facilitating their repair or inducing nuclear translocation of AIF if SSBs are accumulated beyond the repair capacity, while buildup of excessive SSBs in mtDNA is known to cause mitochondrial dysfunction followed by calpain activation (34). We therefore examined the time course and spatial distribution of ssDNA IRs in comparison with those of α-spectrin or PAR IRs in DKO mice after exposure to 3-NP to delineate their interrelation.
LSCIM with an anti-ssDNA antibody revealed that IR for cytoplasmic ssDNA was prominent in the Mth1/Ogg1-DKO striatum in comparison with WT mice after 4 days of 3-NP exposure, while nuclear IR for ssDNA was more dominant in the striatal region where MSNs were largely lost after 7 days of 3-NP exposure (Figure (Figure7,7, A and B). PAR IR was detected only in the DKO striatum exposed to 3-NP for 7 days and mostly colocalized with nuclear ssDNA IR (Figure (Figure7C).7C). On the other hand, cytoplasmic ssDNA IR was detected in neurons with perinuclear α-spectrin IRs (Figure (Figure7D).7D). These observations strongly suggest that 8-oxoG accumulated in either mtDNA in MSNs or nDNA in microglia causes accumulation of SSBs in each type of DNA, and that accumulation of SSBs in mtDNA results in calpain activation in MSNs, while that in nDNA activates the PARP/AIF pathway in microglia.
We previously demonstrated that MUTYH-dependent base excision repair (BER) is responsible for the accumulation of SSBs in nDNA or mtDNA with increased 8-oxoG accumulation in Ogg1-KO cells (34). To analyze the function of MUTYH in the striatum, we next examined the expression of MUTYH protein in striatum (Figure (Figure8A)8A) and found that MUTYH IR was preferentially localized to the cytoplasm and mitochondria and to a lesser extent nuclei in MSNs (Figure (Figure8,8, B and C), while nuclear MUTYH IR in microglia was more apparent than that in MSNs (Figure (Figure88D).
When we challenged Mutyh-KO mice with various doses of 3-NP (120 mg/kg/d to 180 mg/kg/d), we found that Mutyh-KO mice exhibit higher survival rates than do WT mice within 7 days of exposure (data not shown). 3-NP–induced motor symptoms were evaluated in WT and Mutyh-KO mice after exposure to 3-NP (150 mg/kg/d) (Supplemental Figure 6), and WT mice exhibited higher scores than did Mutyh-KO mice. Furthermore, we examined motor function of the survivors in the open-field test (Figure (Figure8E).8E). Mutyh-KO mice exhibited significantly better rearing performance after exposure to 3-NP (120 mg/kg/d) than did WT. Electron microscopic examination and histochemical detection of cytochrome c oxidase activity revealed that mitochondrial structure and function were well maintained in the Mutyh-KO striatum after 3-NP exposure, while mitochondria in the WT striatum were largely degenerated and exhibited lower cytochrome c oxidase activity than did those in the Mutyh-KO striatum (Figure (Figure8,8, F and G), indicating that MUTYH enhances neurodegeneration.
We previously reported that exposure to sodium nitroprusside causes cell death accompanied by 8-oxoG accumulation in the mtDNA of quiescent MTH1-null cells and that knockdown of MUTYH in quiescent MTH1-null cells significantly prevents cell death (40). Considering this finding together with the Mutyh-KO phenotype, it is reasonable to assume that striatal degeneration accompanied by calpain and PARP/AIF pathways in Ogg1-KO or Ogg1/Mth1-DKO mice may be rescued by MUTYH deficiency. Because Ogg1/Mth1/Mutyh–triple KO mice are highly susceptible to spontaneous tumorigenesis within 100 days after birth (our unpublished observation), we tested our hypothesis using Ogg1-KO and Ogg1/Mutyh-DKO mice. Ogg1/Mutyh-DKO mice exhibited significantly better motor performance than did Ogg1-KO mice after exposure to 3-NP for 7 days (Figure (Figure9A),9A), and neither MSN loss nor microgliosis was evident in the DKO striatum (Figure (Figure9,9, B and C). IR for nuclear and cytoplasmic ssDNA was again evident in the Ogg1-KO striatum after exposure to 3-NP; however, there was no such IR in the Ogg1/Mutyh-DKO striatum (Figure (Figure10,10, A and B). In the Ogg1-KO striatum, cytoplasmic ssDNA IR was evident in DARPP32-positive neurons, while nuclear ssDNA IR was detected in F4/80-positive microglia (Figure (Figure10,10, C and D) and cytoplasmic ssDNA IR was colocalized with TFAM IR after exposure to 3-NP for 7 days (Figure (Figure10E).10E). Moreover, we confirmed that cytoplasmic α-spectrin IRs and nuclear PAR IR were detected only in the Ogg1-KO but not the Ogg1/Mutyh-DKO striatum after exposure to 3-NP, as seen in Mth1/Ogg1-DKO mice (data not shown).
Finally, we confirmed that Ogg1/Mutyh-DKO mice exhibit resistance to 3-NP–induced locomotor dysfunction similar to that of Mutyh-KO mice (Supplemental Figure 7).
Taken together, these results indicate that 3-NP–induced striatal degeneration is largely dependent on MUTYH function under conditions of OGG1 deficiency as well as MTH1/OGG1 deficiency.
In the present study, we demonstrated that 8-oxoG accumulated within mtDNA and nDNA under conditions of oxidative stress is differentially involved in neurodegeneration through activation of the calpain and PARP/AIF pathways. MTH1 and OGG1, which prevent accumulation of 8-oxoG in the respective DNAs, protect the brain, while MUTYH, which initiates excision repair of adenine opposite 8-oxoG, triggers neurodegeneration.
We thus propose that the neurotoxicity of oxidative stress in the mammalian brain predominantly results from MUTYH-initiated excision repair of adenine opposite 8-oxoG, which is mostly derived from 8-oxo–dGTP accumulated in the nucleotide pool under oxidative stress.
MTH1 and OGG1 suppress neurodegeneration as well as mutagenesis and carcinogenesis by preventing accumulation of 8-oxoG in cellular DNAs. It has been widely accepted that both degenerative diseases and cancers are tightly associated with DNA damage (1, 4). We previously showed that Mth1-KO and Ogg1-KO mice exhibit increased incidence of spontaneous tumors in liver and lung, respectively. This was accompanied by accumulation of 8-oxoG in their nDNAs, thus demonstrating that MTH1 and OGG1 prevent accumulation of 8-oxoG in nDNA, thereby suppressing mutagenesis and carcinogenesis (12, 13). However, we also found that mice lacking both MTH1 and OGG1 developed no lung tumors, although 8-oxoG was highly accumulated in nDNA (13). This observation suggests that excess accumulation of 8-oxoG under combined deficiency of MTH1 and OGG1 may induce tumor cell death, thereby leading to diminished occurrence of lung tumors (41). Furthermore, we have shown that increased accumulation of 8-oxoG in nDNA and/or mtDNA causes cell death and that MTH1 and OGG1 suppress cell death (34, 40, 42).
We previously reported that Mth1-KO mice exhibit increased accumulation of 8-oxoG in the mtDNA of striatal nerve terminals of dopamine neurons after 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine administration, followed by more severe terminal degeneration of dopamine neurons in comparison to WT mice (31). MTH1 also efficiently suppresses the accumulation of 8-oxoG in mtDNA in the hippocampus caused by kainate-induced excitotoxicity, especially in microglia (43). Moreover, it has been shown that human MTH1 transgenic mice are resistant to 3-NP–induced striatal degeneration (27), suggesting that increased accumulation of 8-oxoG in brain causes neurodegeneration.
In the present study, we showed that double deficiency of MTH1 and OGG1 renders the striatum highly vulnerable to 3-NP–induced neurodegeneration, accompanied by highly increased accumulation of 8-oxoG in both the mtDNA of MSNs and the nDNA of microglia in the striatum. This is the first experimental evidence that OGG1 and MTH1 cooperatively suppress 8-oxoG accumulation in brain DNAs, thus protecting the brain from neuronal loss and microgliosis, which result in severe neurodegeneration and behavioral impairments.
Mitochondria prepared from caudate nucleus as well as other regions of mouse brain possess substantial levels of OGG1 incision activity (44). In the present study, we found that OGG1 was predominantly localized in the mitochondria of MSNs in normal striatum (Figure (Figure1A)1A) and that OGG1 deficiency significantly increased accumulation of mitochondrial 8-oxoG and ssDNA in MSNs after exposure to 3-NP. Considering these findings together, we concluded that mitochondrial OGG1 is crucial to protect striatal MSNs from oxidative stress.
MUTYH suppresses carcinogenesis but promotes neurodegeneration by inducing cell death. We have shown that Mutyh-KO mice exhibit the most significant increase in spontaneous occurrence of tumors among Mth1-, Ogg1-, and Mutyh-KO mice (14). It has been reported that Ogg1/Mutyh-DKO mice (15) and Mth1/Ogg1/Mutyh–triple-KO mice exhibit markedly increased spontaneous tumorigenesis in various tissues over a much shorter period (our unpublished observations). These observations indicate that MUTYH functions as the strongest suppressor of tumorigenesis among the 3, because Mth1/Ogg1-DKO mice develop no lung tumors whereas Ogg1-KO mice do (13). Thus, we hypothesized that MUTYH induces cell death, thereby suppressing tumorigenesis in the absence of MTH1 and OGG1. We previously showed that MUTYH is required for generation of SSBs in DNA and induction of cell death when 8-oxoG is highly accumulated in either nDNA or mtDNA under conditions of oxidative stress (34, 40, 41).
In the present study, we found that MUTYH protein is highly expressed in the striatum and is mostly present in the mitochondria of MSNs and that Ogg1/Mutyh-DKO mice exhibit markedly increased resistance to 3-NP–induced striatal degeneration in comparison with Ogg1-KO mice and even WT mice. These results clearly demonstrate that MUTYH enhances neurodegeneration through the induction of cell death once 8-oxoG becomes accumulated in neurons and/or microglia.
MUTYH-initiated BER induces accumulation of SSBs in DNA, thereby triggering 2 distinct cell death pathways dependent on calpain or PARP-AIF. We previously established that MUTYH initiates the 2 distinct pathways of cell death while 8-oxoG accumulates selectively in mtDNA or nDNA in Ogg1-KO cell lines (34). In the present study, we found that 8-oxoG is significantly accumulated first in the mtDNA of MSNs (P < 0.0001) and then in the nDNA of microglia during striatal degeneration caused by 3-NP in the absence of OGG1 and/or MTH1. This is followed by selective accumulation of SSBs in each type of DNA. Calpain activation was observed in MSNs, while nuclear accumulation of PAR-polymer and AIF was evident in microglia. Calpain activation, PARP-AIF activation, and accumulation of SSBs were observed in Ogg1-KO mice, but were efficiently suppressed by a lack of MUTYH, which initiated BER of adenines inserted opposite 8-oxoG (41). Moreover, administration of inhibitors of calpain or PARP to Mth1/ Ogg1-DKO mice efficiently suppressed 3-NP–induced loss of MSNs and microgliosis in the striatum, and this was followed by improvement of motor impairment. Considering these observations together, we concluded that MUTYH-initiated BER induces accumulation of SSBs in DNA, thereby triggering 2 distinct cell death pathways dependent on calpain or PARP-AIF, as summarized in Figure Figure11.11. The anti-ssDNA antibody used in the present study also recognizes single-stranded regions of double-stranded breaks (DSBs) generated by resection of the ends (45). It is thus possible that DSBs may also be generated through incision of the template strand if there are other types of damaged bases close to 8-oxoG in the template DNA generated under increased oxidative stress.
Recently, Foti et al. reported that cytotoxicity of bactericidal antibiotics such as β-lactams and quinolones predominantly results from lethal double-strand DNA breaks caused by incomplete repair of closely spaced 8-oxoG lesions and is efficiently suppressed by mutM/mutY-double deficiency (5). Considered together with our previous findings (28, 34, 40–42), it is noteworthy that 8-oxoG accumulated in the genome causes cell death through a common mechanism from bacteria to mammals. In the mammalian brain, however, there are different types of cells, such as neurons and microglia, and each cell has 2 distinct genomes, 1 in the nucleus, and 1 in mitochondria. Our findings in the present study show for what we believe is the first time that neurodegeneration is a complex process caused by 8-oxoG accumulated in the 2 distinct genomes in the brain.
Neurodegeneration induced by 8-oxoG and MUTYH depends on both replication of mtDNA in neurons and nDNA in microglia. Overexpression of human MTH1 in mouse striatum efficiently abrogates 3-NP–induced striatal degeneration accompanied by effective suppression of 8-oxoG accumulation in the striatum (27). Mth1/Ogg1-DKO mice exhibited the highest susceptibility to 3-NP–induced striatal degeneration with the highest levels of 8-oxoG in striatal DNAs. These findings indicate that the major source of 8-oxoG accumulated in DNA are 8-oxo–dGTP generated in nucleotide pool; the former is repaired by OGG1 and the latter is hydrolyzed by MTH1.
DNA replication is essential for accumulation of 8-oxoG in the respective DNA; moreover, insertion of adenine opposite 8-oxoG also depends on replication. Neurons are postmitotic and only mtDNA but not nDNA is always replicated in neurons in order to provide the energy essential for maintenance of neuronal functions. 8-oxoG is therefore dominantly accumulated in mtDNA of neurons under conditions of oxidative stress, resulting in SSB accumulation in mtDNA during MUTYH-initiated BER of adenine inserted opposite 8-oxoG. In contrast, microglia are mitotic; thus, 8-oxoG and SSBs accumulated in nDNA dependent on their replication and MUTYH-initiated BER (Figure (Figure11).11). Thus, replication of the respective DNAs in neurons and microglia determines which of the 2 separate signaling pathways must be activated upon accumulation of 8-oxoG under conditions of oxidative stress.
Microgliosis is a delayed event in striatal degeneration induced by 3-NP (Figure (Figure1D);1D); therefore, it is likely that accumulation of 8-oxoG followed by ssDNA accumulation in nDNA of microglia (Figures (Figures3H3H and 7B) is not due to a direct effect of mitochondrial ROS generated in MSNs immediately after exposure to 3-NP. Neurons damaged by mitochondrial ROS thereafter induce a delayed inflammatory response, resulting in microgliosis, which is known to be accompanied by secondary generation of ROS in brain (46). Since inhibition of PARP activity, which is activated by SSBs accumulated in the nDNA of microglia resulting in AIF activation (Figure (Figure5,5, A and D), efficiently suppressed neuronal loss as well as microgliosis, ROS generated in microglia may in turn attack neurons to promote the degenerative process as suggested by Brouillet et al. (36), thus initiating a vicious cycle of neurodegeneration.
Growing evidence suggests that mitochondrial dysfunction is generally associated with neurodegenerative diseases (47) and that microglial activation is also involved in the pathogenesis of most forms of neurodegeneration (48). It is known that activation of calpain, PARP, and AIF is involved in various neurodegenerative diseases such as AD, PD, and HD (38, 49), in which 8-oxoG is a common molecular marker (50), together with altered expression of MTH1, OGG1, and MUTYH (16, 19–22). Inhibition of calpain and PARP has been considered to be a potential therapeutic approach for neurologic disorders (38, 49).
Our findings in the present study show that 2 distinct molecular pathways are separately activated in the mitochondria of neurons and the nuclei of microglia through oxidative DNA damage and its repair, thus providing insights into the molecular mechanisms underlying neurodegeneration under conditions of oxidative stress as well as providing therapeutic targets for neurodegenerative diseases. Suppression of MUTYH may thus be an efficient strategy for protecting the brain under conditions of oxidative stress. However, it will be important not to increase the risk of carcinogenesis under MUTYH suppression; therefore, further experimental evaluation is essential.
We previously established Mth1, Ogg1, and Mutyh gene KO mice (12, 13, 51). Heterozygous mice (Mth1+/–, Ogg1+/–, Mutyh+/–) were backcrossed to C57BL/6J mice (Clea Japan) for more than 15 generations, thereby ensuring the mice were on a pure C57BL/6J genetic background. Homozygous KO mice (Mth1–/–, Ogg1–/–, Mutyh–/–) were generated by crossing of the corresponding heterozygous mice and inbred for 2 to 3 generations to obtain the mice used in each experiment. Mth1+/–Ogg1+/– double-heterozygous mice obtained by crossing of Mth1+/– and Ogg1+/– mice were crossed to obtain Mth1–/–Ogg1–/– (Mth1/Ogg1-DKO) mice and were inbred for 2 to 3 generations to obtain the mice used in each experiment. Ogg1+/–Mutyh+/– double-heterozygous mice were obtained by crossing Ogg1+/– and Mutyh+/– mice to obtain Ogg1–/–Mutyh–/– (Ogg1/Mutyh-DKO), and these were inbred for 2 to 3 generations to obtain the mice used in each experiment. C57BL/6J mice were used as the WT strain. All animals were maintained in an air-conditioned, specific pathogen–free room with time-controlled lighting.
Ten-week-old male mice were used. 3-NP (Sigma-Aldrich) was dissolved in PBS and systemically administered via an i.p. implanted osmotic minipump (1007D, Alzet; delivering 0.5 μl/h for 7 days). All animals were subjected to surgical operations between 9:00 am and 12:00 pm. Mice were anesthetized with pentobarbital (30 mg/kg, i.p.), an incision was made in the abdomen, and the osmotic minipump was implanted into the i.p. cavity. In typical experiments, mice were exposed to 120 mg of 3-NP/kg/d for 7 days, and the control mice were subjected to implantation of the osmotic minipump containing PBS for 7 days. To compare 3-NP resistance between Mutyh-KO and WT mice, a higher dose of 3-NP (150 mg/kg/d) was applied.
A calpain inhibitor (MDL28170, Biomol) or a poly-(ADP-ribose) polymerase inhibitor (3-AB; Sigma-Aldrich) was administered together with 3-NP to Mth1/Ogg1-DKO mice. Two osmotic minipumps prefilled with 3-NP (120 mg/kg/d) and MDL28170 (50 mg/kg/d, dissolved in dimethyl sulfoxide, then diluted in PBS) or 3-AB (30 mg/kg/d, dissolved in PBS) were simultaneously implanted for 7 days. Control mice were implanted with 2 minipumps prefilled with 3-NP (120 mg/kg/d) and PBS.
3-NP–induced motor symptoms were evaluated using a quantitative neurological scale based on previous reports (29, 30). Neurological scores were defined as follows: score = 0, normal behavior; 1, general slowness of displacement resulting from mild hind limb impairment; 2, incoordination and marked gait abnormality; 3, hind limb paralysis; 4, incapacity to move resulting from forelimb and hind limb impairment; 5, recumbency.
Spontaneous locomotor activity in the home cage was measured using a digital counter with an infrared beam sensor (NS-ASS01; Neuroscience, Inc.), placed about 200 mm above the center of the home cage (27 × 14 × 15 cm), and analyzed using DAS-008 software.
To evaluate the extent of motor impairment induced by 3-NP exposure, an open-field test was performed. Behavioral analysis was performed between 9:00 am and 12:00 pm. A mouse was placed in the center of a circular open field (60 cm in diameter, 30 cm deep). Motor activity was quantified using an automated video tracking system, the video-image motion analyzer AXIS-90 (Neuroscience Inc.). Traveling distance and numbers of rearing behaviors were measured for 5 minutes. To avoid a reduction in motor activity due to habituation to the open field, motor activity was recorded only twice, on the day before implantation of the osmotic minipump (baseline performance) and on the seventh day after the operation. The number of animals examined is shown in each figure legend. Relative performance indicates the traveling distance or number of rearing times obtained on the seventh day after implantation of an osmotic minipump as a percentage of that obtained on the day before implantation.
Rotarod testing was performed between 9:00 am and 12:00 pm. The test was performed using an accelerating rotarod for mice (Model 7650; UGO Basile). Prior to 3-NP treatment, all mice were given 3 trials on the rotarod, which was accelerated from 4 to 40 rpm over a 300-second period and kept at the top speed for more than 100 seconds each day for 3 consecutive days. On the third day, latency to fall from the rotarod was measured as a baseline performance. On the seventh day after implantation of the minipump, mice were tested on the rotarod, and latency to fall off was measured after 3-NP treatment.
The primary antibodies used were mouse monoclonal anti-8-oxo–dG (N45.1, JaICa, Fukuroi); rabbit anti-DARPP32 (Cell Signaling Technology); mouse anti-DARPP32 (BD Transduction Laboratories); rat anti-F4/80 (Oxford); goat anti-TFAM (Santa Cruz Biotechnology Inc.); mouse anti–α-spectrin (Chemicon); mouse anti-AIF (Santa Cruz Biotechnology Inc.); mouse anti-PAR (Trevigen), rabbit anti-ssDNA (Immuno-Biological Laboratories), rabbit anti-calbindin D-28K (Chemicon); rabbit anti-GFAP (Dako); rabbit anti-Hsp60 (Abcam); rabbit anti-OGG1, which recognizes the conserved region of human OGG1 proteins (52); rabbit anti-MTH1 (NB100-109, Novus Biologicals); and rabbit anti-MUTYH, which was raised against recombinant human MUTYH (53). The ssDNA antibody recognizes single-stranded regions of DNA with a length of at least 6 deoxynucleotides, which can be generated by SSBs or DSBs with resection of the ends (45, 54).
Tissue preparation and immunohistochemical analyses were performed as described previously (31). Brains were cut on a cryostat at a thickness of 40 μm and collected as free-floating sections in PBS; these were then processed immediately for immunohistochemistry (IHC). Free-floating sections were subjected to appropriate pretreatment and reaction with a primary antibody and then processed using a Vector ABC kit (Vector Laboratories) with the proper biotinylated secondary antibody. Then, the 3,3′-diaminobenzidine (DAB)/nickel (Vector Laboratories) reaction was used to visualize the bound secondary antibody. Some sections were counterstained with cresyl violet. Digital images were acquired using an Axioskop2 Plus microscope, equipped with an AxioCam CCD camera, and Axiovision 3.1 imaging software (Carl Zeiss Inc.) was used.
Free-floating sections incubated with the appropriate primary antibody were then further incubated with a proper Alexa Fluor–labeled secondary antibody (Invitrogen) in a solution containing DAPI (0.05 μg/ml, Sigma-Aldrich). Confocal images were acquired using a confocal laser microscopy system (LSM 510META-V3.2 software; Carl Zeiss Inc.) attached to an inverted microscope (Axiovert 200M). Images of the immunostained sections were taken within 48 hours.
For immunodetection of 8-oxoG in nDNA or mtDNA, free-floating sections were pretreated as described previously (31, 32). Briefly, to detect 8-oxoG in the mtDNA, sections were pretreated only with RNase (5 mg/ml; Sigma-Aldrich) and were directly subjected to IHC with the anti-8-oxo–dG antibody. To detect 8-oxoG in nDNA, RNase-treated sections were further pretreated with 2N HCl to denature the nDNA, thus enabling the antibody to access 8-oxoG in nDNA. These pretreated sections were subjected to IHC with the anti–8-oxo–dG antibody. The intensity of 8-oxoG IR in the whole striatum was measured in each digital image using ImageJ 1.42 (NIH). Pretreated sections reacted with the anti–8-oxo–dG as described above were also subjected to immunofluorescence microscopy with a proper Alexa Fluor–labeled secondary antibody. The mitochondrial 8-oxo–dG index in a single neuron was measured in each laser confocal image of the immunofluorescence samples using ImageJ. Twenty neurons were measured to obtain the mean mitochondrial 8-oxo–dG index for each animal.
Ten-week-old male WT and Mth1/Ogg1-DKO were subjected to PBS or 3-NP administration. The mice were euthanized by cervical dislocation, and the striata were immediately removed and frozen in liquid nitrogen. The tissues were stored at –80°C and analyzed within 7 days of collection. No difference in the level of 8-oxo–dG was observed between tissues stored at –80°C and freshly isolated tissues. DNA was extracted from striata using the DNA Extractor TIS Kit (No. 296-67701, Wako) as described in the instructions for the kit, with a few modifications: 10 mM 2, 2, 6, 6-tetramethylpiperidine-N-oxyl (TEMPO, Wako) and 100 μM 2′-deoxycoformycin, an adenosine deaminase inhibitor provided by the Chemo-Sero-Therapeutic Research Institute (Kumamoto, Japan), were added to all reagents at all stages of manipulation. To allow for accurate quantitation of 8-oxo–dG and to account for the signal variability of the mass spectrometer, [15N5]-labeled 8-oxo–dG (NLM-6715-0, Cambridge Isotope Laboratories Inc.) was added as a stable isotope internal standard prior to enzymatic hydrolysis of the DNA samples. The stable internal standards [15N5] 8-oxo–dG, sodium acetate buffer (ABI, #400884, pH 4.5), nuclease P1 (Yamas), and acid phosphatase (P-0157; Sigma-Aldrich) were added to the DNA samples; samples were then transferred to filter tubes (Ultrafree-MC probind 0.45 μm; Millipore), and centrifuged at 12,000 g for 3 minutes. The digested samples were subjected to LC-MS/MS analysis using a Shimadzu VP-10 HPLC system connected to an API3000 MS/MS system (PE-SCIEX), and the ratios of 8-oxo–dG to 106 dG in the samples were determined as described previously (55).
For detection of mitochondrial cytochrome c oxidase activity in frozen sections, brain sections (thickness, 40 μm) were prepared from WT and Mutyh-KO mice administered PBS or 3-NP (150 mg/kg/d) for 7 days. Brain sections were incubated for 3 hours at 37°C in a solution containing cytochrome c (0.3 mg/ml), DAB (0.5 mg/ml), and sucrose (40 mg/ml) in PBS. Then sections washed in PBS were dehydrated with ethanol, immersed in xylene and mounted on slide glass, and digital images were acquired. The intensity of DAB staining cytochrome c oxidase activity in the whole striatum was measured in each digital image using ImageJ 1.42.
The densities of neurons and glial cells in the dorsolateral striatum (bregma, 0.86 mm to approximately 0.62 mm) (56) were estimated by counting those on cresyl violet–stained sections. Neurons were identified as large cells with typical morphological features including a round, oval, or polygonal shape, and glial cells were identified as cells with profiles distinguished by their small, round, and hyperchromatic nuclei. The density of MSNs in the dorsolateral striatum was estimated by counting DARPP32-positive cells.
Striata were isolated from Mth1/Ogg1-DKO or WT mice on the fourth and seventh days of exposure to 3-NP or on the seventh day of exposure to PBS. Mouse striata were homogenized in ice-cold lysis buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1.0% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitor mixture (Nacalai Tesque). The homogenates were then centrifuged at 89,000 g for 30 minutes at 4°C, and the supernatants were collected. Western blot analysis was performed according to previously described methods (57, 58). Protein concentration in the supernatant was measured using a DC protein assay kit (Bio-Rad). Equal amounts of total protein (5 μg) were separated by SDS-PAGE. Anti–α-spectrin (1:10,000), anti-AIF (1:500), and anti–β-actin (1:10,000) antibodies were used. The intensity of each band was quantified using an LAS1000 image analyzer and Image Gauge, version 3.2 software (Fujifilm).
For transmission electron microscopy, mouse brains were prefixed with 4% paraformaldehyde and 2.5% glutaraldehyde (TAAB Laboratories), washed with a cacodylate buffer, postfixed in osmium tetroxide (Nacalai Tesque), dehydrated in ascending concentrations of alcohol and uranyl acetate (MERCK), and embedded in epoxy resin (TAAB 812 resin. TAAB Laboratories). Ultrathin sections (100 nm) were cut, and examined under a transmission electron microscope (Tecnai 20; FEI Corp.). For electron microscopic immunocytochemistry to detect MUTYH protein, fixed mouse brains were incubated in 2.3 M sucrose and then in 20% polyvinylpyrrolidone (Nacalai Tesque) prepared in 1.8 M sucrose. Ultrathin sections (100 nm) were cut, blocked in 3% BSA, incubated with rabbit anti-MUTYH antibody, then with a secondary antibody, anti-rabbit IgG-gold (10 nm; EY Lab.). Sections were post-fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, mounted with 0.1% uranyl acetate in 2% polyvinyl alcohol (Sigma-Aldrich), and examined under transmission electron microscope.
Statistical analyses were conducted using JMP 9.02 (SAS Institute Inc.). Data are presented as least square (LS) means ± SEM. To assess statistical significance, we performed 1-way ANOVA for effects of treatment or 2-way ANOVA with interaction for effects of genotype and treatment, and interactions between genotype and treatment; results obtained by standard LS fits are shown. For multiple comparison tests, 1-way ANOVA analyses were followed by a post hoc Hsu’s MCB test, while 2-way ANOVA analyses were followed by a post hoc Tukey’s HSD test. Student’s t test was applied for individual pairwise comparisons in all cases. The threshold for statistical significance was P < 0.05.
The handling and killing of all animals were done in accordance with national prescribed guidelines, and ethical approval for the studies was granted by the Animal Care and Use Committee of Kyushu University.
This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (20013034), the Japan Society for the Promotion of Science (18300124, 19390114, 22221004, 22501014), and the Kyushu University Global COE program (Cell-rate Decision Function and Dysfunction in Homeostasis). We thank Yasunari Sakai, Dongchon Kang, Mami Noda, and Mizuki Ohno for helpful discussions; Ryo Ugawa (Laboratory for Technical Support, Medical Institute of Bioregulation, Kyushu University, Kyushu, Japan) for electron microscopy; and Setsuko Kitamura, Noami Adachi, Kazumi Asakawa, and Kaoru Nakabeppu for their technical assistance.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2012;122(12):4344–4361. doi:10.1172/JCI65053.