|Home | About | Journals | Submit | Contact Us | Français|
Increased oxidative stress and concordant DNA methylation changes are found during aging and in many malignant processes including prostate cancer. Increased oxidative stress has been shown to inhibit DNA methyltransferase in in vitro assays, but whether this occurs in vivo is unknown. To generate increased oxidative stress we utilized mice containing mutations in the CuZnSOD (Sod1) gene, a major superoxide dismutase in mammals. Increased 8-hydroxy-2′-deoxyguanosine, an adduct indicating oxidative damage, was found in liver and prostate tissues at 2 and 12 mo Sod1+/− mice compared to controls. prostate tissues from Sod1+/− mice demonstrated decreased weight at 2 mo compared to controls, but this difference was not significant at 12 mo. histologic changes were not seen. Global DNA methylation was significantly decreased at 2 mo in the prostate in Sod1+/− mice. 11p15 containing the epigenetically modulated insulin-like growth factor 2 (Igf2) and H19 genes, both which display oncogenic functions, may be particularly sensitive to oxidative stress. CpG island methylation at an intergenic CTCF binding site and the Igf2 P3 promoter was decreased in Sod1 mutants compared to controls
This is the first in vivo study to show that a deficiency of Sod1 leads to a decrease in DNA methylation. These studies indicate that increased oxidative stress, a factor implicated in neoplasia, can induce DNA hypomethylation in prostate tissues.
DNA methylation plays a critical role in maintaining an appropriate pattern of gene expression during development and aging. Perturbations in epigenetic factors, including DNA methylation, histone modification and genomic imprinting, can contribute to the development of many diseases including diabetes, heart disease and cancer.1–3 There is mounting evidence that environmental exposures to chemical and nutritional factors alter CpG methylation and other epigenetic modifications at specific loci.4 One putative mechanism linking these environmental factors with alterations in DNA methylation is via oxidative stress. In vitro, increased oxidative stress has been shown to result in DNA hypomethylation.5,6 Reactive oxygen species (ROS) associated with oxidative stress result in the conversion of deoxyguanine in CpGs to 8-hydroxy-2'-deoxyguanosine (8-OHdG) and this adduct blocks DNA methyltransferase function at neighboring 5' cytosines.5,6 To date, however, in vivo data is lacking, linking increased oxidative stress with changes in DNA methylation and other epigenetic processes.
A disturbance in the equilibrium between ROS and detoxifying anti-oxidant systems leads to augmented oxidative stress. Oxidative stress-induced superoxide radicals (O2−) are generated during normal metabolic processes in all oxygen utilizing organisms.7 ROS originate both endogenously, produced during normal cellular respiration, and exogenously. Exogenous sources of ROS include various carcinogens such as estrogenic compounds, redox cycling compounds, metals, radiation and chemotherapeutic agents. Increased levels of oxidative stress are associated with augmented damage to proteins, DNA and other cellular constituents.8 The harmful effects of high concentrations of superoxide radicals have been implicated in the aging process9 and in the genesis of a number of diseases including prostate cancer.10
Aging itself is closely linked with a shift in the prooxidant- antioxidant balance of many tissues, including the prostate, toward an oxidative state with reactive oxygen species damage. Inflammation and histologic lesions such as proliferative-inflammatory atrophy (PIA) occur with increased frequency in the aging prostate.11 When compared to other human tissues, the prostate expresses remarkably high levels of lipofuscein, a breakdown product of oxidatively damaged proteins.12,13 Oxidative damage to DNA can be measured by the accumulation of nuclear 8-OHdG and an accumulation of this adduct occurs in aging prostate tissues.14 Associated with this increased oxidative DNA damage are alterations in DNA methylation including genome-wide hypomethylation,15 as well as altered methylation at specific CpG islands.16
One of the major anti-oxidant defense systems that cope with the inevitable production of oxygen free radicals are superoxide dismutases (SODs). In mammals, there are two forms of intra-cellular superoxide dismutases: CuZnSOD (Sod1) which is primarily located in cytosol and mitochondrial intermembrane space17,18 and MnSOD (Sod2) localized in the mitochondrial matrix. CuZnSOD (Sod1) is the most abundant isoform, accounting for the majority of SOD activity in the cell and responsible for removing superoxide radicals generated in cytoplasm and nucleus. Mice homozygously deficient for CuZnSOD (Sod1) show reduced lifespan and neoplastic changes in the liver.19 These Sod1 deficient mice also demonstrate extensive oxidative damage in the cytoplasm as evidenced by increased levels of 8-oxodG and F2-isoprostanes.20
Since oxidative damage is associated with DNA hypomethylation in vitro, we utilized Sod1 deficient mice to test whether increased oxidative stress can induce changes in DNA methylation globally or at specific CpG islands in prostate tissues. The Igf2-H19 region is a well-characterized target and appears susceptible to alterations in methyl-deficient diets21 or methylation inhibition.22 Our demonstration of both global and regional decreases in DNA methylation in Sod1 deficient animals highlight the plasticity of epigenetic controls and provide a link between oxidative stress and DNA methylation in the prostate.
Congenic C57BL/6J mice containing a mutation in Sod1,20,23 were utilized to examine the impact of increased superoxide production on epigenetic changes in the prostate. Prostate tissues from Sod1+/− animals demonstrate roughly half the SOD1 activity that wildtype animals manifest using a previously described gel-based superoxide reporter assay (data not shown).24 Similar activity results have been demonstrated in kidney, liver and other tissues.19 Given the high rate of liver tumor formation, weight change and embryonic lethality in Sod1−/− animals,20 we focused our studies on animals with the heterozygous mutation (Sod1+/−).
Genitourinary tissues were harvested from animal cohorts at early (2 mo) and late timepoints (12 mo). Grossly, examination of ventral prostate (VP), dorsolateral prostate (DLP), coagulating glands and seminal vesicles revealed no macroscopic abnormalities in Sod1+/− mice. No liver tumors were noted in the older cohorts, nor were differences in body weights observed (p = 0.5). VP and DLP weights, corrected for overall body weight, were significantly lower in Sod1+/− mice when compared to Sod1 wild-type mice (p < 0.05; Fig. 1). These differences in VP and DLP weights between Sod1 heterozygous mice and wild-type mice were not observed at 12 mo. Histological analysis of Sod1+/− prostate tissues revealed no significant tissue changes either at 2 or 12 mo timepoint when compared to Sod1 wild-type mice prostates (data not shown).
Several indices of oxidative damage were measured in the tissues of Sod1 wild-type and mutant mice. Increased superoxide has been demonstrated in Sod1+/− tissues by elevated cytosolic aconitase, 8-oxo-dG and F2-isoprostanes.20,25 To confirm these changes in the genitourinary systems of Sod1 mutant mice, we utilized a highly-sensitive 8-OHdG ELISA Kit on DNA extracted from prostate tissues. We first analyzed liver tissues and found 8-OHdG levels were 61% higher in 2 mo mutant mice compared to wildtype animals (0.21 ± 0.01 vs. 0.13 ± 0.02 respectively; p = 0.03). Less marked differences in 8-OHdG levels (24%) were seen in the livers of 12 mo mutant Sod1 mice (0.31 ± 0.01 vs. 0.25; ± 0.01; Sod1+/− and +/+ respectively; p = 0.01; Fig. 2A). In prostate tissues from 2 mo animals, 8-OHdG levels increased 10% (0.44 ± 0.01 vs. 0.40 ± 0.02; Sod1+/− and +/+ respectively; p = 0.02) and 35% at 12 mo (0.54 ± 0.02 vs. 0.40 ± 0.06; Sod1+/− and +/+ respectively; p = 0.03; Fig. 2B).
As a second measure to detect oxidative stress-induced DNA lesions, we employed a quantitative extra-long PCR (XL-PCR) approach.26–28 Decreased amplification ability, signifying DNA damage, was significantly reduced in nuclear DNA samples from 2 mo Sod1 mutant compared to wild-type (1.56 ± 0.31 vs. 2.47 ± 0.25; Sod1+/− and +/+ respectively; p = 0.03; Fig. 3). Relative amplification ratios were reduced in 12 mo Sod1+/− DNA (2.3 ± 0.77 vs. 3.83 ± 1.10; Sod1+/− and +/+ respectively; p = 0.14). These assays confirm the presence of increased oxidative damage in prostate tissues from mice with Sod1 mutation. These changes were more marked in younger mice.
Given data linking oxidative damage and 8-OHdG formation to the inhibition of DNA methyltransferase in vitro,6 we investigated whether global changes in DNA methylation occur in Sod1 deficient mice. Overall DNA methylation was quantified using an ELISA assay employing anti-methylated cytosine antibodies (Fig. 4). When Sod1+/− was compared to Sod1+/+, global DNA methylation was significantly reduced in mouse VP tissues (1.67 ± 0.21 vs. 2.13 ± 0.04; p = 0.05; Fig. 4A) and DLP tissues (0.49 ± 0.06 vs. 0.66 ± 0.02; p = 0.02; Fig. 4B) in 2 mo animals. A reproducible trend towards decreased global DNA methylation was also observed at 12 mo in Sod1+/− tissues, but this did not reach significance (VP 2.19 ± 0.19 vs. 2.40 ± 0.12, p = 0.20; DLP 0.59 ± 0.10 vs. 0.68 ± 0.01, p = 0.21; Sod1+/− versus +/+ respectively).
To further interrogate methylation at specific loci known to be susceptible to diet and other environmental factors,21,29 a methylation-specific quantitative PCR assay was performed. Interrogation of four previously identified loci in the Igf2-H19 region (Igf2 P3, H19 promoter and several CTCF binding sites within the imprint control region; Fig. 5) was performed. In 2 mo Sod1+/− mice, two of the regions (Igf2 P3 and CTCF 3) were hypo methylated (p < 0.05). At 12 mo, Sod1+/− mice showed significant hypomethylation at CTCF 3 (Fig. 5). To confirm the methylation alterations in this gene, bisulfite subclones of the gene were obtained. No GC:AT mutations were observed in the subclones, indicating that the results were not due to the effects of mutations on primer binding efficiency (data not shown).
While increased oxidative stress and altered methylation are common observations in the aging prostate and other human and mammalian tissues, data linking these two processes has been lacking. One relevant approach to study the effects of free radicals is to perturb the enzymatic machinery responsible for their metabolism. The major contributor to superoxide dismutase activity, CuZnSOD (Sodl/SODl), is a cytoplasmic enzyme that is produced constitutively and present in all cells, including the prostate epithelium.30–32 A decrease in the oxidative defenses is refected by levels of superoxide dismutase declining with age in both the human and the rat prostate.31,33 We recapitulated this physiologic situation by employing a well-described mouse model containing a Sod1 mutation to generate increased levels of oxidative stress. We demonstrate for the first time in vivo that inducing oxidative stress leads to perturbation of DNA methylation reflected by both global and regional losses in methylation.
Oxidative damage to DNA may be measured by the accumulation of nuclear 8-hydroxydeoxyguanosine (8-OHdG), a DNA base lesion. Consistent with previous reports,20 8-OHdG modifications were significantly higher in the liver of Sod1+/− mice compared to Sod1 wild-type mice at both the 2 and 12 mo timepoints. 8-OHdG levels were also significantly increased in prostate tissues of Sod1+/− mice (p ≤ 0.05) when compared to wild-type animals. This surrogate marker of increased oxidative stress also accumulates in human prostate with aging,14,34 cancer35 and in several mouse prostate cancer models (e.g., TRAMP 36 and NKX3.137). In addition to effects on methylation in vitro,5 the oxidized DNA base 8-OHdG leads to G:C to T:A transversion mutations prevalent in inactivated oncogenes and tumor suppressor genes38,39 which may contribute to neoplastic transformation.40 Because of limitations in mouse prostate tissues, the closely associated, embryologically-similar coagulating glands were also analyzed using another marker of oxidative stress, oxidized F(2)-isoprostane levels and increased levels noted in Sod1 deficient mice (data not shown). These studies confirm increased ROS and oxidative stress in the prostate and genitourinary tissues of Sod1 deficient animals.
Given in vitro reports linking oxidative stress and 8-OHdG formation to DNA methylation inhibition,6,41 we evaluated global methylation changes in prostate tissues in response to the loss of an Sod1 allele. Global DNA methylation was significantly reduced in VP and DLP tissues of 2 mo Sod1 deficient mice (p ≤ 0.05; Fig. 4). A trend towards decreased global DNA methylation was seen at 12mo in Sod1 deficient mice (p = 0.19). With regard to the analysis of 12mo mice, increased variation in methylation between aging experimental animals was noted and may represent some compensatory mechanisms for reducing oxidative stress in some older animals. Alternatively, the effect of oxidative stress on DNA methylation may be more prevalent in young, developing animals. Previous studies suggest that dietary modulation of DNA methylation is more pronounced in younger animals.42
We also quantitated methylation levels within specific loci known to be susceptible to modulation by environmental factors.21,29 Several CpG islands examined within the imprinted Igf2-H19 region demonstrated hypomethylation in Sod1 deficient mice. This methylation loss was seen both at the biologically significant binding region CTCF 3 (within the imprint control region), as well as in the Igf2 P3 promoter CpG island. Methylation loss (and some gains) occur within both these regions with methyl-deficient diets in mouse prostate tissues21 and livers.43 Changes in gene expression have not been as clearly linked to hypomethylation at these and other CpG islands as hypermethylation has.44 Using a limited number of animals (three in each group) we did find a decrease in Igf2 expression in Sod1+/− mice at 2 mo (0.9-fold; p = 0.01) and an increase at 12 mo (2.4-fold; p = 0.02). These data suggest other transcription factors, possibly regulated by methylation, may play a greater role in the regulation of these genes. Hypomethylation at Igf2-H19 has been implicated in gene expression changes at other loci and can influence other genes through long-range chromatin effects.45
The presence of 8-OHdG in CpG dinucleotide sequences has been shown to strongly inhibit methylation of adjacent cytosine residues5,6 and interfere with the ability of restriction nucleases to cleave DNA.46 In folate and methyl-deficient rats, DNA adducts are less efficiently methylated and an increase in DNA strand breaks proceeds DNA hypomethylation.47 The hypomethylation seen in the current study could be a direct result of DNA methyltransferase inhibition by DNA adducts. Alternatively, Sod1 deficiency can formally affect protein oxidation resulting in loss of DNA methyltransferase activity or changes in the interaction of chromatin binding proteins with DNA.48 The extent of either of these effects was not determined. For many of the experiments in our study, we were limited by the amount of mouse prostate tissue available for additional assays, and also by biological variation present within the mice of each genotype especially notable at older ages.
Difficulty in generating Sod1−/− was encountered and animals were able to be weaned at only a rate of 10% of expected. Also, it has been reported that more that 70% of Sod1−/− mice develop liver tumors at older ages.20 Because of this high rate of malignancy, we focused on studying epigenetic changes in Sod1 heterozygous mice which demonstrate a 50% loss of Sod1 activity. We did not see any phenotypic effects in response to genome-wide hypomethylation in the prostate. No macroscopic abnormalities were observed in genitourinary tissues including the VP, DLP, CG and seminal vesicles in Sod1 mutant mice. We postulate that to see a phenotypic effect, a threshold level of methylation change must be reached beyond the ~20% decrease we generated associated with a 50% SOD1 deficiency. In mice carrying a hypomorphic DNA methyltransferase 1 (Dnmt1) allele, which reduces Dnmt1 expression to 10% of wild-type levels, a substantial genome-wide hypomethylation occurs in all tissues. These mice develop aggressive T-cell lymphomas.49 Future studies will address the functional significance of oxidative-stress induced DNA hypomethylation on prostate cancer susceptible mice.
In conclusion, mice deficient in Sod1 show elevated levels of oxidative stress, as evidenced by increased levels of 8-OHdG levels, as well as global and locus specific hypomethylation in the prostate. This supports in vitro studies indicating DNA base modifications have the ability to interfere with DNA as a substrate for the DNA methyltransferases resulting in global hypomethylation.50 Increased oxidative stress has been proposed as an important factor in the development of cancer in solid organs, and the prostate appears to be especially sensitive.51 Our results suggest a link between elevated oxidative stress, a factor commonly seen in prostate tissues with aging, and hypomethylation also seen in aging and tumorigenesis. Oxidative DNA damage appears to affect patterns of DNA methylation which in turn could lead to aberrant gene expression and contribute to the development of malignancy.41 In compound mutant mice lacking Nkx3.1 and Pten, reduced expression of CuZnSOD and MnSOD is noted and associated with increased oxidative DNA damage.37 These mice develop prostate cancers. A distinguishing feature of epigenetic changes, when compared to genetic changes, is that they are reversible and are attractive targets for therapy.52 These studies in combination with the present data provide further impetus for chemoprevention trials aimed at reducing oxidative stress and suggest DNA methylation as an additional measure of the impact of these agents.
Sod1−/− and the corresponding Sod1+/− mice were generated from a cross between B6 mice. The animals were housed in the Veterans Administration/University of Wisconsin Shared Aging Rodent Facility. Mouse body weights and food consumption were continuously monitored. Difficulty in generating Sod1−/− was noted as animals were able to be weaned only at a rate of 10% of the expected. Mouse organs were then harvested at 2 and 12 mo timepoint and specific assays performed. All studies were performed under institutional approval and conformed to the humane treatment of animals in the conduct of scientific studies.
To determine the effects of Sod1 mutation on the liver and genitourinary systems of Sod1 mutant mice, we used the High sensitivity 8-OHdG ELISA Kit (Japan Institute for the Control of Aging, Fukuroi, Japan) to measure levels of 8-OHdG in DNA extracted from the mouse liver and ventral prostrates. The 8-OHdG test is a competitive in-vitro enzyme-linked immunosorbent assay for the quantitative measurement of 8-OHdG in tissue. The test was performed according to the manufacturer’s instructions and the absorbance read at 450 nm on a SpectraMax Plus 384 microplate reader (Molecular Devices, Sunnyvale, CA, USA).
Oxidative stress induced nuclear DNA damage was measured using qPCR as described previously.26–28 Briefly, total DNA was isolated from mutant and wild-type Sod1 mice VP using DNeasy Blood & Tissue Kit (Qiagen, Valencia, CA, USA). DNA was quantified using PicoGreen (Invitrogen, Carlsbad, CA, USA), and qPCR was performed with identical amounts of input DNA. Specific primers were used to amplify a large fragment of β-globin (10.0 kb) to determine nuclear DNA damage. A small 200 bp fragment of β-globin gene was amplified to monitor changes in nuclear DNA copy number and to nor. malize the data obtained when amplifying the 10.0 kb fragment. Relative amplifications were calculated comparing Sod1 heterozygous mutants to wild-type Sod1 mice VP DNA; these values were used to estimate mathematically the number of lesions present in DNA, assuming a Poisson distribution.26–28 The data shown represent results from four to six animals per group done in duplicate qPCR reactions for each target DNA region were performed; the standard error of the mean is indicated. Statistical significance was evaluated with Student’s unpaired t-test.
Ventral and dorsolateral prostate DNA was extracted using DNeasy Blood & Tissue Kit (Qiagen, Valencia, CA, USA) and extracted DNA was quantified using Quant-iT™ PicoGreen dsDNA quantification kit (Invitrogen, Carlsbad, CA, USA). For each sample, methylation analysis was performed in triplicate aliquots (100 ng DNA each) using anti-methylated cytosine antibody-based Methylamp™ Global DNA Methylation Quantification Ultra Kit (Epigentek, New York, NY, USA). We used methylated DNA standards supplied with the kit, while for blanks were omitted DNA from the assay. The spectroscopic end point (optical density, OD) was read on SpectraMax Plus 384 microplate reader (Molecular Devices, Sunnyvale, CA, USA). After subtracting blank readings from the readings for both the sample and the standard, the value of DNA methylation for each sample was calculated as a ratio of sample OD relative to the OD of the standard. The triplicate values obtained for each tissue from each animal were averaged to give a single data point for the animal. Statistical analysis was conducted using a Student’s unpaired t-test.
Quantification of methylation at specific loci within the Igf2-H19 region was determined by the Normalized Index of Methylation (NIM) MS-qPCR method, as described.21,53 Briefly, 1 or 2 µg of mouse DLP genomic DNA was treated with sodium bisulfite using CpGenome Kit (Chemicon, Temecula, CA, USA), according to the manufacturer’s directions. An SssI hypermethylated standard curve was generated for bisulfite treated DNA using methylation specific primers, and also using control non-CpG-containing MyoDI specific primers. Based on this curve, the copy number of methylated samples in the samples was determined, and the NIM was calculated. Primer sequences used in this experiment have been described before.21
Comparison of 8-OHdG, oxidative DNA damage qPCR and global, local DNA methylation changes experiments were performed using Student’s t-test with values expressed as means ± standard errors with p values less than 0.05 considered significant between groups.
This work was supported by the National Institutes of Health (R01CA97131) and the University of Wisconsin George M. O’Brien Urology Research Center (1P50DK065303) and the John Livesey endowment. J.R.D. is supported through an NIH training grant (T32 AG000213-16) to the Biology of Aging and Age-related Diseases Training Program.