Individual ratios of 8-hydroxyguanine in intact DNA to free 8-hydroxyguanine for subjects with Alzheimer disease and control subjects plotted on a logarithmic scale. The lowest ratio for a subject with Alzheimer disease is 3.5 times higher than the highest control ratio for a control subject.
Lovell MA, Markesbery WR. Ratio of 8-Hydroxyguanine in Intact DNA to Free 8-Hydroxyguanine Is Increased in Alzheimer Disease Ventricular Cerebrospinal Fluid. Arch Neurol. 2001;58(3):392–396. doi:10.1001/archneur.58.3.392
Markers of oxidative stress are increased in cerebrospinal fluid (CSF) of patients with Alzheimer disease (AD), although none of those reported are appropriate diagnostic markers because of the overlap between patients with AD and control subjects.
To determine the ratio of 8-hydroxyguanine (8-OHG) levels in intact DNA to free 8-OHG in the ventricular CSF of patients with AD and age-matched control subjects. The most prominent marker of DNA oxidation is 8-OHG.
Free 8-hydroxy-2′-deoxyguanosine (8-OHdG) was isolated from ventricular CSF taken at autopsy from 18 subjects with AD and 7 control subjects using solid-phase extraction columns. Levels were measured as the hydrolysis product, 8-OHG, using gas chromatography/mass spectrometry with selective ion monitoring. Intact DNA was isolated from the same CSF and the levels of 8-OHG were determined in the intact structures. Stable-labeled 8-OHG was used for quantification.
A statistically significant (P<.05) 108-fold increase in the ratio of 8-OHG in intact DNA to free 8-OHG was observed in patients with AD. Analysis of the data distribution indicated that the lowest AD ratio was 3.5 times higher than the highest control ratio; there was no overlap of the 2 populations.
Although the data for each individual measurement demonstrates overlap between patients with AD and control subjects, the ratio of 8-OHG intact in DNA to free 8-OHG demonstrates a delineation between patients with AD and control 8-OHG subjects and may be useful as a marker of disease progression or the efficacy of therapeutic antioxidant intervention.
INCREASING evidence supports the role of oxidative stress in the pathogenesis of neuronal degeneration in several neurological disorders, including stroke, amyotrophic lateral sclerosis, Parkinson disease, head trauma, and Alzheimer disease (AD). Several studies show that the AD brain has the potential for increased oxidative stress due to elevations of brain iron levels, particularly redox active iron.1,2 Studies of oxidative damage show increased levels of protein oxidation, lipid peroxidation, and markers of lipid peroxidation, including 4-hydroxynonenal2 and acrolein,3,4 F2-isoprostanes,5 and F4-neuroprotanes6 in AD. Increased levels of 4-hydroxynonenal, F2-isoprostanes, and F4-neuroprostanes are present in ventricular cerebrospinal fluid (CSF) in AD. Markers of oxidative stress are present in neurofibrillary tangles and senile plaques in the brain in AD. Two markers of oxidative stress in nuclear and mitochondrial DNA are increased in normal aging7 and in AD.8- 10 Attack of DNA by reactive oxygen species, specifically the hydroxyl radical, leads to the hydroxylation of DNA bases,11 the most prominent of which is 8-hydrodeoxyguanine (8-OHdG). Levels of 8-OHdG are elevated in mitochondrial DNA in the cerebral cortex of patients with AD compared with controls.10 Our laboratory demonstrated statistically significant (P<0.5) elevations of 8-hydroxyguanine (8-OHG) as the hydrolysis product of 8-OHdG, 8-hydroxyadenine, and 5-hydroxyuracil in the parietal, temporal, and frontal lobes in patients with AD compared with age-matched control subjects.8 Of the 6 oxidatively modified base adducts analyzed, 8-OHG demonstrated the highest absolute levels, indicating its prominence as a marker of oxidative stress in AD. Consistent with observations of elevated 8-OHG in the AD brain, we recently demonstrated a statistically significant (P<0.5) decrease in levels of base excision repair enzymes responsible for the repair of oxidized guanine.12 In another study, we showed statistically significant (P<0.5) elevations of 8-OHG in DNA extracted from ventricular CSF in patients with AD compared with age-matched control subjects.13 Analysis of levels of free 8-OHG, resulting from excision from damaged DNA by base-specific glycosylases, demonstrated a statistically significant depletion of the free repair product in AD CSF. Although mean levels of free and intact 8-OHG were significantly (P<0.5) different in 8-OHG AD subjects compared with control subjects, overlap of the data showed that the individual measures alone were not appropriate as diagnostic markers of AD.
This study demonstrates that the ratio of levels of 8-OHdG in intact DNA to free 8-OHG in ventricular CSF removed at autopsy (determined using stable-labeled 8-OHG and gas chromatography/mass spectrometry with selective ion monitoring) is statistically significantly elevated in AD subjects. The data distribution suggests that the ratio delineates AD and control subjects. The ratio also suggests that AD subjects are subject to a dual insult of increased oxidative stress and decreased repair capacities. The combination of these factors (rather than either individually) may contribute to the neurodegeneration observed in the brain in AD.
Approximately 20 to 40 mL of CSF was removed from the lateral ventricles at autopsy using an 18-gauge spinal needle and virgin polyethylene syringes from 18 AD subjects (7 men, 11 women) and 7 control subjects (4 men, 3 women). Data for 7 AD subjects and 5 control subjects are from a previous article13 and are combined with the results of analysis of intact and free 8-OHG from 11 additional AD subjects and 2 additional control subjects. Taken separately, data from the new subjects confirm our prior findings. Demographic data for all subjects are shown in Table 1. The samples were immediately centrifuged to remove particulate matter, and the supernatant was placed in fresh polyethylene tubes and stored at −80°C until used for analysis. The mean ± SEM age was 79.9 ± 2.5 years for AD subjects and 80.0 ± 2.6 years for control subjects. The mean ± SEM postmortem interval was 2.7 ± 0.2 hours for AD subjects and 3.0 ± 0.3 hours for control subjects. Neither age nor postmortem interval was significantly different (P<.05) between AD and control subjects. All AD subjects demonstrated progressive intellectual decline and met the criteria of the National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer′s Disease and Related Disorders Associations Work Group for the clinical diagnosis of probable AD.14 Histopathologic diagnosis was based on the analysis of multiple sections of neocortex, hippocampus, amygdala, entorhinal cortex, basal ganglia, brainstem, and cerebellum stained with hematoxylin-eosin, modified Bielschowsky silver stain, 10D-5 (for β-amyloid, Athena Neurosciences, South San Francisco, Calif),15 ubiquitin, and α-synuclein immunochemistry. Braak staging, an index of neuropathologic severity of AD,16,17 was carried out using the Gallyas stain. All AD subjects met accepted criteria for histopathologic diagnosis of AD.18,19 Control subjects were individuals without a history of dementia or other neurological disorders who underwent annual mental status testing as a part of our normal volunteer control group study. All control subjects had test scores within the normal range. Neuropathologic evaluation of control brains revealed only age-associated gross and histopathologic alterations. Subjects were excluded from this study who had been on a respirator or had prolonged terminal hypoxia, recent or old infarcts, intracranial hemorrhages, drug intoxication, alcoholism, or central nervous system neoplasms. Hypoxic changes were not found in any brain region on microscopic examination in any of the subjects used in this study.
Free 8-OHdG was isolated from ventricular CSF samples as previously described,13 using a modification of the procedure of Shigenaga et al.20 Briefly, the CSF samples (8-33.4 mL) were thawed at room temperature and thoroughly vortexed for 2 minutes, then 20 µL aliquots were removed for protein content determinations using the Pierce bicinchoninic acid method. To allow normalization of levels of free 8-OHG to protein content, accurately measured volumes (8-33.4 mL) of CSF were passed through C18 solid-phase extraction columns preconditioned with 10 mL of high-pressure liquid chromatography–grade methanol, 10 mL of 18-Ω distilled and deionized water, and 10 mL of 50-mmol/L (KH2PO4) monobasic potassium phosphate (KH2PO4)(pH 7.4) under slight vacuum. Stable-labeled 8-OHdG (8C13; 7.9N15; 41.2 nmol) was added to the columns at the same time. The eluent CSF was collected in fresh side-arm test tubes to prevent cross-contamination of the samples and reserved for isolation of intact DNA. The columns were then washed with 4 mL of 50-mM KH2PO4 followed by two 2-mL washes with 5% methanol-KH2PO4. The 8-OHdG and standard 8-OHG were eluted from the column with 3 mL of 15% methanol-KH2PO4.The eluent was added to new C18 columns preconditioned as described above. The columns were washed with 1 mL of distilled and deionized water and dried for 15 minutes under slight vacuum. Purified 8-OHdG and standard 8-OHG were eluted with 2 mL of high-pressure liquid chromatography–grade methanol, placed in 5-mL conical glass tubes, and lyophilized. Using standard solutions of DNA and nonlabeled 8OHdG, the purification allowed passage of approximately 97% of DNA while retaining nearly all 8-OHdG as determined by UV-visible absorption spectrometry.
Intact DNA was isolated from ventricular CSF using a modified procedure of Mecocci et al.7 The CSF was mixed with a 1:10 volume of 1mol/L disodium EDTA, 5% sodium dodecyl sulfate, and 50-mmol/L Tris-hydrochloride (pH 8.0) along with a 1:25 volume of 10-mg/mL proteinase K. The samples were digested for 2 hours at 37°, and 1:10 volume of 5-mmol/L sodium chloride was added. The solution was then extracted 3 times with buffer-saturated phenol containing 5.5-mmol/L 5-hydroxyquinoline to prevent artifactual oxidation of DNA. The samples were extracted 3 times with 24:1 chloroform-isoamyl alcohol, and an additional 1:10 volume of 5-mol/L sodium chloride was added. To isolate DNA, the solution was centrifuged through a 5000 molecular weight cutoff filter at a speed of 2000g for 14 hours at 4°C. The resulting DNA was resuspended in 1 mL of distilled and deionized water and the concentration was determined at 260 nm using a Genesys 5 UV–visible absorbtion spectrophotometer. Ratios of absorbance at 260 and 280 nm showed a mean ± SEM 260-280 ratio of 1.45 ± 0.05, indicating slight protein contamination. Although the isolation procedure should precipitate all DNA, the predominant band demonstrated by polyacrylamide gel electrophoresis was approximately 400 base pairs.
Stable-labeled 8-OHG (41.2 nmol) was added and the samples lyophilized. After lyophilization, the DNA and free 8-OHdG samples were subjected to formic acid hydrolysis, which converted 8-OHdG to the free base, 8-OHG, and bis (trimethylsilyl)trifluoroacetamide derivatization as previously described.13 After derivatization, the samples were lyophilized and suspended in 20 µL bis(trimethylsilyl)trifluoroacetamide immediately before analysis.
Analysis of samples was carried out on a model HP6890 gas chromatograph (Hewlett-Packard; Palo Alto, Calif) equipped with a mass spectrometer operated in selective ion-monitoring mode as previously described.13 Retention times for 8-OHG were essentially constant. Shot-to-shot variation was less than 2% for standard 8-OHG samples. Duplicate analysis of 8-OHG from randomly selected free and intact samples demonstrated variabilities comparable to those observed for standards. Levels of 8-OHG were quantified using stable-labeled 8-OHdG as an internal standard as described by Dizdaroglu.21 For quantification, peaks of interest were integrated and the area of the 8-OHG peak was compared with the internal standard peak, which has a known concentration. For 8-OHG in intact DNA, 8-OHG levels (nanomoles) were normalized to DNA content and converted to nanomoles per milliliter of CSF by multiplying the DNA content by the average DNA yield per milliliter of CSF. Free 8-OHG levels (nanomoles) were normalized to protein content (nanomoles per milligram of protein) and converted to nanomoles per milliliter of CSF by multiplying by the average protein concentration (milligrams of protein per milliliter of CSF). Conversion of levels of 8-OHG to terms of volume of CSF allows calculation of a unitless ratio. Because the gas chromatography/mass spectrometry procedure uses mass spectrometry as the detector, the signals from the labeled and unlabeled compounds may be separated. The advantage of using stable-labeled 8-OHG as an internal standard is that it responds identically to the compound of interest during hydrolysis and derivatization.
The ratio of 8-OHdG in intact DNA to free 8-OHG was calculated, and statistical analyses were performed using a 2-tailed Student t test and the commercially available ABSTAT software (AndersonBell; Arvada, Colo). Correlation analyses were carried out using ABSTAT software.
Statistical comparison of mean age, postmortem interval, DNA content (micrograms per milliliter) and protein content (milligrams per milliliter) indicated no statistical differences between AD and control subjects (Table 1). There was a statistically significant decrease in brain weight in subjects with AD (mean ± SEM, 1080 ± 30 g) compared with age-matched controls (1310 ± 50 g) (P<.001). The mean ± SEM ratio of 8-OHG in intact DNA to free 8-OHdG was significantly elevated (108-fold) in ventricular 8-OHG CSF samples from 18 AD subjects (6.46 ± 1.39) compared with samples from 7 control subjects (0.06 ± 0.02) (P = .01). Levels of 8-OHG in intact DNA (Table 2) were significantly (P = .03) increased (18-fold) in AD subjects' ventricular CSF (912.3 ± 232.7 pmol/mL CSF) compared with control subjects (50.1 ± 22.3 pmol/mL CSF). Levels of free 8-OHG were significantly lower in AD subjects (195.9 ± 40.3 pmol/mL CSF) compared with control subjects (683.7 ± 191.1 pmol/mL CSF) (P = .001) (Table 2). There were no significant correlations between the ratio of 8-OHG in intact DNA to free 8-OHG and age, postmortem interval, protein content, DNA content, or brain weight. There was a positive statistically significant (r = 0.89, P = .05) correlation between intact-to-free ratio and Braak stage.
The distribution of individual data points of the ratio of 8-OHG in intact DNA to free 8-OHG plotted on a log scale ranged from 0.01 to 0.12 for control subjects and from 0.44 to 23.60 for AD subjects (Figure 1). There is no overlap of the data, with the lowest AD value (0.44) 3.5 times the highest control value (0.12). Two of the 18 AD subjects had values close to those observed for control subjects. The other 16 AD subjects had ratios at least 10 times those of the highest control values. Figure 1 shows the data plotted on a log scale to allow comparisons of the distributions.
It is possible that oxidative damage to DNA may contribute to the pathophysiologic alterations found in the brain in AD. The determination of OHdG in urine is thought to be a good index of in vivo oxidation in DNA damage.22 Because ventricular CSF filters and disposes degraded cellular material from the brain, it should more accurately reflect levels of brain DNA oxidation than blood or urine. This study demonstrates a significant, 108-fold increase in the ratio of 8-OHG in intact DNA to free 8-OHG in ventricular CSF from AD patients compared with age-matched control subjects, all with short postmortem intervals. Distribution of the individual data points demonstrates that the lowest AD value is 3.5 times higher than the highest control value. Inspection of values of free 8-OHG and 8-OHG in intact DNA for each subject suggests that for AD subjects, the relatively low level of repair product contributes to the higher ratios observed. There is a statistically significant positive correlation between the ratio of intact 8-OHG in DNA to free 8-OHG and Braak stage (P = .05). Thus, this marker of DNA oxidation seems to mirror brain degeneration and has the potential to be used as an index of disease progression.
This study extends our prior study of levels of 8-OHG in intact DNA and free 8-OHG isolated from ventricular CSF.13 We used stable-labeled 8-OHG and gas chromatography/mass spectrometry with selective ion monitoring with selective ion monitoring to unequivocally identify peaks of interest based on chromatographic retention times and mass spectra. The results of this study for individual measures of free and intact 8-OHG agree with those of our prior study and demonstrate a statistically significant elevation of 8-OHG in intact DNA (P = .03) and a significant depletion of free 8-OHG (P = .001) in patients with AD compared with age-matched control subjects. The levels of 8-OHG measured in ventricular CSF are comparable to those observed in our brain nuclear DNA study, which demonstrated elevations of 8-OHdG in AD frontal, temporal, and parietal lobe structures.8 The studies of Lyras et al9 and Mecocci et al10 also showed increased DNA 8-OHG in the brain in AD. Our finding of an elevation of a marker of DNA oxidation concomitant with a decrease in levels of free repair product correlates well with our previous studies of CSF oxidation, which showed increased levels of 4-hydroxynonenal, a neurotoxic marker of lipid peroxidation, in AD ventricular CSF.2
Free 8-OHG measured in this study results from the excision of 8-OHG through the action of a base-specific glycosylase,23,24 which functions to cleave the altered base. After removal from the brain, cleaved bases are subsequently transported via CSF and blood and eventually excreted in urine,20,25 and they may serve as an efficient marker of oxidative DNA damage in vivo.20 A study of free 8-OHG in urine indicated that excretion decreases with age,26 whereas Mecocci et al10 found that brain 8-OHG in intact DNA increases with age, suggesting a decline in the repair mechanisms responsible for the excision of 8-OHG. These observations are consistent with our study showing decreased activity of the base excision repair enzyme, 8-oxyguanine glycosylase, responsible for excision of 8-OHG in vulnerable regions of the AD brain.13
Although several markers of oxidative stress have been measured in AD ventricular CSF, most demonstrate an overlap between AD and control subjects. Indeed, the distribution of individual measurements of free or 8-OHG in intact DNA demonstrates considerable overlap. However, when we compared the ratio of intact to free 8-OHG, there was no overlap between AD and control populations, with a 3.5-fold difference between the lowest AD and highest control values. Other studies of CSF samples have failed to demonstrate significant differences between AD and control subjects for levels of glutathione27 or selenium.28 Levels of CSF F2-isoprostanes and F4-neuroprostanes show overlaps between AD and control subjects,2 as do levels of CSF tau alone 29,30 or in combination with Aβ1-40 and Aβ1-42.31- 33 It has been suggested that the combination of CSF tau and Aβ may serve as a diagnostic marker for AD; however, these studies used living patients in whom the diagnosis was not absolutely certain. Although the present study used a method requiring larger volumes of CSF than are practical in living subjects, we are pursuing an immunoblot method requiring only a small amount of CSF. Because the study measured markers of oxidative DNA damage, it is unlikely that the results observed are specific to AD, but they may be present in other neurodegenerative diseases in which oxidative stress is involved in the pathogenic mechanism. Although our results represent the analysis of a relatively small number of subjects, they clearly separate patients with late-stage AD from nondemented age-matched controls. The potential of the ratio of 8-OHG in intact DNA to free 8-OHG as a diagnostic marker in AD deserves further study in living subjects at various stages of AD and in other neurodegenerative disorders. It also may be helpful in determining oxidative status and the response of AD patients to therapeutic antioxidant intervention.
Accepted for publication August 17, 2000.
This study was supported by the National Institutes of Health, Bethesda, Md, grants 5-P01-AG0 5119, and 5-P50-AG0 5144 and by a grant from the Abercrombie Foundation, Versailles, Ky.
The authors thank Drs Daron Davis and David Wekstein for CSF procurement, Jane Meara and Paula Thomason for technical and editorial assistance, and Cecil Runyons for demographic data.
Corresponding author: Mark A. Lovell, PhD, 101 Sanders-Brown Bldg, 800 S Limestone St, University of Kentucky, Lexington, KY 40536-0230 (e-mail: firstname.lastname@example.org).