Epigenetic Differences in Cortical Neurons from a Pair of Monozygotic Twins Discordant for Alzheimer's Disease

DNA methylation [1], [2] is capable of modulating coordinate expression of large numbers of genes across many different pathways, and may therefore warrant investigation for their potential role between genes and disease phenotype. In a rare set of monozygotic twins discordant for Alzheimer's disease (AD), significantly reduced levels of DNA methylation were observed in temporal neocortex neuronal nuclei of the AD twin. These findings are consistent with the hypothesis that epigenetic mechanisms may mediate at the molecular level the effects of life events on AD risk, and provide, for the first time, a potential explanation for AD discordance despite genetic similarities.


Introduction
Wide varieties of studies have examined candidate genes for associations with Alzheimer's disease (AD). Although such associations have been found, they are probabilistic rather than inevitable, with the exception of familial AD. A potential explanation of the probabilistic nature of the association between specific genes and disease may be the apparent genetic complexity of AD. However, the existence of rare monozygotic twins discordant for AD offers the opportunity to examine other factors that may be contributing to the probabilistic nature of the association between genes and AD, specifically epigenetic mechanisms. Since epigenetic modifications contribute to the phenotypic differences that emerge in monozygotic twins, including discordant disease states [3], we examined DNA methylation, an important epigenetic mechanism, in monozygotic twins discordant for Alzheimer's disease (AD). As has been described in sporadic AD cases [4], neurons of the AD twin exhibited dramatic decrements across multiple DNA methylation markers compared to the non-AD twin. Epigenetics may therefore, constitute a basic molecular genetic mechanism in the pathophysiology of AD.

Ethics Statement*
Written informed consent for autopsy was obtained for both cases in compliance with institutional guidelines of Boston University. The Boston University Institutional Review Board approved this study including recruitment, enrollment, and autopsy procedures. Both twins and their respective next-of-kin consented to brain autopsy for the purpose of research analysis as participants in the Boston University Alzheimer's Disease Center. The human brain tissue used in this manuscript was from routine existing autopsies, which fully qualifies for 4C exemption by NIH guidelines. In addition samples were analyzed anonymously (e.g. twin 1 and twin 2) throughout the experimental process.

Subjects and brain samples
Using standard protocols of NIH AD Centers, both twins were thoroughly evaluated antemortem and postmortem by boardcertified neurologists and a neuropathologist who determined their respective diagnoses as AD and neurologically normal, nondemented (ND). The AD twin was a white male chemical engineer who had extensive pesticide contact in his work. He developed AD symptoms at age 60, first manifest as the inability to read maps, followed by progressive loss of memory and intellect over 16 years until his death at age 76. His identical twin, also a chemical engineer with an identical education but different work environment, died at 79 years from prostate cancer. At the time of his death, he was cognitively intact. The twins were autopsied at the same facility using the same tissue processing protocols. Post mortem delay for the control twin was 3 h 10 minutes and 7 h 20 minutes for the AD twin. Both subjects were immediately snap frozen on aluminum plates cooled to 280uC on dry ice and immediately transferred to 280uC freezer for long term storage. In the AD twin post-mortem examination confirmed severe AD (NIA-Reagan: high, CERAD plaque: frequent, Braak: VI). In the non-demented twin post-mortem examination revealed sparse neuritic plaques and entorhinal and transentorhinal NFTs (NIA-Reagan: low, CERAD: sparse, Braak: II). The AD twin showed dense plaques and neurofibrillary tangles (NFT) in the anterior temporal cortex, while these stigmata were remarkably rare in the non-demented twin ( Figure 1).

Immunohistochemistry
Temporal neo-cortex was sliced axially into 1-cm thick slabs, immersion fixed for 48 h in buffered 4% paraformaldehyde at 4uC, washed extensively in phosphate buffer (PB), and cryoprotected in ethylene glycol and glycerol. The slabs were then sectioned at 40 mm on a freezing cryostat. Free-floating sections were stored in freezing solution (glycol/glycerol/PB) at 220uC until required for experiments. Tissue sections used for bright field microscopy were immunoreacted using the avidin-biotin complex/diaminobenzidine (DAB) method. Briefly, tissues were washed 26, blocked in 1% hydrogen peroxide for 45 min, washed 36, blocked in 3% bovine serum albumin (BSA) for 1 h, washed 26, and incubated at 4uC overnight in primary antibody solutions containing 0.25% BSA. Unless otherwise stated, all washes were with 16 PBS Triton (PBST). Available information about the antibodies is given in (Table 1). After incubation with primary antibody, sections were washed 36, incubated in biotinylated, species-specific secondary antibodies (Vector) for 2 h, washed 36, and incubated in avidin-biotin complex (Pierce) for 1 h. Following incubation with secondary antibody, the sections were washed 36, once in PBST and twice in 0.05 M Tris buffer, then exposed to DAB solution containing 125 ml of 5 mg/ml DAB (Sigma), 11.125 ml 50 mM Tris buffer pH 7.6, and 500 ml saturated nickel ammonium sulfate. Incubations during chromagen development were no longer than 10 min, and were followed by two quick rinses in 50 mM Tris to stop the reaction. Finally, the sections were dried, taken through graded alcohols, de-fatted in Neoclear (EMD), and mounted with Permount (Pierce). AD and ND sections were immunoreacted simultaneously using netwells in well-less plates. For fluorescence microscopy, the sections were washed 36, blocked with either 3% normal goat serum or 3% BSA, and incubated for 2 h. The sections were then washed 26, incubated in primary antibody in 0.25% BSA at 4uC overnight, washed again, and incubated in species-specific, fluorophoreconjugated secondary antibodies (Molecular Probes) at room temperature for 2 h. After a final wash, the sections were mounted, taken through Sudan Black to reduce autofluorescence, and coverslipped with Vectashield mounting media (Vector). Deletion of primary antibody or incubation with pre-immune serum resulted in abolition of specific immunoreactivity in all cases (data not shown). Adjacent serial sections were stained with cresyl violet for cell layer identification and verification that the island neurons of layer II were intact. For some sections, nuclei were counterstained with 49,69-diamidino-2-phenylindole (DAPI) (Invitrogen) before mounting.
Double-label immunohistochemistry was also employed to evaluate the associations of epigenetic factors with specific neuronal, and glial cell types. Briefly, sections were washed, blocked with either 3% normal goat serum or 3% BSA, and incubated for 1-2 h. Sections were then washed 26, incubated in primary antibodies raised in different species in 0.25% BSA/PBST at 4uC overnight. After primary incubation sections were washed 36 in PBST, and incubated in species-specific, fluorophoreconjugated secondary antibodies (Molecular Probes) at room temperature for 2 h. After a final wash, the sections were mounted, taken through Sudan Black to reduce autofluorescence, and coverslipped with Vectashield mounting media (Vector). Immunostained tissue sections were examined on Olympus IX51 and Olympus IX70 microscopes equipped with epifluorescence illumination or with confocal laser scanning using argon and krypton lasers (Olympus IX70). The findings were documented photographically with an Olympus DP-71 color digital camera or, for confocal microscopy, by Fluoview software (Olympus).

Cell Quantification and statistical analysis
Using the Fluoview software, integrated fluorescence intensities for 5-methylcytosine of layer II and layer III of the anterior  Figure 2a shows representative immunohistochemistry of anterior temporal neocortex for 5-methylcytosine, a marker of methylated CpG sites on DNA. In the AD twin, decreased immunoreactivity relative to the ND twin was readily apparent in the anterior temporal neocortex, a region severely affected in AD. Similar results were obtained in the pathologically-vulnerable superior frontal gyrus ( Figure S1). To develop quantitative data, integrated fluorescence intensities of immunoreactivity for 5-methylcytosine were recorded for layer II and layer III nuclei of DAPI-and neuron specific enolase positive cells from coded slides. By two-tailed t-test, all markers exhibited highly significant (P,0.0001) decrements in immunoreactivity in the AD twin compared to the ND twin ( Figure 2b). These findings are consistent with our previous results for non-twin AD and ND subjects [4].

DNA methylation Immunoreactivity
To show that the results obtained in neocortex were not due to differences in tissue handling, storage time, or quality, DNA methylation markers were also evaluated in cerebellum, a brain region that is largely spared from AD pathology. As in our previous assessment of AD and control cases [4], the cerebella of both twins exhibited virtually identical staining patterns and intensity for 5-methylcytosine (Figure 2a), as well as for another methylation marker, 5-methylcytidine ( Figure S1) and several methylation stabilizing factors (MBD2/3 and HDAC2; Figure S2)

Relationship of DNA methylation marker 5-metylcytosine to neurons and Glial cells
In Figure 3 we examine co-localization of the DNA methylation marker, 5-methylcytosine, with markers for neurons (neuron specific enolase), for reactive astrocytes (GFAP) and for microglia (RCA1). These figures show that DNA methylation is present in all three cell types in non-demented brain and that the decrement in DNA methylation seen in AD extends to all three cell types.

Discussion
Large-scale expression array studies have reported significant up-or downregulation of thousands of genes in AD [5,6]. These alterations in gene expression span multiple pathogenic pathways, including amyloid b peptide (Ab) processing, tau hyperphosphorylation, and inflammation, among others. Because many genes across the genome have methylation sites in their promoters [2], extensive hypomethylation in AD may provide an overarching principle that accounts for significant aspects of the molecular and pathogenic complexity of the disorder. For example, amyloid precursor protein (APP), the Ab precursor, has been shown to be normally methylated, and hypomethylated with age [7] which apparently enhances Ab production [8]. However, more recent data indicates no difference in methylation of the APP gene in AD [9]. Furthermore, inducible nitric oxide synthase, interleukin 1, and tumor necrosis factor-a are all increased as part of the inflammatory response in AD cortex [10]; all their respective genes are methylated; and all show enhanced secretion with hypomethylation [11,12,13]. At the protein level, protein phosphatase 2A is methylated (probably by peptidylarginine methyltransferases or lysine methyltransferases), and its hypomethylation results in tau hyperphosphorylation [14].
The present findings indicate that epigenetic mechanisms may provide a molecular basis for the effect of life events, including exposure to hazardous substances, on AD risk. More specifically, they may provide a rationale for the consistent epidemiologic and neuropathologic association of AD with homocysteine elevation and folate deficiencies [15,16], since folate ultimately provides the methyl group for DNA methylation. Maintaining adequate dietary folate (and B 12 ) or increasing S-adenosylmethionine levels might therefore be useful, inexpensive strategies to decrease risk for AD. Since epigenetic patterns can be passed on to subsequent generations, epigenetics may also constitute a mechanism by which AD in a first degree relative confers increased risk of ''sporadic'' disease. Figure S1 Top four panels: Immunoreactivity for 5-methylcytodine (5MeCd), another methylation marker, in medial temporal gyrus and cerebellum of non-demented and Alzheimer's disease twin. Note similarity of immunoreactivity in granule cell layer of both twins in cerebellum, which is relatively unaffected in Alzheimer's disease. Bottom two panels: 5-methylcytosine (5Me-C)immunoreactivity in the Superior frontal gyrus (SFG)in AD and non-demented twin.