Research
Original Investigation
Association of Brain DNA Methylation in SORL1, ABCA7,
HLA-DRB5, SLC24A4, and BIN1 With Pathological Diagnosis
of Alzheimer Disease
Lei Yu, PhD; Lori B. Chibnik, PhD; Gyan P. Srivastava, PhD; Nathalie Pochet, PhD; Jingyun Yang, PhD; Jishu Xu, MS;
James Kozubek, MS; Nikolaus Obholzer, PhD; Sue E. Leurgans, PhD; Julie A. Schneider, MD;
Alexander Meissner, PhD; Philip L. De Jager, MD, PhD; David A. Bennett, MD
Editorial page 8
IMPORTANCE Recent large-scale genome-wide association studies have discovered several
genetic variants associated with Alzheimer disease (AD); however, the extent to which DNA
methylation in these AD loci contributes to the disease susceptibility remains unknown.
Supplemental content at
jamaneurology.com
OBJECTIVE To examine the association of brain DNA methylation in 28 reported AD loci with
AD pathologies.
DESIGN, SETTING, AND PARTICIPANTS Ongoing community-based clinical pathological cohort
studies of aging and dementia (the Religious Orders Study and the Rush Memory and Aging
Project) among 740 autopsied participants 66.0 to 108.3 years old.
EXPOSURES DNA methylation levels at individual CpG sites generated from dorsolateral
prefrontal cortex tissue using a bead assay.
MAIN OUTCOMES AND MEASURES Pathological diagnosis of AD by National Institute on
Aging–Reagan criteria following a standard postmortem examination.
RESULTS Overall, 447 participants (60.4%) met the criteria for pathological diagnosis of AD.
Brain DNA methylation in SORL1, ABCA7, HLA-DRB5, SLC24A4, and BIN1 was associated with
pathological AD. The association was robustly retained after replacing the binary trait of
pathological AD with 2 quantitative and molecular specific hallmarks of AD, namely, Aβ load
and paired helical filament tau tangle density. Furthermore, RNA expression of transcripts of
SORL1 and ABCA7 was associated with paired helical filament tau tangle density, and the
expression of BIN1 was associated with Aβ load.
CONCLUSIONS AND RELEVANCE Brain DNA methylation in multiple AD loci is associated with
AD pathologies. The results provide further evidence that disruption of DNA methylation is
involved in the pathological process of AD.
Author Affiliations: Author
affiliations are listed at the end of this
article.
JAMA Neurol. 2015;72(1):15-24. doi:10.1001/jamaneurol.2014.3049
Published online November 3, 2014.
Corresponding Author: David A.
Bennett, MD, Rush Alzheimer’s
Disease Center, Rush University
Medical Center, 600 S Paulina St,
Chicago, IL 60612 (david_a_bennett
@rush.edu).
15
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
Research Original Investigation
Brain DNA Methylation and Pathological AD Diagnosis
A
lzheimer disease (AD) is a complex age-related neurodegenerative illness and is considered highly
heritable.1,2 Besides the 3 causal genes of familial AD
(ie, APP, PSEN1, and PSEN2) and the well-known apolipoprotein E (APOE) gene, large-scale genome-wide association
studies3-7 have identified and replicated additional common
variants that are associated with susceptibility of AD, including CR1, BIN1, CD33, CLU, ABCA7, CD2AP, PICALM, EPHA1,
MS4A6A, and MS4A4A. The latest meta-analysis8 to date (with
almost 75 000 individuals) further expands the list to include
HLA-DRB5, PTK2B, SORL1, SLC24A4, DSG2, INPP5D, MEF2C,
NME8, ZCWPW1, CELF1, FERMT2, and CASS4. Studies9,10 report the association of a rare variant in TREM2 with AD risk.
In addition, a large hexanucleotide repeat expansion in
C9oborf72 is implicated in AD.11,12
Genetic marks are tagged by epigenetic information,
such as DNA methylation and histone modification, and
these epigenetic mechanisms are essential in regulating
gene expression and maintaining genomic homeostasis.
Aberrant epigenetic alterations are associated with various
complex human diseases, including AD.13-16 The existence
of cis methylation quantitative trait loci raises the possibility that alteration of DNA methylation in AD loci likely has
an important role in affecting the disease susceptibility.17
We are unaware of any other studies that have systematically assessed the association of brain DNA methylation in
these loci with AD pathologies.
In this study, we target brain DNA methylation in 28 known
susceptibility loci for AD and examine the global association
of methylation in each locus. Our study uses a unique collection of data that integrates pathological measures with brain
DNA methylation from 740 autopsied participants in 2 ongoing clinical pathological studies of aging and dementia.
Methods
Study Participants
Participants were from the Religious Orders Study18 and the
Rush Memory and Aging Project.19 Both studies were approved by the Institutional Review Board of Rush University
Medical Center, and each participant signed a consent form and
an Anatomical Gift Act. Participants enroll without known dementia and agree to annual clinical evaluations and organ donation at the time of death. By January 2010, when the methylation data were generated, more than 2400 participants had
been enrolled, with the follow-up rate among survivors exceeding 90%. More than 800 autopsies had been performed,
and the autopsy rate exceeds 90%.
Neuropathological Assessment of AD
Postmortem brains were processed and examined following a
standard procedure.20 Multiple AD pathology measures were
examined, including neuropathological AD diagnosis, Aβ
load, and paired helical filament tau immunoreactive neurofibrillary tangle (tau tangle) density. Neuropathological AD
diagnosis follows the National Institute on Aging–Reagan criteria, which require an intermediate likelihood of AD (ie, at
16
least Braak stage 3 or 4 and Consortium to Establish a Registry for Alzheimer’s Disease [CERAD] moderate plaques) or a
high likelihood of AD (ie, at least Braak stage 5 or 6 and
CERAD frequent plaques).21-23 Further details of the brain
autopsy procedures and quantification of these pathology
measures are provided in the eMethods in the Supplement.
Brain DNA Methylation and RNA Expression
Frozen tissues of the dorsolateral prefrontal cortex of deceased participants were thawed on ice, and DNA was extracted from gray matter following a mini-protocol (QIAamp
DNA 51306; Qiagen). DNA methylation was interrogated using
a bead assay (Infinium HumanMethylation450; Illumina). Raw
data were processed following a rigorous pipeline for data quality control, as detailed in the eMethods in the Supplement. At
the end of the quality-control pipeline, we obtained distinct
DNA methylation values for 420 132 autosomal CpG sites for
740 samples. In this study, we restricted our analysis to the sites
that cover both genic and 100-kilobase flanking areas around
each of 28 reported AD loci.
RNA-Seq expression data were generated from frozen
dorsolateral prefrontal cortex tissues following the construction of complementary DNA libraries. The paired-end
reads were mapped using the Ensemble human genome
transcriptomic database (http://www.ensembl.org). RNA
expression of the associated AD genes was queried and
examined for an association with AD pathologies. Details on
the RNA-Seq expression profiling are provided in the
eMethods in the Supplement.
Statistical Analysis
The primary analysis examined the association of DNA methylation with pathological AD diagnosis, which was done in 2
steps. First, parallel logistic regression models were conducted with AD diagnosis as the binary outcome and individual CpG site as the predictor. Because age has a large effect on AD pathology and may affect DNA methylation, we
controlled for age at death in all models. Additional covariates included sex, batch, and bisulfite conversion efficiency.
Additional analyses controlled for potential confounding due
to the presence of macroscopic and microscopic infarcts and
cortical Lewy bodies. Second, significance values (P values)
of individual CpGs in a specific AD locus were combined following the Fisher product method,24 denoted by ψ = −2兺 log
pi. Here, pi refers to the significance value of ith CpG. This omnibus statistic ψ assesses a hypothesis that none of the CpGs
in the interrogated locus were associated with the outcome.
Statistical significance of ψ was tested using random permutations (eMethods in the Supplement).
Associated loci identified in the primary analysis were subjected to further interrogation in which we repeated the analyses by replacing the binary outcome of AD diagnosis with continuous measures of Aβ load and separately tau tangle density
in linear regression models. Because both pathology measures are right skewed, we applied square root transformation before the analyses. Similar regression analyses were performed to explore the association of RNA transcript expression
in the associated loci with AD pathologies. Postmortem inter-
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
jamaneurology.com
Brain DNA Methylation and Pathological AD Diagnosis
Original Investigation Research
Table 1. Omnibus Test for DNA Methylation in Associated Alzheimer Disease Loci With Pathological Alzheimer Disease
P Value
Permuted
Adjusteda
289.6763
<1.0 × 10−6
<.0000280
255
666.5471
6.0 × 10−6
.0000840
32385153:32598006
48
157.3848
5.0 × 10−5
.0004667
14
92688924:93067825
62
188.9191
.00013
.0009100
2
127705598:127964903
95
247.3020
.0032
.0179200
INPP5D
2
233825035:234216549
87
214.4398
.0188
.0877333
FERMT2
14
53223988:53517815
54
138.7177
.0239
.0917000
TREM2
6
41026245:41230922
74
182.7488
.0262
.0917000
CLU
8
27354433:27572328
70
173.0821
.0316
.0983111
PTK2B
8
27068998:27416908
101
235.4268
.0517
.1476000
MS4A6A
11
59839079:60050674
23
61.24
.0651
.1657091
CELF1
11
47387488:47674792
11
357.3677
.0847
.1976333
APP
21
27152860:27643446
21
PSEN1
14
73503142:73790399
56
MS4A4A
Locus
Chromosome
SORL1
11
121222911:121604471
69
ABCA7
19
940101:1165570
6
HLA-DRB5
SLC24A4
BIN1
a
Covered Region
No. of CpGs
Observed Test Statistic
53.82021
129.3102
74.74408
.1005
.2164615
.1241
.2415467
11
59948013:60176445
31
.1294
.2415467
CR1
1
207569472:207915110
49
109.341
.1991
.3484250
PSEN2
1
226958272:227183804
68
143.9049
.3032
.4383400
EPHA1
7
142988204:143205985
89
186.9672
.3033
.4383400
CASS4
20
54887167:55134396
84
176.5964
.3070
.4383400
CD2AP
6
47345524:47694996
35
75.1397
.3131
.4383400
APOE
19
45309038:45512650
104
214.1933
.3683
.4910667
PICALM
11
85568213:85880923
32
ZCWPW1
7
99898494:100126302
123
C9orf72
9
27446543:27673842
19
NME8
7
37788198:38040002
48
MEF2C
5
87914057:88299922
146
CD33
19
51628334:51843274
56
DSG2
18
28978026:29228814
21
66.19298
.4035
.5135455
.6369
.7753565
32.78186
.7097
.8279833
84.99657
.7830
.8740308
.8116
.8740308
94.70939
.8789
.8963000
31.04398
.8963
.8963000
237.6435
270.7734
P values are adjusted using the Benjamini-Hochberg procedure.25
vals and RNA degradation could potentially have an effect on
RNA expression; therefore, we adjusted for postmortem intervals and RNA degradation scores in these analyses, in addition to age and sex.
We corrected for multiple testing using the procedure by
Benjamini and Hochberg.25 In identifying associated loci, correction was applied at the locus level. In assessing the top CpG
sites and RNA transcripts within each locus, experimentwise
correction was applied. Statistical analyses were performed
using software programs (SAS, version 9.3; SAS Institute and
R, version 3.0.1; http://www.r-project.org).
Results
Characteristics of the Study Participants
DNA methylation and neuropathology data were available
from 740 study participants. The mean (SD) age at death
was 88.0 (6.7) years (age range, 66.0-108.3 years); 471
(63.6%) were female, and the mean (SD) years of education
was 16.4 (3.6) years (range, 3-28 years). Neuropathological
evaluations showed that 471 (63.6%) had moderate or frejamaneurology.com
quent neuritic plaques as defined by CERAD and 369
(49.9%) had Braak stage 4 or higher. Overall, 447 study participants (60.4%) met the criteria for pathological diagnosis
of AD. The median Aβ load was 2.2% (interquartile range,
0.4%-5.5%), and the median tau tangle density was 3.7/mm2
(interquartile range, 1.2-8.2/mm2).
Association of Methylation in Target Loci
With Pathological AD
For each of 28 AD loci, parallel logistic regression analyses were
performed, and significance values of individual CpGs were
combined for assessment of the global association at the locus level. DNA methylation in SORL1, ABCA7, HLA-DRB5,
SLC24A4, and BIN1 was associated with pathological AD diagnosis (Table 1 and Figure 1). The results for SORL1, ABCA7,
HLA-DRB5, and SLC24A4 also survived more conservative Bonferroni correction. The results were unchanged after controlling for infarcts and Lewy bodies (Table 2). Subsequent analyses focused on these 5 associated loci. Their Online Mendelian
Inheritance in Man accession numbers are SORL1 (602005),
ABCA7 (605414), HLA-DRB5 (604776), SLC24A4 (609840), and
BIN1 (601248).
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
17
Research Original Investigation
Brain DNA Methylation and Pathological AD Diagnosis
Figure 1. DNA Methylation With Pathological Alzheimer Disease Diagnosis in Associated Loci
SORL1
SORL1
4
3
2
0.010
1
100 150 200 250 300
–10
0
10
20
400
Individual Estimate
HLA-DRB5
600
–20
2
Combined Test Statistic
–5
0
0.010
BIN1
20
4
3
2
1
0
0
60
Individual Estimate
10
SLC24A4
0.020
5
0
5
1
–10
–10
Individual Estimate
–Log10 P Value
3
160
1
700
4
0
120
500
0.030
Density
–Log10 P Value
0.010
2
SLC24A4
5
0.020
3
Combined Test Statistic
HLA-DRB5
0.030
4
0
0
–20
Combined Test Statistic
Density
0.020
0
0
40 60 80
5
–Log10 P Value
0.010
ABCA7
0.030
Density
0.020
0
100
140
180
–10 –5
Combined Test Statistic
0
5
10 15
Individual Estimate
BIN1
5
–Log10 P Value
0.030
Density
ABCA7
5
–Log10 P Value
Density
0.030
0.020
0.010
4
3
2
1
0
0
150
200
250
300
–15 –10 –5 0
Combined Test Statistic
5 10 15
Individual Estimate
For each locus, the left panel shows the histogram and density function of the
global test statistic generated from randomly permuted data, with a blue
dashed line superimposed to represent the value of the same statistic observed
from the actual data. The right panel is the volcano plot showing the
significance level vs the regression estimates for individual CpG sites.
Table 2. Omnibus Test for DNA Methylation in Associated Alzheimer Disease Loci With Pathological Alzheimer
Disease, Adjusted for Other Pathologiesa
Model A
Model B
Locus
Observed Test Statistic
Permuted P Value
Observed Test Statistic
Permuted P Value
SORL1
290.3979
<1.0 × 10−5
277.8277
<1.0 × 10−5
ABCA7
665.9884
−5
643.6079
−5
8.0 × 10
HLA-DRB5
159.832
5.0 × 10−5
156.2942
.00012
SLC24A4
190.1051
.0001
180.3139
.00068
BIN1
249.7376
.00248
251.4047
.00179
<1.0 × 10
SORL1
We interrogated 69 CpG sites in the SORL1 locus (sortilinrelated receptor, L [DLR class] A repeats containing), of which
8 showed associations with pathological AD diagnosis (eTable
1 in the Supplement). The top CpG (cg15241519) was observed
in a weakly transcribed region in the gene body, where methylation was associated with greater odds for pathological AD.
Methylation at 7 CpG sites in SORL1 was associated with
Aβ load and 7 CpG sites with tau tangle density (eTable 2 and
eTable 3 in the Supplement). Of these, 4 CpG sites (cg15241519,
cg08441314, cg11606444, and cg22136098) were associated
with both indexes. The regional association plot (Figure 2A)
highlights the location of top hits for the 3 pathological AD
traits.
18
Model A is adjusted for the presence
of macroscopic and microscopic
infarcts, in addition to age, sex,
batch, and bisulfite conversion
efficiency. Model B is further
adjusted for the presence of cortical
Lewy bodies.
We examined the expression of 13 SORL1 transcripts in relation to AD pathologies. We found little association with Aβ
load. By contrast, upregulated expression of the SORL1 transcript (ENST00000524873.1) was associated with higher tau
tangle density (mean [SE] β coefficient, 2.676 [0.687]; adjusted P = .0015). Methylation and expression in the locus were
only weakly correlated (r range, −0.15 to 0.15) (eFigure 1 in the
Supplement).
ABCA7
We analyzed 255 CpG sites in the ABCA7 locus (adenosine triphosphate [ATP]-binding cassette, sub-family A [ABC1], member 7). Nineteen sites showed nominal association with pathological AD diagnosis, and the result was enriched for positively
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
a
jamaneurology.com
Brain DNA Methylation and Pathological AD Diagnosis
Original Investigation Research
Figure 2. Association Plot for the SORL1 and ABCA7 Loci
A Association plot for the SORL1 locus
15
Observed, –Log10 P Value
12
PC_repressed
Strong_promoter
Weak_enhancer
Low_signal
Inactive/poised_promoter
Weak_transcibed/low_signal_proximal_to_active_regions
Heterochromatin
Active_TSS_flanking
Strong_transcription
CNV/rep
Active_enhancer
9
cg25943118
6
5×10–5
cg15241519
cg08441314
3
5×10–2
0
SORL1
121 300
121 400
121 500
Chromosome 11 Position (hg19), Kilobase
B
Association plot for the ABCA7 locus
15
Observed, –Log10 P Value
12
PC_repressed
Strong_promoter
Weak_enhancer
Low_signal
Inactive/poised_promoter
Weak_transcibed/low_signal_proximal_to_active_regions
Heterochromatin
Active_TSS_flanking
Strong_transcription
CNV/rep
Active_enhancer
cg04587220
9
cg04587220
6
5×10–5
cg02308560
3
5×10–2
0
ARID3A
WDR18
POLR2E
GPX4
GRIN3B
SBNO2
C19orf6
CNN2
ABCA7
HMHA1
1000
1100
Chromosome 19 Position (hg19), Kilobase
A, Association plot for the SORL1 locus. P values of individual CpG sites for each
of the 3 Alzheimer disease pathological indexes (red diamond indicates
pathological Alzheimer disease; black diamond, Aβ load; and red circle, tau
tangle density) were plotted against their genomic positions. The color band on
the x-axis represents the chromatic state of the region. The smallest P value for
each outcome is highlighted with larger symbols. B, Association plot for the
ABCA7 locus. CNV/rep indicates copy number variation/repetitive;
PC, Polycomb; and TSS, transcription start site.
associated CpGs (eTable 4 in the Supplement). Specifically, the
regression coefficients for 15 of these 19 CpGs (78.9%) had positive signs such that methylation was associated with greater
odds for AD diagnosis. Of these, 3 CpG sites remained significant after adjusting for multiple testing, among which
cg02308560 and cg04587220 (30 base pairs apart) were observed in a polycomb-repressed region in the HMHA1 gene
proximate to the 3′ untranslated region of ABCA7.
Methylation at 12 CpG sites in the ABCA7 locus was associated with Aβ load and 18 CpG sites with tau tangle density
jamaneurology.com
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
19
Research Original Investigation
Brain DNA Methylation and Pathological AD Diagnosis
(eTable 5 and eTable 6 in the Supplement). Figure 2B shows
that the methylation sites in the HMHA1 gene demonstrate the
strongest signals in relation to AD pathologies. Similar to pathological AD, the results were enriched for positive association
such that methylation at all but one of these CpG sites was associated with increased burden of AD pathologies.
We examined 17 transcripts of ABCA7 for the association
of RNA expression with AD pathologies. We observed that
ENST00000525073.2 expression was associated with higher tau
tangle density (mean [SE] β coefficient, 0.062 [0.018]; adjusted P = .0138). Correlations of DNA methylation at interrogated CpG sites with the expression of ABCA7 transcripts were
weak (r range, −0.2 to 0.2) (eFigure 2 in the Supplement).
HLA-DRB5
Eight of forty-eight CpG sites in the HLA-DRB5 locus (major histocompatibility complex, class II, DR beta 5) showed nominal
association with pathological AD diagnosis; however, none survived multiple testing (eTable 7 in the Supplement). Methylation in the locus was associated with Aβ load and with tau
tangle density. At the CpG level, 3 CpG sites were associated
with Aβ load and 9 were associated with tau tangles (eTable 8
and eTable 9 in the Supplement). An association plot shows
that methylation association with AD pathologies peaked in
the nearby HLA-DRA gene (Figure 3A). We found no association of HLA-DRB5 RNA expression with any of the 3 pathological AD traits.
SLC24A4
We interrogated 62 CpG sites in the SLC24A4 locus (solute carrier family 24, member 4) (Figure 3B). Three CpGs showed associations with pathological AD and 8 with Aβ load (eTable 10
and eTable 11 in the Supplement). We found weaker association with tau tangle pathology (eTable 12 in the Supplement).
We examined the expression of 9 transcripts of SLC24A4 and
found no association with any of the 3 pathological AD traits.
BIN1
In total, 95 CpG sites in the BIN1 locus (bridging integrator 1)
were interrogated. Two CpG sites showed associations with
pathological AD (eTable 13 in the Supplement). The cg22883290
site was located in a weakly transcribed region proximate to
the 3′ untranslated region of BIN1, and cg04019522 was in an
active enhancer region in the gene body. Methylation at both
sites was associated with greater odds for pathological AD
diagnosis.
Methylation at 3 CpG sites in BIN1 was associated with Aβ
load and 5 CpG sites with tau tangle density (eTable 14 and
eTable 15 in the Supplement). An association plot shows that
the strongest signals came from the sites proximate to the 3′
untranslated region of BIN1, as well as the downstream CpG
island harboring cg19153828 (Figure 4).
We examined the expression of 14 BIN1 transcripts. Higher
expression of 3 BIN1 transcripts (ENST00000393040.3,
ENST00000409400.1, and ENST00000462958.1) was associated
with greater Aβ load (eTable 16 in the Supplement). Surprisingly,
we observed that another transcript (ENST00000316724.5)
showed an association in the opposite direction such that higher
20
expression was associated with less amyloid. The pairwise correlations between methylation of individual CpG sites and BIN1
expression showed a weak correlation, with Pearson r between
−0.27 and 0.23 (eFigure 3 in the Supplement).
Mediation Analysis
In the final set of analyses, we explored potential mediation
of amyloid on the association of DNA methylation with tau
tangle pathology. To do so, we repeated the analysis of tau
tangle density by further controlling for Aβ load for the 4 loci
in which CpGs were strongly associated with both amyloid and
tangles. An attenuated association of DNA methylation would
be evidence of mediation.26 We found that DNA methylation
associations with tau tangle density were essentially retained for all the loci after controlling for Aβ load (Table 3).
These data suggest that methylation has an independent effect on these 2 molecular processes.
Discussion
In this study, we targeted brain DNA methylation in 28 previously reported AD loci. We assessed at the locus level the global
association of DNA methylation. We found that DNA methylation in 5 of 28 AD loci was associated with pathological AD
diagnosis, including SORL1, ABCA7, HLA-DRB5, SLC24A4, and
BIN1.
Increasing evidence exists for AD-related alterations in
DNA methylation. Evaluations of immunoreactivity of DNA
methylation markers suggest that levels of global DNA methylation are reduced in entorhinal cortex27 and hippocampus28
in AD. However, this result is inconclusive. Increased global
DNA methylation is observed in frontal cortex of AD brain,29
and findings from another genome-wide DNA methylation association study14 showed hypermethylated and hypomethylated sites in AD brain. In this study, not all interrogated loci
demonstrated evidence of methylation alterations with AD.
Most of the top CpG sites in ABCA7 and BIN1 showed increased burden of AD pathologies with more methylation.
These results suggest that the pathogenesis of AD affects the
brain’s epigenome in a strong but specific manner and hint at
a greater level of complexity in the role of the epigenome in
the disease.
Our findings provide new evidence that brain DNA methylation in AD loci might be involved in the pathological process of AD. Specifically, a recent genome-wide DNA methylation study30 reported 71 discrete CpG sites that are associated
with neuritic amyloid plaques, including cg02308560 in ABCA7
and cg22883290 in BIN1. Herein, we confirm the signals in
ABCA7 and BIN1 with Aβ load. Furthermore, we show that
methylation in these 2 loci is also related to tau tangle density. In addition, we observed associations in 3 additional AD
loci. Of these, the SORL1 locus shows the most significant association with AD. To our knowledge, only one prior study31
has examined brain DNA methylation in SORL1 in relation to
AD. The study targeted the gene’s promoter sequences using
much fewer brain samples, and it demonstrated no differences in methylation between AD and controls. Separately, we
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
jamaneurology.com
Brain DNA Methylation and Pathological AD Diagnosis
Original Investigation Research
Figure 3. Association Plot for the HLA-DRB5 and SLC24A4 Loci
A Association plot for the HLA-DRB5 locus
15
Observed, –Log10 P Value
12
PC_repressed
Strong_promoter
Weak_enhancer
Low_signal
Inactive/poised_promoter
Weak_transcibed/low_signal_proximal_to_active_regions
Heterochromatin
Active_TSS_flanking
Strong_transcription
CNV/rep
Active_enhancer
9
6
5×10–5
cg06060962
cg17606183
cg17606183
3
5×10–2
0
HLA-DRA
HLA-DRB5
32 400
HLA-DRB6
HLA-DRB1
32 500
Chromosome 6 Position (hg19), Kilobase
B
Association plot for the SLC24A4 locus
15
Observed, –Log10 P Value
12
PC_repressed
Strong_promoter
Weak_enhancer
Low_signal
Inactive/poised_promoter
Weak_transcibed/low_signal_proximal_to_active_regions
Heterochromatin
Active_TSS_flanking
Strong_transcription
CNV/rep
Active_enhancer
9
6
cg02902617
5×10–5
cg16757332
3
cg02902617
5×10–2
0
SLC24A4
92 800
92 900
RIN3
93 000
Chromosome 14 Position (hg19), Kilobase
A, Association plot for the HLA-DRB5 locus. P values of individual CpG sites for
each of the 3 Alzheimer disease pathological indexes (red diamond indicates
pathological Alzheimer disease; black diamond, Aβ load; and red circle, tau
tangle density) were plotted against their genomic positions. B, Association plot
for the SLC24A4 locus. CNV/rep indicates copy number variation/repetitive;
PC, Polycomb; and TSS, transcription start site.
observed associations of methylation in HLA-DRB5 and
SLC24A4, and we are unaware of any previous report on these
associations.
Our results suggest that altered methylation in these loci
might involve both Aβ and tau tangle pathologies. SORL1 and
ABCA7 have previously been implicated in the Aβ process.
SORL1 encodes sortilin-related receptor LR11 or SorLA, which
controls Aβ production such that reduced expression of SORL1
tends to increase Aβ production and hence promote AD.32
ABCA7 encodes an ATP–binding cassette transporter that regulates processing of amyloid precursor protein (APP) in vitro.
A 2008 study33 on cultured cells reported that ABCA7 inhibits
Aβ secretion. A more recent study34 on transgenic mice showed
that ABCA7 regulates Aβ homeostasis in the brain and deletion of ABCA7 doubles cerebral Aβ accumulation. Nonetheless, the relationship between DNA methylation and Aβ is complex. One study 35 showed that Aβ reduces global DNA
methylation but increases neprilysin DNA methylation, which
jamaneurology.com
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
21
Research Original Investigation
Brain DNA Methylation and Pathological AD Diagnosis
Figure 4. Association Plot for the BIN1 Locus
15
Observed, –Log10 P Value
12
PC_repressed
Strong_promoter
Weak_enhancer
Low_signal
Inactive/poised_promoter
Weak_transcibed/low_signal_proximal_to_active_regions
Heterochromatin
Active_TSS_flanking
Strong_transcription
CNV/rep
Active_enhancer
9
cg19153828
6
5×10–5
cg22883290
cg22883290
3
5×10–2
0
BIN1
127 700
CYP27C1
127 800
127 900
Chromosome 2 Position (hg19), Kilobase
P values of individual CpG sites for each of the 3 Alzheimer disease pathological
indexes (red diamond indicates pathological Alzheimer disease; black diamond,
Aβ load; and red circle, tau tangle density) were plotted against their genomic
positions. CNV/rep indicates copy number variation/repetitive; PC, Polycomb;
and TSS, transcription start site.
Table 3. Omnibus Test for DNA Methylation in Associated Alzheimer Disease Loci With Aβ Load and Tau Tangle Density
Aβ Loada
Tau Tangle Densitya
Locus
Observed Test
Statistic
Permuted P Value
Observed Test
Statistic
Permuted P Value
Observed Test
Statistic
SORL1
285.6039
<1.0 × 10−5
242.0672
<1.0 × 10−5
203.1709
.00039
<1.0 × 10−5
992.4105
<1.0 × 10−5
ABCA7
<1.0 × 10−5
1035.527
1017.743
Permuted P Value
HLA-DRB5
159.6211
5.0 × 10−5
221.1747
<1.0 × 10−5
176.9269
1.0 × 10−5
BIN1
291.9307
<1.0 × 10−5
368.237
<1.0 × 10−5
289.7298
1.0 × 10−5
a
The models are adjusted for age, sex, batch, and bisulfite conversion efficiency.
b
The model is further adjusted for Aβ load.
subsequently suppresses neprilysin expression, leading to further Aβ degradation. BIN1 is thought to be involved in endocytosis, which could serve as a pathway that leads to APP production and release.36,37 The gene was also shown to mediate
AD risk by modulating tau pathology.38 Little is known about
functional consequences of HLA-DRB5 and SLC24A4 in relation to AD. The HLA-DRB5 locus encodes a major histocompatibility complex class II protein involved in immune
responses, 8 and SLC24A4 may be involved in neural
development.39 In addition, the SLC24A4 gene is located next
to the RIN3 (Ras and Rab interactor 3) gene, which interacts
with BIN1 in the early endocytic pathway.40
Our RNA expression data also reveal some notable results. We find little evidence that SORL1 or ABCA7 expression
is associated with Aβ load. Instead, higher expression of the
SORL1 transcript, and similarly that of ABCA7, is associated
with more tangle pathology. One plausible explanation is that
these genes are overexpressed in response to the neuronal inflammation or degeneration that is induced by AD pathologies, such as tau pathology. We observe that higher expres22
Tau Tangle Densityb
sion of 3 BIN1 transcripts is associated with more Aβ, consistent
with the finding that BIN1 expression is elevated in AD brain.38
Our analysis also reveals another transcript in the BIN1 locus,
of which higher expression is associated with less amyloid. Recent evidence suggests that the BIN1 protein level is lower in
brains of sporadic AD cases.41
Separately, our data show that pairwise correlations between methylation of individual CpGs and expression of the
associated genes are weak. Large-scale data sets examining the
relationship between methylation status and gene expression levels in human brain are just beginning to emerge. Therefore, their relationship at this time remains poorly understood. It is likely that several factors contribute to the poor
associations. First, our methylation scan only captures a fraction of the CpGs across the genome. Second, DNA methylation is just one of many epigenetic changes that contribute to
the expression, and other factors, such as microRNA, might
be at play.42 Third, while there are sparse data from brain regarding these associations at the density measured in our study,
weak associations have been reported for other tissues.43
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
jamaneurology.com
Brain DNA Methylation and Pathological AD Diagnosis
Original Investigation Research
Fourth, we are specifically interrogating genes that harbor genetic variants associated with AD. Genomic variants are known
to affect methylation quantitative trait loci and expression
quantitative trait loci, both cis and trans, in human brain and
other tissues.43,44 Additional studies are warranted to elucidate the relationship between DNA methylation and RNA expression in these target genes.
Conclusions
In summary, investigating the association of DNA methylation in target loci with AD pathology is a first step to better un-
ARTICLE INFORMATION
Accepted for Publication: August 29, 2014.
Published Online: November 3, 2014.
doi:10.1001/jamaneurol.2014.3049.
Author Affiliations: Rush Alzheimer’s Disease
Center, Rush University Medical Center, Chicago,
Illinois (Yu, Yang, Leurgans, Schneider, Bennett);
Department of Neurological Sciences, Rush
University Medical Center, Chicago, Illinois (Yu,
Yang, Leurgans, Schneider, Bennett); Program in
Translational NeuroPsychiatric Genomics, Institute
for the Neurosciences, Department of Neurology,
Brigham and Women’s Hospital, Boston,
Massachusetts (Chibnik, Srivastava, Pochet, Xu,
Kozubek, Obholzer, De Jager); Program in
Translational NeuroPsychiatric Genomics, Institute
for the Neurosciences, Department of Psychiatry,
Brigham and Women’s Hospital, Boston,
Massachusetts (Chibnik, Srivastava, Pochet, Xu,
Kozubek, Obholzer, De Jager); Harvard Medical
School, Boston, Massachusetts (Chibnik, De Jager);
Program in Medical and Population Genetics, Broad
Institute, Cambridge Center, Cambridge,
Massachusetts (Chibnik, Srivastava, Pochet, Xu,
Kozubek, Obholzer, De Jager); Department of
Preventive Medicine, Rush University Medical
Center, Chicago, Illinois (Leurgans); Department of
Pathology, Rush University Medical Center, Chicago,
Illinois (Schneider); Epigenomics Program, Broad
Institute, Cambridge Center, Cambridge,
Massachusetts (Meissner); Harvard Stem Cell
Institute, Harvard University, Cambridge,
Massachusetts (Meissner).
Author Contributions: Dr Yu had full access to all
the data in the study and takes responsibility for the
integrity of the data and the accuracy of the data
analysis.
Study concept and design: Yu, Chibnik, Pochet,
Meissner, De Jager, Bennett.
Acquisition, analysis, or interpretation of data: Yu,
Chibnik, Srivastava, Pochet, Xu, Kozubek, Obholzer,
Schneider, Meissner, De Jager, Bennett.
Drafting of the manuscript: Yu, Pochet.
Critical revision of the manuscript for important
intellectual content: Chibnik, Srivastava, Pochet,
Yang, Xu, Kozubek, Obholzer, Leurgans, Schneider,
Meissner, De Jager, Bennett.
Statistical analysis: Yu, Yang, Leurgans.
Obtained funding: Schneider, De Jager, Bennett.
Administrative, technical, or material support: All
authors.
Study supervision: Pochet, Bennett.
Conflict of Interest Disclosures: None reported.
jamaneurology.com
derstand potential functional pathways that link epigenetic disruptions to the disease. By leveraging genome-wide DNA
methylation profiles and neuropathological data from 740
autopsied older persons, this is the first and largest study to
our knowledge that has interrogated brain DNA methylation
in AD loci for associations with multiple indexes for AD
pathology. Limitations are noted. Brain tissue came from a
single region of the dorsolateral prefrontal cortex. Our present data are not mature enough to derive potential effect
due to cellular composition, and future work needs to collect additional data and develop novel methods to identify
target cell types that drive the association of DNA methylation in each locus.
Funding/Support: This study was supported by
grants P30AG10161, R01AG15819, R01AG17917,
R01AG36042, R01AG36836, and U01AG46152
from the National Institutes of Health.
8. Lambert JC, Ibrahim-Verbaas CA, Harold D, et al;
European Alzheimer’s Disease Initiative (EADI);
Genetic and Environmental Risk in Alzheimer’s
Disease; Alzheimer’s Disease Genetic Consortium;
Cohorts for Heart and Aging Research in Genomic
Epidemiology. Meta-analysis of 74,046 individuals
identifies 11 new susceptibility loci for Alzheimer’s
disease. Nat Genet. 2013;45(12):1452-1458.
Role of the Funder/Sponsor: The funding source
had no role in the design and conduct of the study;
collection, management, analysis, and
interpretation of the data; preparation, review, or
approval of the manuscript; and decision to submit
the manuscript for publication.
Additional Contributions: We thank all the
participants of the Religious Orders Study and the
Rush Memory and Aging Project, as well as the staff
at the Rush Alzheimer’s Disease Center, for this
work.
REFERENCES
1. Gatz M, Pedersen NL, Berg S, et al. Heritability
for Alzheimer’s disease: the study of dementia in
Swedish twins. J Gerontol A Biol Sci Med Sci. 1997;52
(2):M117-M125.
2. Wilson RS, Barral S, Lee JH, et al. Heritability of
different forms of memory in the Late Onset
Alzheimer’s Disease Family Study. J Alzheimers Dis.
2011;23(2):249-255.
3. Hollingworth P, Harold D, Sims R, et al;
Alzheimer’s Disease Neuroimaging Initiative;
CHARGE Consortium; EADI1 Consortium. Common
variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33
and CD2AP are associated with Alzheimer’s disease.
Nat Genet. 2011;43(5):429-435.
4. Lambert JC, Heath S, Even G, et al; European
Alzheimer’s Disease Initiative Investigators.
Genome-wide association study identifies variants
at CLU and CR1 associated with Alzheimer’s disease.
Nat Genet. 2009;41(10):1094-1099.
9. Guerreiro R, Wojtas A, Bras J, et al; Alzheimer
Genetic Analysis Group. TREM2 variants in
Alzheimer’s disease. N Engl J Med. 2013;368(2):117127.
10. Jonsson T, Stefansson H, Steinberg S, et al.
Variant of TREM2 associated with the risk of
Alzheimer’s disease. N Engl J Med. 2013;368(2):
107-116.
11. Harms M, Benitez BA, Cairns N, et al;
NIA-LOAD/NCRAD Family Study Consortium.
C9orf72 hexanucleotide repeat expansions in
clinical Alzheimer disease. JAMA Neurol. 2013;70
(6):736-741.
12. Kohli MA, John-Williams K, Rajbhandary R, et al.
Repeat expansions in the C9ORF72 gene contribute
to Alzheimer disease in Caucasians. Neurobiol Aging.
2013;34(5):1519.e5-1519.e12. doi:10.1016/j
.neurobiolaging.2012.10.003.
13. Robertson KD. DNA methylation and human
disease. Nat Rev Genet. 2005;6(8):597-610.
14. Bakulski KM, Dolinoy DC, Sartor MA, et al.
Genome-wide DNA methylation differences
between late-onset Alzheimer’s disease and
cognitively normal controls in human frontal cortex.
J Alzheimers Dis. 2012;29(3):571-588.
15. Rakyan VK, Down TA, Balding DJ, Beck S.
Epigenome-wide association studies for common
human diseases. Nat Rev Genet. 2011;12(8):529-541.
5. Naj AC, Jun G, Beecham GW, et al. Common
variants at MS4A4/MS4A6E, CD2AP, CD33 and
EPHA1 are associated with late-onset Alzheimer’s
disease. Nat Genet. 2011;43(5):436-441.
16. Lu H, Liu X, Deng Y, Qing H. DNA methylation, a
hand behind neurodegenerative diseases. Front
Aging Neurosci. 2013;5:85. doi:10.3389/fnagi.2013
.00085.
6. Seshadri S, Fitzpatrick AL, Ikram MA, et al;
CHARGE Consortium; GERAD1 Consortium; EADI1
Consortium. Genome-wide analysis of genetic loci
associated with Alzheimer disease. JAMA. 2010;
303(18):1832-1840.
17. Zhang D, Cheng L, Badner JA, et al. Genetic
control of individual differences in gene-specific
methylation in human brain. Am J Hum Genet.
2010;86(3):411-419.
7. Harold D, Abraham R, Hollingworth P, et al.
Genome-wide association study identifies variants
at CLU and PICALM associated with Alzheimer’s
disease [published corrections appear in Nat Genet.
2009;41(10):1156 and 2013;45(6):712]. Nat Genet.
2009;41(10):1088-1093.
18. Bennett DA, Schneider JA, Arvanitakis Z, Wilson
RS. Overview and findings from the Religious
Orders Study. Curr Alzheimer Res. 2012;9(6):628645.
19. Bennett DA, Schneider JA, Buchman AS, Barnes
LL, Boyle PA, Wilson RS. Overview and findings
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
23
Research Original Investigation
Brain DNA Methylation and Pathological AD Diagnosis
from the Rush Memory and Aging Project. Curr
Alzheimer Res. 2012;9(6):646-663.
Alzheimer’s disease patients. Neurobiol Aging.
2013;34(9):2091-2099.
20. Bennett DA, Schneider JA, Arvanitakis Z, et al.
Neuropathology of older persons without cognitive
impairment from two community-based studies.
Neurology. 2006;66(12):1837-1844.
29. Rao JS, Keleshian VL, Klein S, Rapoport SI.
Epigenetic modifications in frontal cortex from
Alzheimer’s disease and bipolar disorder patients.
Transl Psychiatry. 2012;2:e132. doi:10.1038/tp.2012
.55.
21. The National Institute on Aging and Reagan
Institute Working Group on Diagnostic Criteria for
the Neuropathological Assessment of Alzheimer’s
Disease. Consensus recommendations for the
postmortem diagnosis of Alzheimer’s disease.
Neurobiol Aging. 1997;18(4 suppl 1):S1-S2.
22. Braak H, Braak E. Neuropathological stageing
of Alzheimer-related changes. Acta Neuropathol
(Berl). 1991;82:239-259.
23. Mirra SS, Heyman A, McKeel D, et al. The
Consortium to Establish a Registry for Alzheimer’s
Disease (CERAD), part II: standardization of the
neuropathologic assessment of Alzheimer’s
disease. Neurology. 1991;41:479-486.
24. Fisher RA. Questions and answers #14. Am Stat.
1948;2(5):30-31.
25. Benjamini Y, Hochberg Y. Controlling the false
discovery rate: a practical and powerful approach to
multiple testing. J R Stat Soc B. 1995;57:289-300.
26. Baron RM, Kenny DA. The moderator-mediator
variable distinction in social psychological research:
conceptual, strategic, and statistical considerations.
J Pers Soc Psychol. 1986;51(6):1173-1182.
27. Mastroeni D, Grover A, Delvaux E, Whiteside C,
Coleman PD, Rogers J. Epigenetic changes in
Alzheimer’s disease: decrements in DNA
methylation. Neurobiol Aging. 2010;31(12):20252037.
28. Chouliaras L, Mastroeni D, Delvaux E, et al.
Consistent decrease in global DNA methylation and
hydroxymethylation in the hippocampus of
24
30. De Jager PL, Srivastava G, Lunnon K, et al.
Alzheimer’s disease: early alterations in brain DNA
methylation at ANK1, BIN1, RHBDF2 and other loci.
Nat Neurosci. 2014;17(9):1156-1163.
31. Furuya TK, da Silva PN, Payão SL, et al. SORL1
and SIRT1 mRNA expression and promoter
methylation levels in aging and Alzheimer’s Disease.
Neurochem Int. 2012;61(7):973-975.
32. Rogaeva E, Meng Y, Lee JH, et al. The neuronal
sortilin-related receptor SORL1 is genetically
associated with Alzheimer disease. Nat Genet.
2007;39(2):168-177.
33. Chan SL, Kim WS, Kwok JB, et al. ATP-binding
cassette transporter A7 regulates processing of
amyloid precursor protein in vitro. J Neurochem.
2008;106(2):793-804.
34. Kim WS, Li H, Ruberu K, et al. Deletion of Abca7
increases cerebral amyloid-β accumulation in the
J20 mouse model of Alzheimer’s disease. J Neurosci.
2013;33(10):4387-4394.
35. Chen KL, Wang SS, Yang YY, Yuan RY, Chen RM,
Hu CJ. The epigenetic effects of amyloid-β1-40 on
global DNA and neprilysin genes in murine cerebral
endothelial cells. Biochem Biophys Res Commun.
2009;378(1):57-61.
36. Itoh T, De Camilli P. BAR, F-BAR (EFC) and
ENTH/ANTH domains in the regulation of
membrane-cytosol interfaces and membrane
curvature. Biochim Biophys Acta. 2006;1761(8):
897-912.
37. Koo EH, Squazzo SL. Evidence that production
and release of amyloid β-protein involves the
endocytic pathway. J Biol Chem. 1994;269(26):
17386-17389.
38. Chapuis J, Hansmannel F, Gistelinck M, et al;
GERAD Consortium. Increased expression of BIN1
mediates Alzheimer genetic risk by modulating tau
pathology. Mol Psychiatry. 2013;18(11):1225-1234.
39. Larsson M, Duffy DL, Zhu G, et al. GWAS
findings for human iris patterns: associations with
variants in genes that influence normal neuronal
pattern development. Am J Hum Genet. 2011;89(2):
334-343.
40. Kajiho H, Saito K, Tsujita K, et al. RIN3: a novel
Rab5 GEF interacting with amphiphysin II involved
in the early endocytic pathway. J Cell Sci. 2003;116(pt
20):4159-4168.
41. Glennon EB, Whitehouse IJ, Miners JS, et al.
BIN1 is decreased in sporadic but not familial
Alzheimer’s disease or in aging. PLoS One. 2013;8
(10):e78806. doi:10.1371/journal.pone.0078806.
42. Van den Hove DL, Kompotis K, Lardenoije R,
et al. Epigenetically regulated microRNAs in
Alzheimer’s disease. Neurobiol Aging. 2014;35(4):
731-745.
43. van Eijk KR, de Jong S, Boks MP, et al. Genetic
analysis of DNA methylation and gene expression
levels in whole blood of healthy human subjects.
BMC Genomics. 2012;13:636. doi:10.1186/1471
-2164-13-636.
44. Gibbs JR, van der Brug MP, Hernandez DG,
et al. Abundant quantitative trait loci exist for DNA
methylation and gene expression in human brain.
PLoS Genet. 2010;6(5):e1000952. doi:10.1371
/journal.pgen.1000952.
JAMA Neurology January 2015 Volume 72, Number 1
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 06/14/2020
jamaneurology.com