LETTER
doi:10.1038/nature10849
An epigenetic blockade of cognitive functions in the
neurodegenerating brain
Johannes Gräff1,2,3, Damien Rei1,2, Ji-Song Guan1,2,3, Wen-Yuan Wang1,2,3, Jinsoo Seo1,2, Krista M. Hennig3,4,
Thomas J. F. Nieland3, Daniel M. Fass3,4, Patricia F. Kao5, Martin Kahn1, Susan C. Su1,2, Alireza Samiei1, Nadine Joseph1,2,3,
Stephen J. Haggarty3,4, Ivana Delalle5 & Li-Huei Tsai1,2,3
Cognitive decline is a debilitating feature of most neurodegenerative diseases of the central nervous system, including Alzheimer’s
disease1. The causes leading to such impairment are only poorly
understood and effective treatments are slow to emerge2. Here we
show that cognitive capacities in the neurodegenerating brain are
constrained by an epigenetic blockade of gene transcription that
is potentially reversible. This blockade is mediated by histone
deacetylase 2, which is increased by Alzheimer’s-disease-related
neurotoxic insults in vitro, in two mouse models of neurodegeneration and in patients with Alzheimer’s disease. Histone
deacetylase 2 associates with and reduces the histone acetylation
of genes important for learning and memory, which show a
concomitant decrease in expression. Importantly, reversing the
build-up of histone deacetylase 2 by short-hairpin-RNA-mediated
knockdown unlocks the repression of these genes, reinstates structural and synaptic plasticity, and abolishes neurodegenerationassociated memory impairments. These findings advocate for the
development of selective inhibitors of histone deacetylase 2 and
suggest that cognitive capacities following neurodegeneration are
not entirely lost, but merely impaired by this epigenetic blockade.
Long-lasting forms of memories require stable gene expression
changes3, which are in part orchestrated by chromatin-templated
epigenetic processes4. Of the epigenetic modifications identified so
far in the nervous system, histone acetylation has been unequivocally
associated with facilitating learning and memory4. Acetylation
diminishes the electrostatic affinity between neighbouring histones
and the DNA and, consequently, can promote a more open chromatin
structure that allows for memory-related gene transcription5.
Over the past decade, several studies have reported sporadic cases of
reduced histone acetylation in animal models of neurodegeneration that
are characterized by cognitive decline, including models of Alzheimer’s
disease4. Accordingly, pharmacological treatments aimed at increasing
histone acetylation have shown promising results in reversing cognitive
deficits in some of these models, predominantly by the use of nonselective histone deacetylase (HDAC) inhibitors6. However, the causative
agent of such memory-impairing histone acetylation changes, and,
hence, the best targets for pharmacological strategies, remain unknown.
One likely candidate is HDAC2, a class I HDAC that negatively regulates
memory and synaptic plasticity in the healthy mouse brain7,8.
To investigate whether HDAC2 mediates cognitive deficits associated
with neurodegeneration, we measured its abundance in CK-p25 mice9,10,
which inducibly and forebrain-specifically overexpress p25, a truncated
version of p35. p25 aberrantly activates cyclin-dependent kinase 5
(CDK5), and is implicated in various neurodegenerative diseases11,
including Alzheimer’s disease12. After 6 weeks of p25 induction, CKp25 mice display Alzheimer’s-disease-related pathologies such as neuronal loss9, b-amyloid accumulation10, reactive astrogliosis9 and reduced
synaptic density13, most prominently in the hippocampus and the cortex,
two brain areas important for memory formation and storage, respectively14. Accordingly, 6-week-induced CK-p25 (hereafter referred to as
CK-p25) mice also display spatial and associative memory deficits13.
Using immunohistochemistry and western blot analysis, we
found that HDAC2 was significantly increased in neuronal nuclei in
hippocampal area CA1 in CK-p25 mice compared with control
littermates (Fig. 1a, d, e; see Supplementary Fig. 2a for a specificity
control for the HDAC2 signal). No changes in HDAC2 were observed
in hippocampal area CA3 or the dentate gyrus (Supplementary Fig. 3a,
b), explaining the overall marginal increase in the entire hippocampus.
Interestingly, this effect appears to be non-cell-autonomous, as both
p25-positive and p25-negative cells displayed elevated HDAC2
(Fig. 1a). In contrast, levels of the structurally highly related
HDAC1, and of HDAC3, another class I HDAC involved in memory
formation15, were not changed (Fig. 1b, c, e). Furthermore, HDAC2,
but not HDAC1 or HDAC3, was also increased in the prefrontal
cortex of CK-p25 mice (Supplementary Fig. 4), whereas in the amygdala, a brain area not affected by neurodegeneration in the CK-p25
mice, its levels remained unchanged (Supplementary Fig. 3c). This
neurodegeneration-associated increase of HDAC2 was confirmed in
another mouse model of Alzheimer’s-disease-related pathologies and
cognitive decline, the 5XFAD mouse16,17 (Supplementary Fig. 5).
Next, we aimed to determine the functional consequences of elevated
HDAC2. Because HDAC2 has been shown to associate with the promoter region of genes involved in memory and synaptic plasticity7, we
proposed that it is enriched at these genes following neurodegeneration. Of the known HDAC2 targets7, we focused on those that, in
several independent studies, had been demonstrated to be downregulated in the human brain with Alzheimer’s disease (Supplementary
Table 1). These include the immediate-early genes Arc, Bdnf exons I,
II and IV, Egr1, Homer1, and Cdk5, implicated in learning and
memory, and genes related to synaptic plasticity such as the glutamate
receptor subunits GluR1, GluR2, NR2A and NR2B (also known as Gria1,
Gria2, Grin2a and Grin2b), as well as Nfl (neurofilament light chain, also
known as Nefl), Syp (synaptophysin) and Syt1 (synaptotagmin 1). Using
chromatin immunoprecipitation (ChIP, for primers see Supplementary
Table 2), we found that HDAC2 is significantly enriched at these genes
in the CK-p25 hippocampus, the exception being the promoter regions
of the activity-dependent Bdnf exons I and II18, and the housekeeping
genes b-actin, b-globin and b-tubulin (Fig. 1f). In contrast, binding of
HDAC1 and HDAC3 was unaltered (Supplementary Fig. 6a, b).
Interestingly, in agreement with previous reports showing that
HDAC2 can also bind to a gene’s coding region19, we also found
HDAC2 more abundantly bound to the coding sequence of the same
genes (see Supplementary Fig. 7 and Supplementary Table 3 for primer
sequences).
1
Picower Institute for Learning and Memory, Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. 2Howard Hughes Medical
Institute, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. 3Stanley Center for Psychiatric Research, Broad Institute of Harvard University and Massachusetts Institute of
Technology, Cambridge, Massachusetts 02142, USA. 4Center for Human Genetic Research, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts 02114, USA. 5Department of
Pathology and Laboratory Medicine, Boston University School of Medicine, Boston, Massachusetts 02118, USA.
2 2 2 | N AT U R E | VO L 4 8 3 | 8 M A R C H 2 0 1 2
©2012 Macmillan Publishers Limited. All rights reserved
*
*
*
*
**
0
HDAC 2
1
3
Housekeeping
CON
*
CK-p25
*
*
*
**
**
0
1.5
1.0
H4K12 acetylation
*
*
*
*
0
**
**
*
*
**
*
1.0
*
*
*
*
*
* 1
0
*
SYP/
DAPI
0
h
1.5 mRNA expression
1.0
0
*
f
*
*
*
*
*
*
*
*
**
*
*
**
*
*
*
*
*
βAc
β- tin
G
lo
β- bin
Tu
bu
lin
B
Sy
p
Sy
t1
N
R2
fl
A
N
R2
N
R1
lu
lu
G
R2
0
G
mRNA
(fold change CON)
i
1
*
*
2
*
RNA Pol II binding
Ar
Bd c
n
Bd f I
n
Bd f II
nf
IV
C
dk
5
E
H gr1
om
er
1
RNA Pol II ChIP
(fold change CON)
h
1.5
3
2 **
Figure 1 | Elevated HDAC2 levels epigenetically block the expression of
neuroplasticity genes during neurodegeneration. a–c, Representative
immunohistochemical images depicting HDAC1–3 levels in area CA1 of CK-p25
mice and control littermates; scale bar, 20 mm. d, Quantitative assessment of
a–c (n 5 3–6 slices from three or four mice each). e, Representative western blot
images and quantification of HDAC1–3 in the CK-p25 and control hippocampus
(n 5 6–9 mice each). f–h, Quantitative PCR results of (f) HDAC2-,
(g) AcH4K12- and (h) RNA Pol II-immunoprecipitated chromatin at the
promoter of neuroplasticity and housekeeping genes in the CK-p25 and control
hippocampus. i, Quantitative RT–PCR results of the same genes (f–i, n 5 4–8
animals each). *P # 0.05; **P # 0.01; ***P # 0.001; values are mean 6 s.e.m.
We next assessed the acetylation of several histone (H) residues in the
promoter region of these genes, for which acetylation has been shown to
be important for learning, memory, and synaptic plasticity, such as H2B
lysine (K) 5, H3K14, H4K5 and H4K12 (ref. 4). ChIP analyses revealed a
hypoacetylation for all residues at the neuroplasticity genes (Fig. 1g and
Supplementary Fig. 6c–e), albeit to different extents. Importantly, the
acetylation of housekeeping genes was not altered. The effects of
elevated HDAC2 levels further appear to be restricted to histones, as
we found no overall acetylation changes on other proteins regulated by
this modification, such as tau (also known as MAPT), protein 53 (p53;
also known as TP53) and tubulin, nor in overall nuclear or cytoplasmic protein acetylation (Supplementary Fig. 8).
Next, to determine the functional consequences of promoter
hypoacetylation, we assessed the binding of activated (that is, phosphorylated) RNA polymerase II (RNA Pol II), and found it to be markedly
reduced (Fig. 1h). This prompted us to measure the messenger RNA
(mRNA) expression of these genes by quantitative reverse-transcription
PCR (RT–PCR) (primers in Supplementary Table 4). We found
reduced expression for all genes with elevated HDAC2 binding and
CON,scr
CK-p25,scr
CK-p25,shHDAC2
60
** *
50
40
30
20
10
0
25
i
*
1
2
3 4
Day
2
*
0
*
2
*
*
1
5
**
75
1
0
*
50
25
j
CON,scr
CK-p25,scr
CK-p25,shHDAC2
70
60
50
40
30
20
10
0
CK-p25,scr
CK-p25,shHDAC2
CON,scr
CK-p25,scr
CK-p25,shHDAC2
*
**
50
expression
*
e
*
75
d mRNA
3
Escape latency (s)
2
c H4K12 acetylation
Mean SYP
optical density (a.u.)
4
100
0
0.5
Synaptic plasticity
*
*
*
* *
**
*
**
*
** **
GluR1
GluR2
Nfl
NR2A
NR2B
Syp
Syt1
Learning and memory
HDAC2 binding
125
g
CON,
scr
Merge
CON, scr
CK-p25, scr
CK-p25, shHDAC2
MAP2/
DAPI
CK-p25, CK-p25,
shHDAC2
scr
Merge
b
Merge
HDAC2
(mean grey value)
Merge
HDAC2
HDAC1
HDAC3
β-Tubulin
*
1.0
3
DAPI
HDAC2
Arc
Bdnf I
Bdnf II
Bdnf IV
Cdk5
Egr1
Homer1
DAPI
1
GFP
Stratum
radiatum
AcH4K12 ChIP
(fold change CON)
g
6
DAPI
0
HDAC 2
Freezing (% of time)
HDAC2 ChIP
(fold change CON)
f
DAPI
1.0
a
0
Time spent in quadrant (s)
e
CON
CK-p25
***
mRNA
(fold change CON,scr)
HDAC3
2.0
GluR1
GluR2
Nfl
NR2A
NR2B
Syp
Syt1
HDAC1
HDAC2
d
Arc
Bdnf I
Bdnf II
Bdnf IV
Cdk5
Egr1
Homer1
CK-p25
GFP
CON,
scr
c CON
CK-p25
GFP
CK-p25, CK-p25,
shHDAC2
scr
b CON
AcH4K12 ChIP
(fold change CON,scr)
CK-p25
GFP
HDAC level CA1
(fold change CON)
a CON
HDAC level hippocampus
(fold change CON)
LETTER RESEARCH
CON,
scr
O
R
L
T
CK-p25,
scr
O
R
L
T
CK-p25,
shHDAC2
O
R
L
T
30 ***
30
30 ***
20
20
20
10
10
10
0
T RO L
0
T RO L
0
T RO L
Figure 2 | Reducing HDAC2 levels alleviates memory deficits.
a, Representative immunohistochemical images depicting HDAC2 in
hippocampal area CA1 of CK-p25, shHDAC2, CK-p25, scr and CON, scr
animals; scale bar, 20 mm. b, Quantitative assessment of a, n 5 4 or 5 sections from
four mice each. c, Quantitative PCR results of AcH4K12-immunoprecipitated
chromatin in CK-p25, scr and CK-p25, shHDAC2 compared with CON, scr
mice. d, Quantitative RT–PCR results of the same genes. (c, d, n 5 4–6 animals
each.) e, g, Representative immunohistochemical images depicting (e) SYP and
(g) MAP2 immunoreactivity in the hippocampus stratum radiatum; scale bars:
e, 25 mm; g, 20 mm. f, Quantitative assessment of e, n 5 4 mice each; a.u., arbitrary
units. h, Freezing responses of CON, scr (n 5 18), CK-p25, scr (n 5 16) and CKp25, shHDAC2 (n 5 16) mice 24 h after contextual fear conditioning. i, Escape
latencies in a water maze task of CON, scr (n 5 19), CK-p25, scr (n 5 17) and CKp25, shHDAC2 (n 5 19) animals. Data points are averages of two trials per day.
j, Representative swim traces and time spent per quadrant during the water maze
test (T, target quadrant; R, right; O, opposite; L, left of target). *P # 0.05;
**P # 0.01; ***P # 0.001; values are mean 6 s.e.m.
a concomitant decrease in histone acetylation and RNA Pol II binding
(Fig. 1i). Of note, HDAC2 probably acts with the transcriptional
repressor complexes CoREST, NuRD and SIN3, as we found increased
association of HDAC2 with these complexes in hippocampal CK-p25
extracts by co-immunoprecipitation (Supplementary Fig. 9). Taken
together, these results indicate that HDAC2 mediates a local chromatin
compaction of neuroplasticity genes, which decreases their expression
and may contribute to cognitive decline during neurodegeneration.
To examine causally such a possibility, we generated adeno-associated
viral vectors carrying either short-hairpin RNAs (shRNAs) directed
against HDAC2 or scrambled control shRNAs (Supplementary Fig.
8 M A R C H 2 0 1 2 | VO L 4 8 3 | N AT U R E | 2 2 3
©2012 Macmillan Publishers Limited. All rights reserved
RESEARCH LETTER
and physiological changes in the surviving neurons. To this end, we
measured synaptic density by SYP immunohistochemistry, labelling
the presynaptic terminals of functional synapses, and dendritic abundance by microtubule-associated protein 2 (MAP2) immunoreactivity.
We found that, whereas SYP and MAP2 levels were reduced in the
stratum radiatum of CK-p25, scr animals (as previously described for
CK-p25 mice13,20), their abundance was markedly increased in CK-p25,
shHDAC2 animals, to levels comparable to CON, scr animals (Fig. 2e–g).
Interestingly, however, the number of surviving neurons, as evidenced by
NeuN immunohistochemistry, was not altered by HDAC2 reduction
(Supplementary Fig. 12). Next, we measured synaptic plasticity by electrophysiological recordings and observed that long-term potentiation in
CA1 neurons was undistinguishable between CK-p25, shHDAC2 and
CON, scr animals, but significantly improved over CK-p25, scr animals.
A similar effect was observed for basal synaptic plasticity (Supplementary
Fig. 13). Together, these data indicate that, although HDAC2 normalization did not impact neuronal survival, it did reinstate morphological
and synaptic plasticity in the surviving neurons.
We suggested that the reduction of HDAC2 would also counteract
the cognitive deficits associated with neurodegeneration. For this, we
assessed associative and spatial memory on a fear-conditioning and
water maze task, respectively, two types of hippocampus-dependent
memory that are severely perturbed in CK-p25 animals13. We observed
2
*
*
1
0
f
Amyloid-β42–1
Amyloid-β1–42
Hdac2 mRNA
(fold change CON)
2
H2O
H2O2
1
0
0
H2O
H2O2
1.5
***
1.0
0.5
0
CON
CK-p25
g
GRE consensus sequence
2
GRE sequence in
Hdac2 promoter
**
1
DAPI
d
+25
+42
h
2
**
1
l
Amyloid-β42–1
Amyloid-β1–42
1.5
**
0.5
0
1.5
***
1.0
0.5
0
CON
CK-p25
j
90 CON
70
70
50
50
30
30
k
90 CK-p25
30
50 70 90 30 50 70
pGR1 (mean grey value)
90
1.5
5
0
luc
1.0
*
0.5
**
0
Figure 3 | Neurotoxic insults increase HDAC2 through stress elements in its
promoter. a, c, Representative pictures of HDAC2 and pGR1 labelling of
primary hippocampal neurons treated with (a) H2O2 and (c) amyloid-boligomers (n 5 20–40 neurons per group); scale bar, 10 mm.
b, d, Quantification of a and c. e, f, Quantitative RT–PCR results showing
Hdac2 expression in (e) H2O2- and amyloid-b-treated primary hippocampal
neurons, and (f) in the CK-p25 hippocampus (n 5 7–9 mice each).
g, Alignment of the vertebrate GRE consensus sequence with the GRE in the
proximal promoter of mouse Hdac2. h, Quantification and representative
western blot images of hippocampal extracts of CK-p25 versus control mice
(n 5 3 each). i, Representative images of immunohistochemical labelling of
pGR1 and HDAC2 in the CK-p25 hippocampus (n 5 3–6 slices from three
mice each); scale bar, 20 mm. j, Regression analysis of i showing a significant
correlation between pGR1 and HDAC2 in CK-p25 (R2 5 0.686, P # 0.001), but
*
Cdk5 fl/fl
Cdk5cKO
CDK5
pGR1
GR1
HDAC2
β-Actin
luc
***
**
***
***
***
5
4
3
2
1
0
H2O2 −
Amyloid-β1–42 −
pGR1/
CDK5 GR1 HDAC2
**
CON
CK-p25
10
+
−
−
+
−
−
luc
Luciferase activity
(fold change CON)
DAPI
+300
15
m
0
Protein (f.c. CON)
pGR1
HDAC2
CON
CK-p25
HDAC2
TSS
−1000
n
i
GRE
1.0
pGR1
GR1
β-Tubulin
0
Hdac2 promoter region
pGR1 binding
(fold change CON)
0.5
pGR1
HDAC2 level
(fold change
CON)
1.0
HDAC2
pGR1 level
(fold change
CON)
H2O2
Hdac2 mRNA
(fold change CON)
e
c
***
1.5
pGR1/GR1
(fold change CON)
b
DAPI
Amyloid-β1–42 Amyloid-β42–1
pGR1
H2O
HDAC2
pGR1 level
HDAC2 level
(fold change CON) (fold change CON)
a
Luciferase activity
(fold change CON)
10a, b). The knockdown efficiency of the shRNAs in culture was about
25–30% (Supplementary Fig. 10c–f), ideally suited for the targeted normalization of the 20–50% increase of HDAC2 in the CK-p25 mice. We
injected these vectors bilaterally into hippocampal area CA1 of 2-weekinduced CK-p25 and control mice, and assessed HDAC2 levels 4 weeks
after viral injection (Supplementary Fig. 10g). CK-p25 animals injected
with an shRNA against HDAC2 (CK-p25, shHDAC2) showed reduced
HDAC2 levels compared with CK-p25 animals injected with control
scrambled shRNA (CK-p25, scr), indistinguishable from control mice
injected with scrambled shRNA (CON, scr) (Fig. 2a, b). Protein levels of
HDAC1 and HDAC3 remained unchanged (Supplementary Fig. 11).
We observed transduction efficiencies of 53–61% (mean 6 s.e.m.
57.4 6 2.5; n 5 3–4 mice per group analysed) and comparable infection
rates in both control and CK-p25 hippocampi (Supplementary Fig. 10h).
Next, we sought to determine whether reducing HDAC2 levels
would alter the promoter histone acetylation and mRNA expression
of neuroplasticity genes. We found that H4K12 acetylation was significantly enhanced on most of these genes, and they showed increased
expression (Fig. 2c, d). Importantly, most of these genes showed comparable, or even higher, expression in CK-p25, shHDAC2 mice compared with CON, scr animals (black dotted lines).
Based on these findings, we investigated whether such regained chromatin and transcriptional plasticity might translate into morphological
***
3
**
+
−
−
+
GR
GRS221A
***
***
2
1
0
H2O2 −
Amyloid-β1–42 −
−
−
+
−
+
−
−
+
−
+
not control mice (R2 5 0.019, not significant). k, Quantification and
representative western blot images of Cdk5cKO and control Cdk5fl/fl forebrain
extracts (n 5 3 each). f.c., fold change. l, Quantitative PCR results of pGR1immunoprecipitated chromatin around the GRE in a 1.3-kb-wide Hdac2
promoter region (schematically shown above the graph; TSS, transcriptional
start site) in the CK-p25 and control hippocampus (n 5 3–6 animals each);
green lines represent fragments amplified by primer pairs. m, Luciferase
activity of CAD cells transfected with the Hdac2 promoter with (orange) or
without (blue) GRE (schematic of constructs shown above graph), and treated
with H2O2 and amyloid-b1–42. n, Luciferase activity of CAD cells transfected
with Hdac2-GRE in the presence of endogenous glucocorticoid receptor or of
cotransfected GRS211A. In vitro results are from at least three independent
experiments. *P # 0.05; **P # 0.01; ***P # 0.001; values are mean 6 s.e.m.
2 2 4 | N AT U R E | VO L 4 8 3 | 8 M A R C H 2 0 1 2
©2012 Macmillan Publishers Limited. All rights reserved
LETTER RESEARCH
that the Hdac2-GRE also responds to neurotoxic stimuli. When CAD
cells were treated with either H2O2 or amyloid-b1–42, luciferase activity
was significantly increased, but only with the GRE present (Fig. 3m,
orange bars). Furthermore, the capability of GR to activate Hdac2
critically depends upon its phosphorylation. When S211 was mutated
to alanine (GRS211A), the glucocorticoid receptor was no longer capable
of activating Hdac2 in vitro (Fig. 3n). This dependency on S211 phosphorylation also occurs in vivo, as Cdk5cKO forebrain extracts26 had
reduced pGR1 and HDAC2 levels (Fig. 3k). Taken together, these results
indicate that Alzheimer’s-disease-related neurotoxic stimuli lead to an
increase in Hdac2 gene transcription by mechanisms involving glucocorticoid receptor activation and its interaction with the Hdac2-GRE.
Lastly, to assess the relevance of these findings in humans, we compared HDAC2 abundance in post-mortem brain samples from
patients with varying degrees of non-familial Alzheimer’s disease,
the most common form of neurodegeneration-associated dementia
worldwide27. The cases used here (Supplementary Table 5) are defined
by the Braak and Braak stages28, which are characterized by the accumulation of hyperphosphorylated tau protein in the cortices (Supplementary Fig. 18), and by increasing neurodegeneration28 and
cognitive impairment29. We found that, in all Alzheimer’s-diseaserelated Braak and Braak stages, HDAC2 levels were significantly elevated in hippocampal area CA1 (Fig. 4a, d) and the entorhinal cortex
(Supplementary Fig. 19a, d), which are among the earliest and most
affected brain areas in Alzheimer’s disease2 and crucial for memory
formation and storage14. HDAC2 accumulation was visible beginning
2.0
c
HDAC1
HDAC3
CON
CON
CON
BB I–II
BB I–II
BB I–II
BB III–IV
BB III–IV
BB III–IV
BB V–VI
BB V–VI
BB V–VI
**
1.5
1.0
0.5
0
e
2.0
1.5
1.0
0.5
0
f
HDAC3 protein level
(fold change CON)
d
b
HDAC2
HDAC1 protein level
(fold change CON)
a
HDAC2 protein level
(fold change CON)
that associative memory of CK-p25, shHDAC2 animals returned to
levels of CON, scr animals (Fig. 2h). Likewise, CK-p25, shHDAC2
animals showed significantly reduced escape latencies compared with
CK-p25, scr animals during training in the water maze (Fig. 2i) and,
24 h later, they spent significantly more time in the target quadrant,
indistinguishable from the performance of CON, scr mice (Fig. 2j).
Overall, swimming behaviour was similar between the different groups
(Supplementary Fig. 14a, b), and altering HDAC2 levels per se did not
affect locomotor activity or anxiety as assessed by an open field test
(Supplementary Fig. 14c–f). Together, these results indicate that elevated HDAC2 levels are causally involved in the cognitive decline
associated with neurodegeneration in CK-p25 mice, but that the prevention of HDAC2 upregulation rescues memory capacities.
To gain insight into the mechanisms underlying the increase in
HDAC2, we exposed primary hippocampal neurons to neurotoxic
stimuli characteristic of Alzheimer’s-disease-related neurodegeneration, hydrogen peroxide (H2O2) and amyloid-b oligomers21,22. As
revealed by immunocytochemistry and western blot analysis, treatment with either H2O2 or amyloid-b1–42, but not control amyloidb42–1, oligomers was sufficient to increase HDAC2 (Fig. 3a–d, left
panels, and Supplementary Fig. 15a–d). Importantly, both neurotoxic
stimuli increased Hdac2 at the mRNA level (Fig. 3e), and increased
Hdac2 transcription was also evident in the CK-p25 hippocampus
(Fig. 3f), suggesting the involvement of transcriptional mechanisms.
This prompted us to screen the Hdac2 promoter for potential binding sites of transcriptional regulators. Using transcription factor binding databases23, we found a well-conserved recognition element for the
glucocorticoid receptor 1 (GR1, also known as NR3C1) in the proximal
promoter region of Hdac2 (Fig. 3g). Glucocorticoid receptors are activated by phosphorylation following behavioural or cellular stress and,
upon binding to the glucocorticoid responsive element (GRE) in a
gene’s promoter region, they can act as transcriptional activators or
repressors, depending, in part, on the residue phosphorylated24. Of its
known phosphorylation sites, serine (S) 211 has been robustly associated with activated forms of GR1 (ref. 25).
Based on this knowledge, we examined whether S211 phosphorylation on GR1 was increased after neurotoxic insults in vitro, and in the
CK-p25 brain in vivo. Immunocytochemical labelling and western
blot analysis of cultured hippocampal neurons following H2O2 and
amyloid-b1–42 treatment revealed a significantly increased phosphorylation of GR1 (pGR1) on S211 compared with control conditions
(Fig. 3a–d, middle panels, and Supplementary Fig. 15a–d; see Supplementary Fig. 2b for anti-pGR1S211 specificity). Furthermore, the
CK-p25 hippocampus showed similarly increased pGR1 levels
(Fig. 3h, i and Supplementary Fig. 16). Remarkably, we observed that
the increase in pGR1 occurred concurrently with that of HDAC2
following neurotoxicity (Fig. 3a, c, i, j) and, using forebrain extracts
of conditional Cdk5 knockout (Cdk5cKO) and control mice26, we identified CDK5 as a GR1 kinase in vivo (Fig. 3k).
We then sought to determine whether GR1 phosphorylation
increases Hdac2 transcription. We first examined the binding of
pGR1 to the Hdac2 promoter by pGR1-ChIP and primer pairs (Supplementary Table 2) spanning a region from 21000 to 1300 base pairs
around the Hdac2 transcriptional start site (Fig. 3l, top). We found that
pGR1 binding to the Hdac2-GRE was significantly increased in the CKp25 hippocampus (Fig. 3l, bottom). Similar results were obtained after
amyloid-b1–42 treatment of primary hippocampal cultures (Supplementary Fig. 15e). Second, to determine whether GR1 is directly
capable of transcriptionally activating Hdac2, we cloned the Hdac2 promoter with and without the GRE into a luciferase construct and tested its
activity in CAD cells, a primary neuron-like cell line. We found that the
presence of the GRE alone increased the luciferase activity by approximately threefold but that, upon the addition of a constitutively active
form of glucocorticoid receptor, GR526 (see Methods), this activity was
further doubled. However, without the Hdac2-GRE, the addition of
GR526 had no effect (Supplementary Fig. 17). Importantly, we found
2.0
1.5
CON
BB I–II
BB III–IV
BB V–VI
1.0
0.5
0
Figure 4 | HDAC2 expression is increased in patients with Alzheimer’s
disease. a–c, Representative immunohistochemical images depicting nuclear
HDAC1–3 levels (white dotted circles) in neurons (arrow points to magnified
neuron in inset) of hippocampal area CA1 from patients with Braak and Braak
(BB) stages I–II (n 5 4), III–IV (n 5 7) and V–VI (n 5 8) compared with
healthy BB0 control brains (CON, n 5 7); scale bar, 100 mm. d–f, Quantitative
assessment of a–c. **P # 0.01; values are mean 6 s.e.m.
8 M A R C H 2 0 1 2 | VO L 4 8 3 | N AT U R E | 2 2 5
©2012 Macmillan Publishers Limited. All rights reserved
RESEARCH LETTER
at Braak and Braak stage I/II, implicating it as an early event in the
progression of Alzheimer’s disease. In contrast, levels of HDAC1 and
HDAC3 were not altered (Fig. 4b, c, e, f and Supplementary Fig. 19b, c,
e, f). Thus, elevated levels of HDAC2 may also accompany the cognitive decline of the human neurodegenerating brain.
The findings presented in this study describe that epigenetic mechanisms substantially contribute to the cognitive decline associated with
Alzheimer’s disease-related neurodegeneration. Although it is well
documented that neuronal loss and amyloid-b- or tau-induced
neurotoxicity acutely disable synaptic functions, in turn leading to
cognitive deficits1,2,30, the HDAC2-mediated epigenetic blockade of
neuroplasticity-related gene expression could delineate a process by
which memory functions become permanently impaired in the
Alzheimer’s disease brain (Supplementary Fig. 1a). This blockade
appears to be induced by GR1; thus, glucocorticoid receptors may function as molecular mediators between neurodegeneration-associated
neurotoxic stressors and cognitive impairment (Supplementary Fig. 1b).
Intriguingly, our findings may also provide a potential explanation, at
least in part, as to why, in some clinical trials, cognitive impairments in
patients with Alzheimer’s disease persist despite successful amyloid-b
clearance2: once the epigenetic blockade is in place, reducing amyloid-b
generation and deposition alone may not be sufficient to rescue against
cognitive dysfunction. A more efficacious strategy may therefore lie
in the combination of amyloid-b reduction with the inhibition of
HDAC2. By extension, these findings pinpoint HDAC2 as the
probable target of non-selective HDAC inhibitors that counteract
cognitive decline in mouse models of Alzheimer’s disease6 and, as a
result, strongly advocate for the development of HDAC2-selective
inhibitors. Finally, our finding that HDAC2 inhibition probably reinstates transcriptional, morphological and synaptic plasticity in the
surviving neurons of the neurodegenerating brain raises hope that
such plasticity is not irrevocably lost, but merely constrained by the
epigenetic blockade.
METHODS SUMMARY
Please refer to Methods for more detail. Human material was used with informed
consent from all donors, and all mouse work was approved by the Committee for
Animal Care of the Division of Comparative Medicine at the Massachusetts
Institute of Technology.
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 25 July 2011; accepted 13 January 2012.
Published online 29 February 2012.
1.
Walsh, D. M. & Selkoe, D. J. Deciphering the molecular basis of memory failure in
Alzheimer’s disease. Neuron 44, 181–193 (2004).
2. Holtzman, D. M., Morris, J. C. & Goate, A. M. Alzheimer’s disease: the challenge of
the second century. Sci. Transl. Med. 3, 77sr1 (2011).
3. Kandel, E. R. The molecular biology of memory storage: a dialogue between genes
and synapses. Science 294, 1030–1038 (2001).
4. Gräff, J., Kim, D., Dobbin, M. M. & Tsai, L. H. Epigenetic regulation of gene
expression in physiological and pathological brain processes. Physiol. Rev. 91,
603–649 (2011).
5. Brownell, J. E. & Allis, C. D. Special HATs for special occasions: linking histone
acetylation to chromatin assembly and gene activation. Curr. Opin. Genet. Dev. 6,
176–184 (1996).
6. Kazantsev, A. G. & Thompson, L. M. Therapeutic application of histone deacetylase
inhibitors for central nervous system disorders. Nature Rev. Drug Discov. 7,
854–868 (2008).
7. Guan, J. S. et al. HDAC2 negatively regulates memory formation and synaptic
plasticity. Nature 459, 55–60 (2009).
8. Akhtar, M. W. et al. Histone deacetylases 1 and 2 form a developmental switch that
controls excitatory synapse maturation and function. J. Neurosci. 29, 8288–8297
(2009).
9. Cruz, J. C., Tseng, H. C., Goldman, J. A., Shih, H. & Tsai, L. H. Aberrant Cdk5 activation
by p25 triggers pathological events leading to neurodegeneration and
neurofibrillary tangles. Neuron 40, 471–483 (2003).
10. Cruz, J. C. et al. p25/cyclin-dependent kinase 5 induces production and
intraneuronal accumulation of amyloid beta in vivo. J. Neurosci. 26, 10536–10541
(2006).
11. Cruz, J. C. & Tsai, L. H. A Jekyll and Hyde kinase: roles for Cdk5 in brain
development and disease. Curr. Opin. Neurobiol. 14, 390–394 (2004).
12. Patrick, G. N. et al. Conversion of p35 to p25 deregulates Cdk5 activity and
promotes neurodegeneration. Nature 402, 615–622 (1999).
13. Fischer, A., Sananbenesi, F., Pang, P. T., Lu, B. & Tsai, L. H. Opposing roles of
transient and prolonged expression of p25 in synaptic plasticity and
hippocampus-dependent memory. Neuron 48, 825–838 (2005).
14. Frankland, P. W. & Bontempi, B. The organization of recent and remote memories.
Nature Rev. Neurosci. 6, 119–130 (2005).
15. McQuown, S. C. et al. HDAC3 is a critical negative regulator of long-term memory
formation. J. Neurosci. 31, 764–774 (2011).
16. Oakley, H. et al. Intraneuronal beta-amyloid aggregates, neurodegeneration, and
neuron loss in transgenic mice with five familial Alzheimer’s disease mutations:
potential factors in amyloid plaque formation. J. Neurosci. 26, 10129–10140
(2006).
17. Kimura, R., Devi, L. & Ohno, M. Partial reduction of BACE1 improves synaptic
plasticity, recent and remote memories in Alzheimer’s disease transgenic mice.
J. Neurochem. 113, 248–261 (2010).
18. Lauterborn, J. C. et al. Differential effects of protein synthesis inhibition on the
activity-dependent expression of BDNF transcripts: evidence for immediate-early
gene responses from specific promoters. J. Neurosci. 16, 7428–7436 (1996).
19. Sun, J. M., Chen, H. Y. & Davie, J. R. Differential distribution of unmodified and
phosphorylated histone deacetylase 2 in chromatin. J. Biol. Chem. 282,
33227–33236 (2007).
20. Fischer, A., Sananbenesi, F., Wang, X. Y., Dobbin, M. & Tsai, L. H. Recovery of
learning and memory is associated with chromatin remodelling. Nature 447,
178–182 (2007).
21. Markesbery, W. R. & Carney, J. M. Oxidative alterations in Alzheimer’s disease. Brain
Pathol. 9, 133–146 (1999).
22. Yankner, B. A. Mechanisms of neuronal degeneration in Alzheimer’s disease.
Neuron 16, 921–932 (1996).
23. Sandelin, A., Alkema, W., Engstrom, P., Wasserman, W. W. & Lenhard, B. JASPAR: an
open-access database for eukaryotic transcription factor binding profiles. Nucleic
Acids Res. 32, D91–D94 (2004).
24. Chrousos, G. P. & Kino, T. Glucocorticoid signaling in the cell. Expanding clinical
implications to complex human behavioral and somatic disorders. Ann. NY Acad.
Sci. 1179, 153–166 (2009).
25. Kino, T. et al. Cyclin-dependent kinase 5 differentially regulates the transcriptional
activity of the glucocorticoid receptor through phosphorylation: clinical
implications for the nervous system response to glucocorticoids and stress. Mol.
Endocrinol. 21, 1552–1568 (2007).
26. Guan, J. S. et al. Cdk5 is required for memory function and hippocampal plasticity
via the cAMP signaling pathway. PLoS ONE 6, e25735 (2011).
27. Citron, M. Alzheimer’s disease: strategies for disease modification. Nature Rev.
Drug Discov. 9, 387–398 (2010).
28. Braak, H. & Braak, E. Diagnostic criteria for neuropathologic assessment of
Alzheimer’s disease. Neurobiol. Aging 18, S85–S88 (1997).
29. Grober, E. et al. Memory and mental status correlates of modified Braak staging.
Neurobiol. Aging 20, 573–579 (1999).
30. Palop, J. J. & Mucke, L. Amyloid-beta-induced neuronal dysfunction in Alzheimer’s
disease: from synapses toward neural networks. Nature Neurosci. 13, 812–818
(2009).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements We thank A. Mungenast, S. Jemielity, R. Madabushi, F. Calderon de
Anda and M. Horn for reading the manuscript, A.M. for manuscript editing, M. Eichler for
mouse colony maintenance, K. Fitch for sectioning the human brain samples and M.H. for
quantification of Fig. 2a. This work was partly supported by the Stanley Medical Research
Institution (to S.J.H. and L.-H.T.), National Institutes of Health/National Institute on Drug
Abuse (RO1DA028301, to S.J.H.) and National Institutes of Health/National Institute of
Neurological Disorders and Stroke (RO1NS078839, to L.-H.T.). J.G. was supported by a
Bard Richmond fellowship and by the Swiss National Science Foundation, W.Y.W. by the
Simons Foundation and M.K. by the Theodor und Ida Herzog-Egli foundation. L.-H.T. is an
investigator of the Howard Hughes Medical Institute.
Author Contributions This study was designed by J.G. and L.-H.T., and directed and
coordinated by L.-H.T. J.G. planned and performed the in vitro, CK-p25 and Cdk5cKO
mouse and human in vivo biochemical characterization, and all behavioural
experiments. D.R. planned and contributed to the in vitro and CK-p25 in vivo
experiments, generated the GR526 and the shRNA constructs, and contributed to the
stereotaxic injections. J.S.G. initiated and contributed to the CK-p25 biochemical
characterization, and performed the 5XFAD and HDAC22/2 experiments. W.Y.W.
generated the luciferase constructs. J.S. performed the electrophysiological
experiments. K.M.H., T.J.F.N., D.F. and S.J.H. characterized the shRNA constructs. M.K.
contributed to the quantitative RT–PCR experiments and performed the quantification
of the human data. S.C.S. performed the site-directed mutagenesis. A.S. contributed to
the immunohistochemistry and the quantitative reverse transcription/quantitative
PCR experiments. N.J. contributed to the behavioural and quantitative reverse
transcription/quantitative PCR experiments. P.F.K. and I.D. provided the human
samples and contributed to the optimization of their staining. The manuscript was
written by J.G. and L.-H.T. and commented on by all authors.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Readers are welcome to comment on the online version of this article at
www.nature.com/nature. Correspondence and requests for materials should be
addressed to L.-H.T. ([email protected]).
2 2 6 | N AT U R E | VO L 4 8 3 | 8 M A R C H 2 0 1 2
©2012 Macmillan Publishers Limited. All rights reserved
LETTER RESEARCH
METHODS
Human material. Human material was obtained from the Massachusetts Alzheimer
Disease Research Center at Massachusetts General Hospital (protocol number 2004P-001613/4) and from the Boston Medical Center (H-24454) with informed consent
by all donors. The cases analysed in this study are strictly defined according to the
Braak and Braak staging28, with control cases being brains completely devoid of tau
pathologies (Braak and Braak stage 0)38. Hippocampal blocks (at the level of the
lateral geniculate nucleus) were fixed in paraformaldehyde, paraffin-embedded and
sectioned at 10 mm thickness. After antigen retrieval (Biogenex), brain slices were
rehydrated using xylene/ethanol, blocked with 5% milk serum in TBS (pH 7.4) at
room temperature and incubated at 4 uC overnight in TBS containing 5% milkserum in the following antibodies: amyloid-b, tau (Dako), HDAC1, HDAC2
(Abcam) or HDAC3 (Santa Cruz). Next, they were washed in PBS (pH 7.4),
incubated in HRP-conjugated secondary antibodies (Biogenex) and visualized
with DAB (Biogenex). Slides were counterstained with haematoxylin and dehydrated with ethanol/xylene. Images were quantified by an experimenter blind to
Braak and Braak stages using ImageJ 1.42q.
Animal models. All mouse work was approved by the Committee for Animal
Care of the Division of Comparative Medicine at the Massachusetts Institute of
Technology. Adult (3–6 months old) male double transgenic CK-p25 mice9,10,13,
6-month-old male transgenic 5XFAD mice16, and 3- to 4-month-old Cdk5cKO
mice26 and their respective control littermates were used for the experiments
unless otherwise noted; for CK-p25 mice, all behavioural experiments took place
between 6 and 8 weeks of p25 induction, the time when cognitive deficits are first
visible13. Behavioural experiments were conducted blindly and essentially as
described13. Open field behaviour was monitored using the VersaMax system
(Accuscan) for 20 min. For fear conditioning, mice were put in the conditioning
chamber (TSE systems) for 3 min, after which they received a one-time 2 s
footshock (0.8 mA). Animals were then left in the box for another 30 s. Twentyfour hours later, the mice were put into the same box and their freezing behaviour
was scored during 3 min. For the water maze that took place in a round tank (1.2 m
in diameter) filled with white opaque water, mice were first habituated to the task,
with the platform being visible for two trials. During habitation and the acquisition
phase, mice were allowed to swim for 60 s or until they reached the platform
(monitored by HVS Image). Animals that did not reach the platform after 60 s
were gently guided towards it; all animals were allowed to remain on the platform
for 15 s. For testing, mice were put back into the water without the platform 24 h
after that last training session, from a starting position different of the last starting
position during the acquisition phase, and their time spent in each quadrant was
recorded (HVS Image).
Electrophysiology. To record field excitatory postsynaptic potentials, transverse
hippocampal slices were prepared from CON, scr, CK-p25, scr and CK-p25,
shHDAC2 mice. In brief, the brain was rapidly removed and transferred to icecold, oxygenated (95% O2 and 5% CO2) cutting solution containing 211 mM
sucrose, 3.3 mM KCl, 1.3 mM NaH2PO4, 0.5 mM CaCl2, 10 mM MgCl2, 26 mM
NaHCO3 and 11 mM glucose. Hippocampal slices were cut with a VT1000S
vibratome (Leica) and transferred for recovery to a holding chamber containing
oxygenated artificial cerebrospinal fluid consisting of 124 mM NaCl, 3.3 mM KCl,
1.3 mM NaH2PO4, 2.5 mM CaCl2, 1.5 mM MgCl2, 26 mM NaHCO3 and 11 mM
glucose at 28–30 uC for at least 1 h before recording. CA1 field potentials evoked by
Schaffer collateral stimulation were measured. After recording of a stable baseline
(at least 20 min), long-term potentiation was induced by four episodes of theta
burst stimulation (TBS) with 10 s intervals. TBS consisted of ten bursts (each with
four pulses at 100 Hz) of stimuli delivered every 200 ms. Recordings were performed using an AM-1800 microelectrode amplifier (A-M systems) and a
Digidata 1440A analogue to digital converter (Axon Instruments). All data were
digitized and analysed by the use of pClamp10 software (Axon Instruments). Basal
synaptic input/output relationship was obtained by plotting field excitatory postsynaptic potential slopes against stimulation intensities. All experiments were
performed by an experimenter blind to treatment groups.
In vitro studies. Primary mouse hippocampal neuronal cultures (days in vitro 14–
17) were treated with sense and antisense amyloid-b oligomers (1 mM, Bachem)
for 24 h or H2O2 (50 mM, Mallinckrodt Chemicals, removed after 5 min, and
assessed 8 h later) unless otherwise noted. For ChIP experiments, cortical cultures
(days in vitro 10–14) were used.
Dual luciferase assays were conducted on CAD cells31 (ATCC) that were transfected using lipofectamine (Invitrogen) with 0.5 mg of the proximal promoter
region of Hdac2 containing the GRE consensus sequence (CAAGAAGAAAGTG
GCTAC) or with the proximal promoter region without the GRE sequence subcloned into the pGL3 reporter vector (Promega) according to the manufacturer’s
instructions. Cells were cotransfected with 0.05 mg of the constitutively active form
of GR32, GR526 or GRS211A, and treated with 1 mM amyloid-b1–42 oligomers or
with 50 mM H2O2.
For site-directed mutagenesis, the complementary DNA for the full-length
human glucocorticoid receptor (Addgene) was used and serine 211 was replaced
by alanine using the QuikChange Lightning Kit (Agilent Technologies) as per the
manufacturer’s instructions. All constructs were verified by sequencing
(Genewiz). The following mutagenesis primers were used: 59-GTAAAGAGA
CGAATGAGGCTCCTTGGAGATCAGACC-39 (forward); 59-GGTCTGATCT
CCAAGGAGCCTCATTCGTCTCTTTAC-39 (reverse).
Immunohistochemistry and immunocytochemistry. Immunocytochemistry
and mouse immunohistochemistry were performed as described7. Immunocytochemistry and immunohistochemistry on different experimental conditions
were performed with the same antibody solution at the same time to assure
identical staining conditions. A negative (that is, no antibody) control was
included simultaneously. In brief, for immunohistochemistry, mice were perfused
with 10% paraformaldehyde under deep anaesthesia (ketamine, xylazine) and
their brains sectioned at 0.35 mm thickness using a vibratome (Leica). For immunocytochemistry, cells were fixed using 4% paraformaldehyde. Slices/cells were
permeabilized with 0.1% Triton X-100, blocked and incubated overnight with
0.1% Triton X-100/10% fetal bovine serum in PBS containing primary antibodies:
HDAC1, HDAC2 (Abcam), HDAC3 (Santa Cruz), phospho-GR1 (S211) (Cell
Signaling), GR (Abcam) or GFP (Aves Labs). Primary antibodies were visualized
with Alexa-Fluor 488, Cy3 and Cy5 antibodies (Molecular Probes), neuronal
nuclei with Hoechst 33342 (Invitrogen). Note that for staining of shRNA-injected
animals, mCherry was visualized without staining, and only Cy2 and Cy5 secondary
antibodies were used. Images were acquired using a confocal microscope (LSM 510,
Zeiss) at identical settings at the highest intensity for each of the conditions. Images
were quantified using ImageJ 1.42q by an experimenter blind to treatment groups,
whenever possible. For each experimental condition, 20–40 representative cells per
section were analysed, and the mean signal intensity was measured. To assess the
specificity of the HDAC2 immunostaining, adult HDAC22/2 mice were used as
previously described7. For a specificity control of the pGR1 antibody, calf intestine
phosphatase (CIP) (New England Biolabs) treatment on immunocytochemistry
slides was performed as described previously33. Calf intestine phosphatase or
H2O were applied for 40 min in Buffer 3 (New England Biolabs).
Molecular analyses. Western blots, co-immunoprecipitation, ChIP and gene
expression analyses were performed as described elsewhere34 with the following
modifications, and expressed as fold change of the respective control conditions.
For western blots, proteins were extracted using 13 RIPA buffer containing
proteinase (complete, Roche) and phosphatase inhibitors (1 mM b-glycerophosphate,
10 mM NaF, 0.1 mM Na3VO4), transferred onto PVDF membranes (Biorad) and
stripped using stripping buffer (Thermo Scientific). The following primary antibodies
were used: acetyl-K (Cell Signaling), a- and b-tubulin, b-actin (Sigma), Cdk5
(Santa Cruz), HDAC1, HDAC2 (Abcam), HDAC3, phospho-GR1 (S211) (Cell
Signaling), GR1 (Abcam), p53 (Cell Signaling) or tau (Invitrogen). Secondary
antibodies were horseradish peroxidase-linked (GE Healthcare). Signal intensities
were quantified using ImageJ 1.42q and normalized to values of b-actin, a- or
b-tubulin. Phospho-GR1 was first normalized to GR1. Three to eight animals were
used per condition. Cytoplasmic and nuclear fractionation was performed as
described elsewhere34.
For co-immunoprecipitation, hippocampal lysates were incubated with HDAC2
(Abcam) or IgG (Sigma) and the immunoprecipitated extracts probed for SIN3A,
MTA2 (Abcam), CoRest (Millipore), LSD1 (Cell Signaling) or HDAC2. Signal
intensities were quantified using ImageJ 1.42q, and normalized to input. Three or
four animals were used per condition.
For ChIP, tissue samples were homogenized in cell lysis buffer containing
proteinase (complete, Roche) and phosphatase inhibitors (1 mM b-glycerophosphate,
10 mM NaF, 0.1 mM Na3VO4) and chromatin was sonicated using a Branson
Digital Sonifier with 10 rounds of 15 s at 25% power per sample on ice to 200–
400 base pairs in length. For ChIP of primary cortical cultures, approximately
1 3 106 cells were crosslinked in 37% formaldehyde (Sigma), quenched with
203 glycine (Sigma), washed with PBS and cell lysis buffer containing both phosphatase and proteinase inhibitors (Roche) and sonicated in nuclear lysis buffer
using a Vibra Cell Sonifier with 3 3 3 pulses of 5 s at 35% power (50% duty) on ice.
Sheared chromatin was immunoprecipitated with antibodies against HDAC1,
HDAC2 (Abcam), HDAC3 (Santa Cruz), acetyl H2BK5, acetyl H3K14
(Abcam), acetyl H4K5, acetyl H4K12 (Millipore), phospho-GR1 (S211) (Cell
Signaling) or phospho-RNA Pol II (Abcam). DNA was extracted by phenol/
chloroform/isoamyl alcohol (American Bioanalytical) and subjected to quantitative
PCR (Bio-Rad Thermal Cycler) using primers specific to the promoter or coding
regions of the genes assayed (see Supplementary Tables 2 and 3 for primer
sequences). The fluorescent signal of the amplified DNA (SYBR green, Bio-Rad)
was normalized to input. Four to eight samples were used per condition.
For gene expression analysis, mRNA was extracted (Qiagen), reversetranscribed (Invitrogen) and quantitatively amplified on a thermal cycler
©2012 Macmillan Publishers Limited. All rights reserved
RESEARCH LETTER
(Bio-Rad) using SYBR green (Bio-Rad) and gene-specific primers (see
Supplementary Table 4). The comparative Ct method35 was used to examine
differences in gene expression. Values were normalized to expression levels of
Gapdh. Four to eight samples were used per condition.
Experimental manipulations. For validation of the knockdown efficacy of
HDAC2 RNA interference in neurons, short hairpins targeting the open reading
frame of mouse Hdac2 mRNA from the Broad Institute’s RNAi consortium
shRNA library (www.broadinstitute.org/rnai/trc/lib) were packaged into lentiviral
vectors36 and used to infect dissociated primary mouse embryonic (gestation
day 18) cortical cultures, prepared as described7. Primary cultures were transduced
after 4 days in vitro. Cells were collected ten days after transduction, and protein
levels were measured by western blotting. The target sequences for the two effective shRNAs were CCCAATGAGTTGCCATATAAT (HDAC2 shRNA 2-1,
TRCN0000039395) and CGAGCATCAGACAAACGGATA (HDAC2 shRNA
2-4, TRCN0000039397).
After validation, scramble shRNA37 or HDAC2-shRNA constructs were subcloned under the CaMKII-U6 promoter into the plasmid AAV entry vector fused
to mCherry, tested again in mouse primary hippocampal cultures (day in vitro 7),
and high titre (1 3 1012 to 4 3 1012 viral particles) adeno-associated viruses
(serotype 2.5) were produced at the University of North Carolina Vector Core facility.
One microlitre of shRNA-containing adeno-associated viruses was stereotaxically
injected into hippocampal area CA1 (anterior–posterior position 22.0 mm, medial–
lateral position 61.6 mm, dorso-ventral 21.5 mm from Bregma) of both
hemispheres at 0.1 ml min21. Injection needles were left in place 5 min after injection
to assure even distribution of the virus. Injections were performed 4 weeks before
behavioural testing. All infusion surgeries were performed under aseptic conditions
and anaesthesia (ketamine/xylazine) in accordance with the Massachusetts Institute
of Technology’s Division of Comparative Medicine guidelines.
Statistics. Statistical analyses were performed using GraphPad Prism 5. One-way
analyses of variance followed by Tukey post-hoc tests, or one-tailed Student’s t-tests
were used unless otherwise indicated. All data are represented as mean 6 s.e.m.
Statistical significance was set at P # 0.05.
31. Qi, Y., Wang, J. K., McMillian, M. & Chikaraishi, D. M. Characterization of a CNS cell
line, CAD, in which morphological differentiation is initiated by serum deprivation.
J. Neurosci. 17, 1217–1225 (1997).
32. Li, L. & Lindquist, S. Creating a protein-based element of inheritance. Science 287,
661–664 (2000).
33. Xie, Z., Sanada, K., Samuels, B. A., Shih, H. & Tsai, L. H. Serine 732 phosphorylation
of FAK by Cdk5 is important for microtubule organization, nuclear movement, and
neuronal migration. Cell 114, 469–482 (2003).
34. Koshibu, K. et al. Protein phosphatase 1 regulates the histone code for long-term
memory. J. Neurosci. 29, 13079–13089 (2009).
35. Livak, K. J. & Schmittgen, T. D. Analysis of relative gene expression data using realtime quantitative PCR and the 22DDCT. Methods 25, 402–408 (2001).
36. Moffat, J. et al. A lentiviral RNAi library for human and mouse genes applied to an
arrayed viral high-content screen. Cell 124, 1283–1298 (2006).
37. Sarbassov, D. D., Guertin, D. A., Ali, S. M. & Sabatini, D. M. Phosphorylation and
regulation of Akt/PKB by the rictor-mTOR complex. Science 307, 1098–1101
(2005).
38. Price, J. L. et al. Neuropathology of nondemented aging: presumptive evidence for
preclinical Alzheimer disease. Neurobiol. Aging 30, 1026–1036 (2009).
©2012 Macmillan Publishers Limited. All rights reserved
SUPPLEMENTARY INFORMATION
doi:10.1038/nature10849
a
neurotoxic insults
acute synaptic dysfunctions
chronic cognitive deficits
epigenetic blockade of gene transcription
b
e.g. Bdnf IV,
synaptophysin
neurotoxic insults
Aβ
H2O2
Cdk5/p25
P
P
GR1
GR
GR1
GR
GRE
HDAC2
Hdac2
other
Supplementary Figure 1. Model of the potential consequences and causes of elevated HDAC2
levels in the neurodegenerating brain. a, Neurotoxic insults are known to acutely constrain synaptic
functions, which in turn, can lead to more chronic cognitive dysfunctions. Here, we propose that neurotoxic insults can directly lead to cognitive dysfunctions by epigenetically silencing neuroplasticity genes
via the mechanisms described below. b, Upon neurotoxic insults such as extracellular Aβ-fibrils, H2O2,
intracellular accumulation of p25 and/or other stimuli, GR1 becomes phosphorylated on S211, a mark
for its activation. Phoshorylated GR1 binds to its responsive element (GR responsive element, GRE) in
the proximal Hdac2 promoter region in the neurodegenerating mouse brain, and stimulates the expression of Hdac2. In the presence of Aβ-fibrils and H2O2, this stimulation is further potentiated, but abolished when S211 is mutated to A211 (not shown). Elevated HDAC2 protein levels bind to the promoter
region of learning, memory and synaptic plasticity-related genes such as BdnfIV, synaptophysin and
others. There, HDAC2 binding co-occurs with reduced histone acetylation, and thus more compacted
chromatin, preventing RNA polymerase from binding (not shown), supposedly the driving force behind
the reduced gene expression observed. Such decreased expression of neuroplasticity genes is correlated with reduced synaptic plasticity and poor memory performance in mice with neurodegeneration.
W W W. N A T U R E . C O M / N A T U R E | 1
RESEARCH SUPPLEMENTARY INFORMATION
a
HDAC2 antibody specificity
HDAC2
DAPI
CON
HDAC2 -/-
b
pGR1 antibody specificity
pGR1
DAPI
Mock
CIP
Supplementary Figure 2. Specificity of HDAC2 and pGR1 antibody signals. a, Immunohistochemical images showing HDAC2 detection in the hippocampus of wild-type mice, but not in
HDAC2-/- mice. Scale bar upper panel, 80µm, lower panel, 100µm. b, Immunocytochemical
images showing pGR1 immunoreactivity in mock-treated, but not in calf intestinal phosphatase
(CIP)-treated primary hippocampal neurons (DIV14). Scale bar, 10µm.
2 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
a
b
Dentate Gyrus
HDAC2
DAPI
CA3
HDAC2
DAPI
CON
CK-p25
c
Amygdala
HDAC2
DAPI
CON
CK-p25
Supplementary Figure 3. The elevation of HDAC2 is restricted to area CA1 of the hippocampus,
and further not observed in the amygdala. a, b, Representative immunohistochemical images of
HDAC2 levels in the (a) dentate gyrus and (b) hippocampal area CA3 of CK-p25 mice compared to
control littermates. c, Representative immunohistochemical images of HDAC2 levels in the amygdala
of CK-p25 mice compared to control littermates; scale bar, 100µm (a, b), 200µm (c).
W W W. N A T U R E . C O M / N A T U R E | 3
RESEARCH SUPPLEMENTARY INFORMATION
CON
CK-p25
a
HDAC2
DAPI
b
HDAC2 protein level
(fold change of control)
GFP
CON
CK-p25
2.0
1.5
**
1.0
0.5
c
GFP
HDAC1
DAPI
CON
CK-p25
d
2.0
HDAC1 protein level
(fold change of control)
0.0
1.5
1.0
0.5
0.0
GFP
CON
CK-p25
HDAC3
DAPI
f
HDAC3 protein level
(fold change of control)
e
2.0
1.5
1.0
0.5
0.0
Supplementary Figure 4. HDAC2, but not HDAC1 and HDAC3, is increased in the neurodegenerating cortex of CK-p25 mice. a, c, e, Representative immunohistochemical images depicting (a)
HDAC2, (c) HDAC1, and (e) HDAC3 levels in the prefrontal cortex of CK-p25 mice and control littermates (n=3-6 sections from 3 mice each). b, d, f, Quantification of (a, c, e). Scale bar, 20µm. **p≤
0.01; values are mean ± s.e.m.
4 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
a
HDAC2
DAPI
MERGE
b
CON
5XFAD
2.0
HDAC2 protein level
(fold change of control)
CON
5XFAD
*
1.5
1.0
0.5
0.0
c
HDAC1
DAPI
MERGE
d
2.0
HDAC1 protein level
(fold change of control)
CON
5XFAD
1.5
1.0
0.5
0.0
e
HDAC2
DAPI
MERGE
f
2.0
HDAC2 protein level
(fold change of control)
CON
5XFAD
*
1.5
1.0
0.5
0.0
g
HDAC1
DAPI
MERGE
h
5XFAD
HDAC1 protein level
(fold change of control)
2.0
CON
1.5
1.0
0.5
0.0
Supplementary Figure 5. HDAC2, but not HDAC1 levels are increased in the neurodegenerating
forebrain of 5XFAD mice. a, c, Representative immunohistochemical images depicting (a) HDAC2 and
(c) HDAC1 in the hippocampus of 6-month-old 5XFAD mice compared to control littermates (n=3-6
sections from 3-4 mice each). b, d, Quantitative assessment of (a) and (c). e, g, Representative immunohistochemical images depicting (e) HDAC2 and (g) HDAC1 in the prefrontal cortex of 6-month-old 5XFAD
mice compared to control littermates (n=3-4 sections from 3 mice each). f, h, Quantitative assessment of
(e) and (g). Scale bar, 50µm. *p≤0.05; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 5
RESEARCH SUPPLEMENTARY INFORMATION
a
CON
CK-p25
HDAC1 ChIP
(fold change of control)
HDAC1 binding
HDAC3 ChIP
(fold change of control)
b
c
Learning and Memory
5.0
Synaptic Plasticity
5.0
Housekeeping
5.0
2.5
2.5
2.5
0.0
0.0
0.0
HDAC3 binding
5.0
5.0
2.5
2.5
0.0
0.0
*
5.0
2.5
*
0.0
AcH2BK5 ChIP
(fold change of control)
H2BK5 acetylation
2.5
2.5
2.5
2.0
2.0
2.0
1.5
1.5
1.5
1.0
0.0
AcH3K14 ChIP
(fold change of control)
d
e
*
*
**
*
0.5
*
1.0
0.0
*
*
**
*
0.5
1.0
**
0.5
0.0
H3K14 acetylation
2.5
2.5
2.5
2.0
2.0
2.0
1.5
1.5
1.5
*
1.0
**
0.5
1.0
*
*
0.5
1.0
*
*
*
0.5
0.0
0.0
2.5
2.5
2.5
2.0
2.0
2.0
1.5
1.5
1.5
0.0
1.0
0.5
*
**
*
0.5
ul
in
p
1
Sy
t
Sy
2A
fl
2B
R
N
R
N
N
1
R
lu
G
R
er
2
0.0
1
r1
om
H
Eg
5
dk
IV
nf
C
II
Bd
nf
Bd
Bd
nf
I
0.0
in
0.0
1.0
*
lo
b
**
βTu
b
*
βG
*
*
Ac
tin
*
β-
*
lu
0.5
G
1.0
Ar
c
AcH4K5 ChIP
(fold change of control)
H4K5 acetylation
Supplementary Figure 6. HDAC1 and HDAC3 binding to neuroplasticity genes is overall not increased
in the hippocampus of CK-p25 mice, but the reduction of histone acetylation in the promoter region
of neuroplasticity genes is not restricted to H4K12. a, b, Quantitative PCR results of (a) HDAC1- and (b)
HDAC3-immunoprecipitated chromatin in the hippocampus of CK-p25 mice versus control littermates (n=4-6
mice each). c-e, Quantitative PCR results of (c) AcH2BK5-, (d) AcH3K14- and (e) AcH4K5-immunoprecipitated chromatin of the hippocampus of CK-p25 mice versus control littermates (n=3-8 animals each). *p≤0.05;
**p≤0.01; ***p≤0.001; values are mean ± s.e.m.
6 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
4.0
tin
Ac
β-
p
t1
Sy
Sy
N
R
2B
N
fl
N
R
2A
lu
G
lu
G
Eg
C
dk
Ar
R
2
0.0
R
1
0.0
r1
H
om
er
1
0.0
5
2.0
nf
2.0
c
2.0
in
*
*
ul
*
n
*
*
bi
*
*
4.0
βTu
b
*
*
CON
CK-p25
6.0
lo
4.0
*
6.0
βG
*
6.0
Bd
HDAC2 ChIP
(fold change of control)
HDAC2 binding to coding sequences
Supplementary Figure 7. HDAC2 binding to the coding region of neuroplasticity genes is
increased in the hippocampus of CK-p25 mice. Quantitative PCR results of HDAC2immunoprecipitated chromatin in the coding sequence of neuroplasticity genes in the hippocampus of
CK-p25 mice versus control littermates (n=4-5 animals each). *p≤0.05; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 7
RESEARCH SUPPLEMENTARY INFORMATION
a
CON
b
CK-p25
CON
CK-p25
Total Tau
Ac-p53
Total p53
Ac-tubulin
Protein level
(fold change of control)
Ac-Tau
Total tubulin
c
CON
CK-p25
2.0
2.0
1.5
1.5
1.5
1.0
1.0
1.0
0.5
0.5
0.5
0.0
0.0
0.0
Ac-Tau/Tau
d
Cytoplasm
2.0
Ac-p53/p53
Ac-tubulin/tubulin
Nucleus
CON
CK-p25
Ac-K
Ac-K
β-Actin
β-Actin
Supplementary Figure 8. The elevation of HDAC2 does not induce general acetylation changes
on proteins other than histones. a, Representative images of general acetyl western blot analysis of
immunoprecipitated protein complexes with tau, p53 and tubulin (n=3-6 mice each); note that the
increase in tau and p53 protein levels in the CK-p25 mice has been previously described by our lab,
and that p53 was found to be hyperacetylated on lysine 382 (Kim et al., 2007). b, Quantification of (a).
c, d, Representative images of western blot analysis of protein acetylation in cytoplasmic (c) and
nuclear (d) hippocampal extracts of CK-p25 mice and control littermates (n=3 mice each); values are
mean ± s.e.m.
8 | W W W. N A T U R E . C O M / N A T U R E
b
CoREST
WB: LSD1
Protein level
(fold change of control)
WB: CoREST
1.5
*
1.0
0.5
0.0
LSD1
Protein level
(fold change of control)
AC
2
In
pu
t
CON
CK-p25
D
H
Ig
G
D
H
IP:
Ig
G
a
AC
2
In
pu
t
SUPPLEMENTARY INFORMATION RESEARCH
4
***
3
2
1
0
WB: SIN3A
CON
CK-p25
2.5
2.0
1.5
1.0
0.5
0.0
**
MTA2
Protein level
(fold change of control)
WB: HDAC2
Protein level
(fold change of control)
SIN3A
WB: MTA2
4
*
3
2
1
0
Supplementary Figure 9. HDAC2 is more abundantly bound to co-repressor complexes in the hippocampus of CK-p25 mice. a, Representative images of western blot analysis of protein levels of CoREST
and LSD1, members of the CoREST complex, of SIN3A, a member of the Sin3 complex, and of MTA2, a
member of the NuRD complex, following immunoprecipitation of hippocampal extracts with HDAC2 (n=3-4
animals each). b, Quantification of (a). Protein levels were normalized to input, *p≤0.05; **p≤0.01; ***p≤
0.001; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 9
RESEARCH SUPPLEMENTARY INFORMATION
a
CONTROL
shRNA 2-1
shRNA 2-2
shRNA 2-3
shRNA 2-4
HDAC2
α-tubulin
ITR
c
CaMKII
mCherry
U6
d
shRNA shRNA
CON
2-1
2-4
HDAC2
β-tubulin
HDAC2 protein level
(fold change of control)
b
e
shRNA
CON
shRNA 2-1
shRNA 2-4
***
1.5
1.0
0.5
0.0
f
HDAC2
DAPI
shRNA
2-4
untransfected
shRNA
1.0
HDAC2 protein level
(fold change of control)
mCherry
shRNA
2-1
***
0.5
2-1
0.0
**
1.0
0.5
2-4
0.0
g
ITR
p25 induction
0
h
behavioral testing,
sample collection
stereotaxic injection
2
CON, scr
6
CK-p25, scr
weeks
CK-p25, shHDAC2
mCherry
Supplementary Figure 10. Strategy to reduce HDAC2 expression by RNA interference. a, Validation of
the target specificity of different HDAC2 shRNA clones as determined by lentiviral transduction in primary
cortical neurons (DIV7) by western blot analysis. A significant downregulation was observed for constructs
shRNA2-1 and shRNA2-4. These constructs and the scramble shRNA were then subcloned into the AAV
vector under (b). b, Schematic of the construct used for AAV production; “shRNA” signifies either control
scramble shRNA, or shRNA constructs targeting HDAC2. c, Representative images of western blot analysis
of HDAC2 in primary hippocampal neurons (DIV14) following transfection with the AAV-shRNA constructs. d,
Quantification of (c). e, Representative immunocytochemical images depicting HDAC2 levels in primary
hippocampal neurons (DIV14) following transfection with the AAV-shRNA constructs; scale bar, 10µm. f,
Quantification of (e). g, Schematic of the experimental timeline. Mice were 3-4 months of age when p25 was
induced. Note that shRNA2-1 was used as shHDAC2 for all in vivo experiments. h, Representative pictures
of mCherry showing comparable infection rates in hippocampal area CA1 of control mice injected with
scrambled shRNA (CON, scr), CK-p25 mice injected with scrambled shRNA (CK-p25, scr) and CK-p25 mice
injected with the shRNA against HDAC2 (CK-p25, shHDAC2); scale bar, 100µm. **p≤0.01; ***p≤0.001; values
are mean ± s.e.m.
1 0 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
a
GFP
HDAC1
DAPI
MERGE
b
CON, scr
CK-p25, scr
CK-p25, shHDAC2
CON, scr
HDAC1 level (mean grey value)
150
CK-p25,
scr
CK-p25,
shHDAC2
100
50
0
c
GFP
HDAC3
DAPI
MERGE
d
CON, scr
CK-p25,
scr
CK-p25,
shHDAC2
HDAC3 level (mean grey value)
150
100
50
0
Supplementary Figure 11. AAV-shRNA mediated reduction of HDAC2 does not alter levels of HDAC1 and HDAC3.
a, c, Representative immunohistochemical images depicting hippocampal (a) HDAC1 and (c) HDAC3 levels in CON, scr,
CK-p25, scr, and CK-p25, shHDAC2 animals (n=4-5 sections from 4 mice each). b, d, Quantitative assessment of (a, c);
scale bar, 20µm; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 1 1
RESEARCH SUPPLEMENTARY INFORMATION
Neuron counts
CON, scr
NeuN
DAPI
CON, scr
CK-p25, scr
CK-p25, shHDAC2
b
CK-p25, scr
CK-p25, shHDAC2
Number of
NeuN-positive cells
a
2000
**
1500
1000
500
0
Supplementary Figure 12. Reducing HDAC2 does not alter the course of neurodegeneration.
a, Representative immunohistochemical images depicting NeuN staining in hippocampal area CA1
of CON, scr, CK-p25, scr, and CK-p25, shHDAC2 animals (n=4-5 sections from 3-4 mice each). b,
Quantification of (a); scale bar, 80µm; **p≤0.01; values are mean ± s.e.m.
1 2 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
CON, scr
CK-p25, scr
CK-p25, shHDAC2
a
b
LTP
250
c
fEPSP slope (% baseline)
fEPSP slope (% baseline)
400
300
200
100
4 TBS
0
-20
-10
0
10
20
30
40
50
60
*
**
200
150
100
50
0
Input-Output
*
*
1000
fEPSP slope (mV/s)
** *
CON, scr vs CK-p25, scr
*
*
CK-p25, shHDAC2 vs CK-p25, scr
800
600
400
200
0
0
0.3
0.6
0.9
1.2
1.5
1.8
Stimulation intensity
Supplementary Figure 13. Synaptic plasticity is restored upon reducing HDAC2. a, Field excitatory postsynaptic potential (fEPSP) slopes in hippocampal area CA1 of CON, scr, CK-p25, scr and
CK-p25, shHDAC2 animals (n=7-8 slices from 4 mice each). b, Average slopes of fEPSP during the
last 10min of recording; sample traces above the bar chart represent fEPSPs at 1min before (gray)
and 1h after (black) theta-burst stimulation (TBS). c, Input/output relationship of baseline synaptic
transmission. *p≤0.05; **p≤0.01; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 1 3
RESEARCH SUPPLEMENTARY INFORMATION
CON, scr
CK-p25, scr
CK-p25, shHDAC2
Water Maze
b
30
Swim speed (m/s)
Distance travelled (m)
a
20
10
0
Open Field
**
200
**
600
400
200
0
0
400
300
200
100
0
Time spent in center (s)
f
e
Time spent rearing (s)
0.1
d
600
400
0.2
0.0
Time spent moving (s)
Distance travelled (cm)
c
0.3
400
300
200
100
0
Supplementary Figure 14. Overall locomotion and anxiety behavior is not affected by HDAC2
reduction. a, b, Distance travelled (a) and swim speed (b) were comparable between CON, scr,
CK-p25, scr and CK-p25, shHDAC2 in the water maze task. c, d, Distance travelled (c) and time spent
moving (d) were comparable between CK-p25, scr and CK-p25, shHDAC2 in an open field test. Note
that both groups of CK-p25 animals were hyperactive as compared to CON, scr animals. e, f, Time
spent rearing (e) and time spent in center (f) were comparable between CON, scr, CK-p25, scr, and
CK-p25, shHDAC2 mice during an open field test. **p≤0.01; values are mean ± s.e.m.
1 4 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
H2O
H2O2
pGR1
GR1
HDAC2
β-Actin
c
Protein level
(fold change of Aβ42-1)
3.0
2.5
*
b
Aβ42-1
Aβ1-42
*
2.0
Aβ42-1 Aβ1-42
1.5
pGR1
1.0
GR1
HDAC2
0.5
β-Actin
0.0
HDAC2
pGR1/GR
HDAC2
Aβ42-1
Aβ1-42
2.5
*
2.0
d
*
2.5
*
*
2.0
1.5
1.0
0.5
0.0
HDAC2
pGR1/GR
Aβ1-42
0
6
12
24
48 h
Aβ42-1
0
6
12
24
48 h
pGR1
GR1
1.5
HDAC2
1.0
0.5
β-Actin
0
3.0
Protein level
(fold change of Aβ42-1)
Protein levels (fold change H2O)
H2O
H2O2
Protein levels (fold change Aβ42-1)
a
10
20
pGR1
30
40
50
*
*
pGR1
2.5
2.0
Aβ42-1
Aβ1-42
1.5
1.0
GR1
HDAC2
β-Actin
0.5
0
e
pGR1 binding
(fold change of 0h)
8.0
10
20
30
40
pGR1 binding to Hdac2 promoter
6.0
*
*
4.0
Aβ1-42
2.0
0.0
50
0
10
20
30
40
50
time (h)
Supplementary Figure 15. Timecourse of pGR1 and HDAC2 induction following neurotoxic
insults. a, b, Representative images (left) of western blot analyses of pGR1, GR1, and HDAC2 levels
of primary hippocampal neurons (DIV14-17) (a) 8h after H2O2- or H2O-treatment and (b) 24h after exposure to Aβ1-42 or Aβ42-1 oligomers and quantification thereof (right); pGR1 levels were normalized to
total GR1 levels. c, d, Quantification (c) and representative images (d) of western blot analysis of the
development of pGR1 and HDAC2 induction in primary hippocampal neurons (DIV14-17) treated with A
β1-42 or Aβ42-1 oligomers. pGR1 was normalized to GR1. e, Timecourse of pGR1 binding to the GRE in
the Hdac2 promoter as revealed by ChIP analysis in primary cortical neurons (DIV10-14) treated with A
β1-42. All results are from at least three independent experiments. *p≤0.05; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 1 5
RESEARCH SUPPLEMENTARY INFORMATION
HDAC2
pGR1
***
***
90
80
pGR1 level (mean grey value)
HDAC2 level (mean grey value)
90
70
60
50
40
30
20
10
80
70
60
50
40
30
20
10
0
0
CON
CK-p25
CON
CK-p25
Supplementary Figure 16. HDAC2 and pGR1 protein levels are concomitantly increased in
CK-p25 mice. Scatter plot representation of HDAC2 and pGR1 immunohistochemical labeling of neurons in hippocampal area CA1 in CK-p25 versus control mice (n=50 neurons from comparable rostrocaudal positions from 3 mice each); ***p≤0.001.
1 6 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
luc
GRE
luc
***
**
Luciferase activity
(fold change of control)
7
**
6
5
4
3
2
1
0
GR526
-
+
-
+
Supplementary Figure 17. GR526 potentiates the transcriptional potential of the GRE in the
Hdac2 promoter. Luciferase activity of CAD cells (≥3 independent experiments) transfected with the
proximal promoter region of Hdac2 with (orange) or without (blue) GRE (schematic of constructs are
shown above the graph) and co-transfected with a constitutively active form of GR, GR526 (see
Methods). **p≤0.01; ***p≤0.001; values are mean ± s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 1 7
RESEARCH SUPPLEMENTARY INFORMATION
a
tau
b
Hippocampus
Entorhinal Cortex
c
β-amyloid
d
Hippocampus
Entorhinal Cortex
CON
CON
CON
CON
BB I-II
BB I-II
BB I-II
BB I-II
BB III-IV
BB III-IV
BB III-IV
BB III-IV
BB V-VI
BB V-VI
BB V-VI
BB V-VI
Supplementary Figure 18. Tau tangles and β-amyloid accumulate in the hippocampus and the
entorhinal cortex as Alzheimer’s disease progresses. a, b, Representative images of the cases analyzed in this study showing the accumulation of tau tangles (dark-brown spots) during the development
of the Braak and Braak (BB) stages of Alzheimer’s disease in (a) hippocampal area CA1 and (b) the
entorhinal cortex. c, d, Representative images of the cases analyzed in this study showing the accumulation of β-amyloid (dark-brown aggregations) during the development of the Braak and Braak (BB)
stages of the disease in (c) hippocampal area CA1 and (d) the entorhinal cortex; scale bar, 100µm.
1 8 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
a
b
HDAC2
c
HDAC1
CON
HDAC3
CON
CON
➚
➚
➚
BB I-II
BB I-II
BB I-II
➚
➚
➚
➚
BB III-IV
BB III-IV
BB III-IV
➚
➚
BB V-VI
BB V-VI
BB V-VI
➚➚
➚
➚
1.5
1.0
0.5
0.0
**
e
2.0
1.5
1.0
0.5
0.0
f
HDAC3 protein level
(fold change of control)
2.0
HDAC1 protein level
(fold change of control)
HDAC2 protein level
(fold change of control)
d
2.0
CON
BBI-II
BBIII-IV
BBV-VI
1.5
1.0
0.5
0.0
Supplementary Figure 19. HDAC2, but not HDAC1 or HDAC3 levels are increased in the entorhinal
cortex during the progression of Alzheimer’s disease. a-c, Representative immunohistochemical images
depicting (a) HDAC2, (b) HDAC1, and (c) HDAC3 levels in neuronal nuclei in the entorhinal cortex of patients
with Braak and Braak (BB) stages I/II (n=4), III/IV (n=7) and V/VI (n=7) compared to BB0 healthy control
brains (CON, n=7); insets show magnification of one neuron (indicated by an arrow, nucleus surrounded by
white dotted circles); scale bar, 100µm. d-f, Quantitative assessment of a-c, **p≤0.01; values are mean ±
s.e.m.
W W W. N A T U R E . C O M / N A T U R E | 1 9
RESEARCH SUPPLEMENTARY INFORMATION
Supplementary Table 1
Evidence for reduced expression of Alzheimer’s disease-related HDAC2 target genes
Gene
Evidence
Sample type
Reference
Arc
mRNA
Bdnf
mRNA
Human post-mortem AD brain
Amyloid-containing post-mortem
human brain regions
-amyloid treated primary cortical
neurons
Human post-mortem AD brain
Cdk5
mRNA
Human post-mortem AD brain
Egr1
mRNA
-amyloid treated human brain
pericytes
Amyloid-containing post-mortem
human brain regions
Amyloid-containing post-mortem
human brain regions
-amyloid treated primary cortical
neurons
Human post-mortem AD brain
Human post-mortem AD brain
protein
Homer1
mRNA
protein
GluR1
mRNA
protein
GluR2
mRNA
mRNA/protein
Nfl
mRNA
Human post-mortem AD brain
Galanin-negative neurons in
human post-mortem AD brain
Human post-mortem AD brain
NR2A
mRNA
Human post-mortem AD brain
NR2B
mRNA
Human post-mortem AD brain
Syp
mRNA
Human post-mortem AD brain
Syt1
mRNA
Human post-mortem AD brain
Ginsberg et al., 2002
Dickey et al., 2005
Wang et al., 2005
Philips et al., 1991;
Murray et al., 1994;
Colangelo et al., 2002;
Tan et al., 2010
Ginsberg et al., 2002;
Liang et al., 2008;
Tan et al., 2010
Rensink et al., 2002/4
Dickey et al., 2005
Dickey et al., 2005
Roselli et al., 2009
Ginsberg et al., 2002
Wakabayashi et al.,
1999
Ginsberg et al., 2002
Counts et al., 2009;
Tan et al., 2010
McLachlan et al., 1988;
Clark et al., 1989;
Kittur et al., 1994;
Colangelo et al., 2002;
Ginsberg et al., 2002
Ginsberg et al., 2002;
Bi and Sze 2002;
Tan et al., 2010
Ginsberg et al., 2002;
Bi and Sze 2002
Heffernan et al., 1998;
Callahan et al., 1999;
Colangelo et al., 2002;
Ginsberg et al., 2002;
Mufson et al., 2002
Ginsberg et al., 2002;
Mufson et al., 2002;
Tan et al., 2010
Arc, activity-regulated cytoskeleton-associated protein; Bdnf, brain-derived neurotrophic factor; Cdk5,
cyclin-dependent kinase 5; Egr1, early growth response protein 1; GluR, glutamate receptor; Nfl,
neurofilamin; NR2, N-methyl D-aspartate receptor subtype 2; Syp, synaptophysin; Syt1,
synaptotagmin.
2 0 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
Supplementary Table 2
Primer sequences used for promoter amplification
Promoter
Forward (5’-3’)
Reverse (5’-3’)
Arc
CAGCATAAATAGCCGCTGGT
GAGTGTGGCAGGCTCGTC
Bdnf 1
TGATCATCACTCACGACCACG
CAGCCTCTCTGAGCCAGTTACG
Bdnf 2
TGAGGATAGTGGTGGAGTTG
TAACCTTTTCCTCCTCC
Bdnf 4
GCGCGGAATTCTGATTCTGGTAAT
GAGAGGGCTCCACGCTGCCTTGACG
Cdk5
CGCAGCCTGTTGGACTTTGT
GCGTTGCAGAGGAGGTGGTA
Egr1
GTGCCCACCACTCTTGGAT
CGAATCGGCCTGTATTTCAA
Homer1
CTGCCTGAGTGTCGTGGAAG
ATGATTTCACTCGCGCTGAC
GluR1
GGAGGAGAGCAGAGGGAGAG
TTCCTGCAATTCCTTGCTTG
GluR2
GCGGTGCTAAAATCGAATGC
ACAGAGAGGGGCAGGCAG
Nfl
GCCAGACGAAAGTCAGGAAG
TTTTCACCAGCAGTTTGCAG
NR2A
TCGGCTTGGACTGATACGTG
AGGATAGACTGCCCCTGCAC
NR2B
CCTTAGGAAGGGGACGCTTT
GGCAATTAAGGGTTGGGTTC
Syp
CTAGCCTCCCGAATGGAATG
CAGCAGCAGCATCAGCAATG
Syt1
CTGAACAGGTTGAGGGCATT
CCTGAGGAGAGGGGTTTAGG
-Actin
CCCATCGCCAAAACTCTTCA
GGCCACTCGAGCCATAAAAG
-Globin
TGACCAATAGTCTCGGAGTCCTG
AGGCTGAAGGCCTGTCCTTT
-Tubulin
TCCAGGGATGAAGAATGAGG
TGAGCACTGGTAGGGAGCTT
Segment 1
TCTGGCTAGAAGCACATCCA
TAGGTGTGGGCAAAAGAAGG
Segment 2
CCTTCTTTTGCCCACACCTA
TGGCAGACTCCTTGTATCTCC
Segment 3
CCGAGTCCTGGAAATGCTTA
ACCTGGTGGTCCGATATTCC
Segment 4
ATATCGGACCACCAGGTAGC
TTTAGCCCCTTGCTCAGAGA
Segment 5
TAAGACCGAGGGGTGAACCT
CCAGGGCGACAGTAGTGTTT
Segment 6
CCCGTAGAAACACTACTGTCG
GTAGGGCACAGAGCGGGATA
Segment 7
TATCCCGCTCTGTGCCCTAC
CGCCTCCTTGACTGTACGC
Segment 8
CATGGCGTACAGTCAAGGAG
TCTATGAGGCTTCATGGGATG
Segment 9
TATTATGGCCAGGGTCATCC
GACGTTAAATCTCTGCATCTGCT
Hdac2
Arc, activity-regulated cytoskeleton-associated protein; Bdnf 1, brain-derived neurotrophic factor,
promoter region of exon I; Cdk5, cyclin-dependent kinase 5; Egr1, early growth response protein 1;
GluR1, glutamate receptor 1; Nfl, neurofilamin; NR2, N-methyl D-aspartate receptor subtype 2; Syp,
synaptophysin; Syt1, synaptotagmin; Hdac2, histone deacetylase 2; segment 1, first segment
amplified in the 1300bp-long Hdac2 promoter region (see Fig. 3l).
W W W. N A T U R E . C O M / N A T U R E | 2 1
RESEARCH SUPPLEMENTARY INFORMATION
Supplementary Table 3
Primer sequences used for coding sequence amplification
Coding
Forward (5’-3’)
Reverse (5’-3’)
region
Arc
GAAGTGGTGGGAGTTCAAGC
CTCCTCAGCGTCCACATACA
Amplicon
position (bp
from TSS)
753
Bdnf
CATTCAGCACCTTGGACAGA
CAGCCTACACCGCTAGGAAG
707
Cdk5
GGCTAAAAACCGGGAAACTC
CCCAATACCAGCCCAGTCTA
622
Egr1
GGGAGGGTTTGTTTTGATGA
TTTCAACAGCTGACGCAAAC
553
Homer1
CATAGCCAAAAGCCGGTCTA
ATCATTGCCAACCTTGTTCC
645
GluR1
CACATGTAGCCGGAGTGATG
CACTCAAGAGGATGGGGAAA
622
GluR2
ATTTCGGGTAGGGATGGTTC
ACCATCCTTCACTGGCATTC
640
Nfl
GACAGCCTGATGGACGAGAT
GGCTCTTGAACCACTCTTCG
640
NR2A
TGAGAATTGCTCGGTGTCTG
ACCTGGCACTGTAGGAATGG
564
NR2B
GGCTACGGCTACACATGGAT
CCTCTTCTCGTGGGTGTTGT
582
Syp
CTCCTCGGCTGAATTCTTTG
CATTGGCCCTTTGTTGTTCT
306
Syt1
CGATGCTGAAACTGGACTGA
GGCAGCAGGAAGACTTTGAC
363
-Actin
TGTTACCAACTGGGACGACA
ACCTGGGTCATCTTTTCACG
312
-Globin
CCTTTTTAGGCTGCTGGTTG
AGAATAGCCAGGGGAAGGAA
200
-Tubulin
GCCAAGTCACAATGGAGGTT
ACAGCTTTCAACAGCCCAGT
745
Arc, activity-regulated cytoskeleton-associated protein; Bdnf, brain-derived neurotrophic factor; bp,
base pairs; Cdk5, cyclin-dependent kinase 5; Egr1, early growth response protein 1; GluR1,
glutamate receptor 1; Nfl, neurofilamin; NR2, N-methyl D-aspartate receptor subtype 2; Syp,
synaptophysin; Syt1, synaptotagmin; TSS, transcriptional start site.
2 2 | W W W. N A T U R E . C O M / N A T U R E
SUPPLEMENTARY INFORMATION RESEARCH
Supplementary Table 4
Primer sequences used for exon amplification
Gene
Forward (5’-3’)
Reverse (5’-3’)
Arc
GTTGACCGAAGTGTCCAAGC
CGTAGCCGTCCAAGTTGTTC
Bdnf I
CTCAAAGGGAAACGTGTCTCT
TCACGTGCTCAAAAGTGTCAG
Bdnf II
CTAGCCACCGGGGTGGTGTAA
TCACGTGCTCAAAAGTGTCAG
Bdnf IV
TGCGAGTATTACCTCCGCCAT
TCACGTGCTCAAAAGTGTCAG
Cdk5
CTCATGAGATTGTGGCTCTG
GACGTGGAGTACAGCTTGGC
Egr1
AGCGAACAACCCTATGAGCA
TCGTTTGGCTGGGATAACTC
Gapdh
AGAGAGGGAGGAGGGGAAATG
AACAGGGAGGAGCAGAGAGCAC
GluR1
GTGGTGGTGGACTGTGAATC
TTGGCGAGGATGTAGTGGTA
GluR2
TGTGTTTGTGAGGACTACGGCA
GGATTCTTTGCCACCTTCATTC
Hdac2
GGGACAGGCTTGGTTGTTTC
GAGCATCAGCAATGGCAAGT
Homer1
AGCAGAAGGAAGGCTTGACT
CACGGTACGGCCAATAACTA
Nfl
AGCTGGGTGATGCTTACGAC
AGCTGCACTTGAGCCTTCTC
NR2A
TGCAAGTTACACAGCCAACC
ATCGGAAAGGCGGAGAATAG
NR2B
CCCAGATCCTCGATTTCATT
GCCAAACTGGAAGAACATGG
Syp
GCCACGGACCCAGAGAACAT
GGAAGCCAAACACCACTGAG
Syt1
CATCGACCAGATCCACTTGT
TCGTTTCCTACTTGGCACAC
-Actin
GGGAAATCGTGCGTGACATT
CGGATGTCAACGTCACACTT
-Globin
AGCTGCATGTGGATCCTGAGA
GATAGGCAGCCTGCACTGGT
-Tubulin
TAGTGGAGAACACAGACGAGA
CTGCTGTTCTTACTCTGGATG
Arc, activity-regulated cytoskeleton-associated protein; Bdnf I, brain-derived neurotrophic factor exon I;
Cdk5, cyclin-dependent kinase 5; Egr1, early growth response protein 1; GAPDH, glyceraldehyde-3phosphate dehydrogenase; GluR, glutamate receptor; Hdac2, histone deacetylase 2; Nfl, neurofilamin;
NR2, N-methyl D-aspartate receptor subtype 2; Syp, synaptophysin, Syt1, synaptotagmin.
W W W. N A T U R E . C O M / N A T U R E | 2 3
RESEARCH SUPPLEMENTARY INFORMATION
Supplementary Table 5
Case details
Sample ID
Diagnosis
BM1
BM9
BM11
BM16
BM17
MADRC912
MADRC1301
MADRC1314
MADRC1315
MADRC1323
MADRC1325
MADRC1360
MADRC1368
MADRC1377
MADRC1401
MADRC1424
MADRC1433
MADRC1434
MADRC1454
MADRC1476
MADRC1477
MADRC1483
MADRC1492
MADRC1510
MADRC1522
MADRC1531
MADRC1564
MADRC1576
BB0
BB0
BB0
BB0
BB0
BBIII/IV
BBI/II
BB0
BBV/VI
BBIII/IV
BBV/VI
BBV/VI
BBIII/IV
BBV/VI
BBIII/IV
BBV/VI
BBIII/IV
BBV/VI
BBI/II
BBV/VI
BBI/II
BBV/VI
BB0
BBIII/IV
BBI/II
BBIII/IV
BBV/VI
BBI/II
Sex Age
(years)
Postmortem Interval
(hours)
m
f
f
f
f
f
m
m
f
m
f
f
f
m
m
m
m
f
m
f
f
f
f
m
f
f
m
m
19
23
13
26
47
5
24
36
9
24
30
31
24
24
24
25
48
6
9
14
12
30
4
22
23
12
12
n.d.
58
72
73
73
77
103
85
56
64
94
80
88
95
78
90
89
87
89
85
86
92
90
87
73
68
85
90
88
BB, Braak and Braak stage; BB0, cases with no neurofibrillary tangles and
thus considered healthy controls; BM, Boston Medical Center; MADRC,
Massachusetts Alzheimer Disease Research Center at Massachusetts
General Hospital; n.d., not defined.
2 4 | W W W. N A T U R E . C O M / N A T U R E