Sodium selenate mitigates tau pathology,
neurodegeneration, and functional deficits
in Alzheimer’s disease models
Janet van Eersela,b, Yazi D. Kea, Xin Liua, Fabien Deleruea, Jillian J. Krilb, Jürgen Götza,1, and Lars M. Ittnera,1
aAlzheimer’s and Parkinson’s Disease Laboratory, Brain and Mind Research Institute, University of Sydney, Camperdown, New South Wales 2050, Australia;
and bDepartment of Pathology, University of Sydney, Camperdown, New South Wales 2016, Australia
Communicated by Etienne-Emile Baulieu, Institut National de la Sante et de la Recherche Médicale, Le Kremlin-Bicetre, France, June 25, 2010 (received for
review June 7, 2010)
Alzheimer’s disease (AD) brains are characterized by amyloid-
β-containing plaques and hyperphosphorylated tau-containing
neurofibrillary tangles (NFTs); however, in frontotemporal demen-
tia, the tau pathology manifests in the absence of overt amyloid-β
plaques. Therapeutic strategies so far have primarily been target-
ing amyloid-β, although those targeting tau are only slowly begin-
ning to emerge. Here, we identify sodium selenate as a compound
that reduces tau phosphorylation both in vitro and in vivo. Impor-
tantly, chronic oral treatment of two independent tau transgenic
mouse strains with NFT pathology, P301L mutant pR5 and K369I
mutant K3 mice, reduces tau hyperphosphorylation and completely
abrogates NFT formation. Furthermore, treatment improves contex-
tual memory and motor performance, and prevents neurodegener-
ation. As hyperphosphorylation of tau precedes NFT formation, the
effect of selenate on tau phosphorylation was assessed in more
detail, a process regulated by both kinases and phosphatases. A
major phosphatase implicated in tau dephosphorylation is the ser-
ine/threonine-specific protein phosphatase 2A (PP2A) that is re-
duced in both levels and activity in the AD brain. We found that
selenate stabilizes PP2A-tau complexes. Moreover, there was an
absence of therapeutic effects in sodium selenate-treated tau trans-
genic mice that coexpress a dominant-negative mutant form of
PP2A, suggesting a mediating role for PP2A. Taken together, sodium
selenate mitigates tau pathology in several AD models, making it
a promising lead compound for tau-targeted treatments of AD and
related dementias.
frontotemporal lobar degeneration | protein phosphatase 2A |
neurofibrillary tangle | transgenic | treatment
Alzheimer’s disease (AD) is the most prevalent neurodegen-erative disorder, characterized by progressive loss of cogni-
tion. Histopathologically, AD is defined by two lesions, plaques
and neurofibrillary tangles (NFTs), which result from deposition
of amyloid-β (Aβ) and hyperphosphorylated tau, respectively. Aβ
forms upon cleavage of the amyloid precursor protein by β- and
γ-secretases, and accumulates extracellularly (1). Tau accumu-
lates intracellularly as it becomes increasingly phosphorylated at
both physiological and pathological sites, resulting in reduced
affinity to microtubules and redistribution from the axonal to the
somato-dendritic compartment (2). The amyloid cascade hy-
pothesis places Aβ upstream of tau, a concept supported by AD
mouse models (3, 4). Interestingly, tau depletion in mice prevents
Aβ pathology, suggesting that Aβ toxicity is also tau-dependent
(5). This finding highlights a central pathogenic role of tau in AD.
This role extends to diseases such as frontotemporal dementia,
the second most common form of dementia, where tau lesions are
frequent without overt Aβ pathology (2). Thus, tau can induce
neurodegeneration in the absence of Aβ. Accordingly, expression
of tau in transgenic mouse models recapitulates features of AD
and frontotemporal lobar degeneration (FTLD) (6).
Selenium is a vital trace element enriched in brain (7, 8); its
levels decline with age, and particularly low levels have been linked
to cognitive impairment and AD (9, 10). Short-term administra-
tion of selenium improvesmemory deficits in an acute ratmodel of
dementia (11) and reduces tau phosphorylation in WT rats (12).
Therefore, selenium has been attributed neuroprotective proper-
ties, but the underlying mechanism and its therapeutic potential
remain elusive (11).
To date, treatment of AD and related dementias is limited to
symptomatic relief, with no cure available. Although Aβ has been
the main focus of drug development until recently, tau is in-
creasingly recognized as a target for the treatment of AD and
FTLD (13). Here, we tested putative therapeutic effects of so-
dium selenate, an oxidized form of selenium, on tau pathology,
using cell culture and several transgenic mouse models.
Results
Sodium Selenate Reduces Tau Phosphorylation in Vitro. To test sel-
enate in vitro, we first treated SH-SY5Y neuroblastoma cells that
stably express human tau carrying the FTLD pathogenic mutation
P301L (SH-P301L). In these cells, tau is phosphorylated at mul-
tiple sites, including the pathological epitope Ser422 (pS422),
a site correlating with NFT formation in vivo (3, 14) (Fig. 1 A
and B). Selenate treatment reduced pS422 staining without af-
fecting tau expression levels (Fig. 1A andB, and Fig. S1).Western
blotting of extracts obtained from both selenate-treated and un-
treated cells in the presence of the protein phosphatase 2A
(PP2A) inhibitor okadaic acid (OA) revealed a dose-dependent
reduction of tau phosphorylation at multiple sites, including
pS422, 12E8, and PHF-1 (Fig. 1 C).
Toxicity has been reported for some forms of selenium, such as
sodium selenite (15), which would severely compromise a thera-
peutic use. We thus assessed neurotoxicity of sodium selenate
compared with selenite, a less oxidized state of selenium, by mea-
suring lactate dehydrogenase (LDH) release in primary hippo-
campal cultures (Fig. 1 D). Although selenite was toxic already
at low concentrations, selenate showed no toxicity, even at high
100 μM concentrations. Furthermore, chronic treatment of mice
with selenate over 4 mo did not cause any overt side effects or any
form of neurotoxicity (see below).
Selenate Improves Tau Transgenic K3 Mice. Given the profound
effects on tau phosphorylation in vitro, we next tested if selenate
also reduces hyperphosphorylation of tau in vivo. K3 mice express
human tau carrying the pathogenic K369I mutation in neurons
(16). These neurons are characterized by a particularly early-onset
Author contributions: J.v.E., J.G., and L.M.I. designed research; J.v.E., Y.D.K., X.L., F.D., and
L.M.I. performed research; J.v.E., Y.D.K., X.L., F.D., J.G., and L.M.I. analyzed data; and
J.v.E., J.J.K., J.G., and L.M.I. wrote the paper.
The authors declare no conflict of interest.
1To whom correspondence may be addressed. E-mail: jgoetz@med.usyd.edu.au or
littner@med.usyd.edu.au.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1009038107/-/DCSupplemental.
13888–13893 | PNAS | August 3, 2010 | vol. 107 | no. 31 www.pnas.org/cgi/doi/10.1073/pnas.1009038107
Parkinsonism associated with abundant tau phosphorylation, and
axonal transport defects (16, 17). We first treated 6-wk-old K3
mice with 12 μg/mL selenate added to the drinking water ad libi-
tum, with controls receiving no drug, and monitored the pro-
gression of motor symptoms. Selenate did not affect total water
consumption (Fig. S2). K3 mice normally exhibit motor deficits
on the Rota-Rod already at 6 wk, compared with WT (Fig. 2A).
However, although untreated K3 mice (further) declined dra-
matically over time until they failed to stay on the rod at all, sele-
nate-treated K3 mice showed a gradually improving motor
performance, such that after 6 wk of treatment they were no longer
distinguishable fromWT littermates. The motor improvement was
associated with decreased tau phosphorylation in the motor cortex,
hippocampus, and substantia nigra (Fig. 2B and Fig. S3). Staining
of brain sections for tau phosphorylated at Thr231/Ser235 (AT180
epitope), a major phosphorylation site in K3 mice (16), revealed
significantly reduced phosphorylation in selenate- compared with
untreatedmice. Importantly, transgenic taumRNA levels were not
altered upon treatment (Fig. S1). Hence, selenate reduces both tau
phosphorylation and early-onset motor deficits in young K3 mice.
Tau pathology in K3 mice is progressive. Beyond the age of
4 mo, deposition of tau results in overt Bielschowsky-silver-posi-
tive NFT-like inclusions and axonal spheroids in many brain
areas, as well as in neurodegeneration in the substantia nigra (16,
17). When we treated 4-mo-old K3 mice until 8 mo of age with
selenate, numbers of inclusions were significantly reduced, as
shown for the hippocampal subiculum and midbrain (Fig. 2 C–E).
Furthermore, numbers of spheroids, a result of axonal transport
defects in K3 mice (17), were significantly lower in treated mice
(Fig. 2F). In fact, spheroids were completely absent from frontal
cortex, suggesting that selenate reverts functional impairments
leading to spheroids. K3 mice are characterized by a substantial,
age-dependent degeneration of cerebellar basket neurons, re-
sulting in the absence of pinceau terminals formed by clustered
axons surrounding Purkinje cells (Fig. 2G). Selenate treatment
fully prevented this degeneration. Taken together, these findings
show that selenate reduces tau phosphorylation and deposition,
mitigates pathological spheroid formation, and prevents axonal
degeneration of distinct neuronal populations in K3 mice.
Selenate Halts Pathology in Tau Transgenic pR5 Mice. Next, we
treated 8-mo-old pR5mice that express P301Lmutant human tau
in neurons (18) for 4 mo with 12 μg/mL selenate added to the
drinking water. The pR5 mice present with a progressive tau pa-
thology, including NFT formation initiated at around 6 mo of age
(18). These mice are characterized by an amygdala-dependent
impairment in the conditioned taste aversion (CTA) paradigm
(19). As expected, untreated pR5 mice displayed a significant
impairment, although selenate-treated pR5 mice showed no im-
pairment, performing similar to wild-type littermates and hence,
suggestive of improved contextual memory (Fig. 3A). Basic taste
qualities were normal in untreated and treated pR5 mice (Fig.
S2). Next, we determined whether the functional improvement of
selenate-treated pR5 mice was associated with changes in tau
pathology. Double-staining for human tau (HT7) and tau phos-
phorylated at Ser422 (pS422) revealed reduced phosphorylation
in CA1 neurons of selenate-treated compared with untreated
pR5 mice, although staining for total tau was comparable (Fig.
3B). As for the K3 mice, transgenic tau mRNA levels were not
altered by selenate (Fig. S1). Next, we determined if selenate
treatment reduces levels of insoluble tau, a step critical in NFT
formation. Therefore, we extracted pR5 brain tissue with either
formic acid (FA) or sarkosyl to obtain insoluble proteins (16, 20).
Consistent with the histopathological finding, Western blotting
revealed reduced phosphorylation of tau in selenate-treated pR5
brains, although total levels of soluble tau were comparable in
treated and untreated pR5 mice (Fig. 3 C and D). However, both
FA and sarkosyl extraction revealed markedly less insoluble tau in
selenate-treated compared with untreated pR5 mice. The amyg-
dala is a major site of NFT formation in pR5mice (18). Here, both
numbers of neurons stained with pS422, a marker of severe tau
pathology (3), and numbers of Gallyas-positive NFTs were sig-
nificantly reduced in selenate-treated compared with untreated
pR5 brains (Fig. 3 E and F). Specifically, NFT numbers in 12-mo-
old selenate-treated pR5 mice were similar to those in 8-mo-old
untreated pR5 mice (3), suggesting that treatment had halted
disease progression.
Selenate-Induced Improvements Require PP2A. Reduced PP2A ac-
tivity has been implicated in AD (21, 22). Because selenate antag-
onized the PP2A inhibitor OA in SH-SY5Y cells (Fig. 1C), we
addressed PP2A function in more detail. Regulation of tau phos-
phorylation by PP2A involves direct binding (23). Interestingly,
coimmunoprecipitation of PP2A from SH-SY5Y cells with a tau-
specific antibody revealed a markedly increased tau-PP2A inter-
action in the presence of increasing doses of selenate (Fig. 4A and
Fig. S4), suggesting that selenate enhances tau binding of PP2A.
To determine the role of PP2A in mediating the effects of sel-
enate on tau pathology in vivo, we crossed pR5 with Dom5 mice
(pR5.Dom5). Dom5 mice express a substrate-specific dominant-
negative mutant form of PP2A, L309A (24). The two transgenes
show an overlapping expression pattern in pR5.Dom5 mice, in-
cluding the hippocampus (25). Although selenate treatment of
pR5 mice reduced levels of tau phosphorylation significantly,
treatment had no effect in pR5.Dom5 mice, as determined by
Western blot analysis of hippocampal extracts (Figs. 2C and 4B
D
controlA
B
224
Sp
7T
H
egre
m
selenate
HT7
*
†
pS422
Gapdh
C 20nM OA
†
- - 1.0 1 01
001
0001
selenate:
(µM)
12E8 †
PHF-1 †
selenate
selenite
* *
*
4
3
2
1
0
yti ci xot
ev ita ler
)rt
Cf o
d lof (
conc (mM)
0000.0
1 000.0
1 00.0
1
0.0
1 .0 1 10
50
100)
%(
sllec
#
HT7pS422
I
selenate control
I
Fig. 1. Sodium selenate mitigates aberrant tau phosphorylation in SH-SY5Y
neuroblastoma cells. (A) Immunocytochemistry of human P301L mutant tau
expressing SH-SY5Y cells treated with selenate. Phosphorylation of tau at
the pathological epitope Ser422 (pS422) is markedly reduced in selenate-
treated cells, although total tau (HT7) is comparable to untreated controls.
(B) Flow cytometry confirms reduced pS422 phosphorylation, but similar HT7
staining upon selenate treatment, compared with control. I, fluorescence
intensity. (C) OA induces phosphorylation of tau at multiple epitopes (pS422,
12E8, PHF-1), resulting in a molecular weight shift (†) in SH-SY5Y cells.
Cotreatment with increasing doses of selenate shows a dose-dependent
reduction of tau phosphorylation, but levels of nonphosphorylated tau
(asterisk) remain unaltered. (D) Dose-dependent increase in toxicity of sel-
enite, but not selenate in primary hippocampal neurons, as determined by
LDH release (*P < 0.001).
van Eersel et al. PNAS | August 3, 2010 | vol. 107 | no. 31 | 13889
N
EU
RO
SC
IE
N
CE
and Fig. S4). Furthermore, extraction of insoluble proteins
revealed high levels of insoluble tau in selenate-treated pR5.
Dom5, but not pR5 mice. Consistent with these findings, the his-
tological analysis revealed that selenate had no effect on pre-
venting tau phosphorylation and NFT formation in pR5.Dom5
mice (Fig. 4 C and D). Therefore, the absence of therapeutic
effects in pR5.Dom5mice suggests a role for PP2A in the selenate-
induced reduction of tau phosphorylation.
Increased kinase activities have been implicated in AD (26).
Although selenate may alter kinase function, Western blot
analysis of wild-type and K3 brains with phosphorylation (ac-
tivity)- and total kinase-specific antibodies (GSK3β, ERK1/2, p38
Mapk, and cdk5) revealed no alterations upon chronic selenate
treatment (Fig. S5). Therefore, altered kinase activity seems not
to contribute to the effects of selenate on tau phosphorylation.
Discussion
In the present study, we show that chronic low doses of sodium
selenate reduce tau phosphorylation in both cell culture and
mouse models of disease. Treatment prevents memory and motor
deficits, NFT formation, and degeneration in two tau transgenic
mouse lines with robust pathologies. These effects are, at least in
part, mediated by PP2A, because selenate does not reduce tau
phosphorylation in mice coexpressing a defective PP2A subunit
(which causes reduced activity for substrates such as tau).
PP2A is a heterotrimeric complex composed of a structural A,
a catalytic C, and a variable regulatory B subunit (27). Four classes
of regulatory subunits, B (PR55), B’ (B56 or PR61), B’’ (PR72),
and B’’’ (PR93/PR110), with several members in each subfamily,
define substrate specificity and subcellular localization of the
holoenzyme. This result is also true for the PP2A substrate tau
BA
ED
C control selenate controlWT
** * * * * *** * * **
* * *
** * * * * * * * * * *
selenate
K3 selenate
K3 control
WT control
WT selenate
0
0 1 2 3 4 5 6 7 8 9 101112
60
120
180
)ces(
do
R-ato
R
no
e
mit
duration of treatment (wks)
F
G
cortex CA1 SN
yks
wohcslei
B
e tan eles
l ortnoc
yks
wohcslei
B
F
N
/
AP
lortnoc
etaneles
60
40
20
0
noitc es/
sn ois el
*
lortnoc
etaneles
150
100
50
0
noi tces /
s diorehps
*
lortnoc
etaneles
12
8
4
0
noitc es/
s nois el
*
#
*
Fig. 2. Chronic selenate treatment improves pathology in both young and aged K369I mutant tau transgenic K3 mice. (A) Rota-Rod testing in acceleration
mode of selenate-treated and untreated (control) WT and K3 mice. Despite pronounced deficits in K3 compared with WT mice at the onset of treatment
(6 wk of age), after 5 wk of chronic selenate treatment motor performance of K3 mice is improved such that performance does not differ significantly from
WT (#n.s.), although untreated K3 control mice continue to deteriorate, staying only for a minimal time on the rod (*P < 0.05, treated vs. untreated
K3 mice). (B) Selenate treatment for 12 wk significantly reduces staining of K3 brains with the phospho-tau antibody AT180 (Thr231/Ser235), a dominant
phospho-epitope in K3 mice, in neurons of cortex, hippocampal CA1, and substantia nigra (SN), compared with untreated K3 mice (for quantification, see
Fig. S3). (Scale bars, 50 μm.) (C) Bielschowsky-silver positive NFT-like lesions (arrows) are numerous in the superior colliculus of the midbrain of 8-mo-old
untreated K3 mice (control), but are significantly reduced after 4 mo of treatment with selenate. (Scale bar, 50 μm.) (D) Quantification reveals that numbers
of lesions are 4.4-fold lower in the superior colliculus of selenate-treated compared with untreated (control) K3 mice (*P < 0.0001). (E) Numbers of NFT-like
lesions in the subiculum are 2.87-fold lower in selenate-treated compared with untreated (control) K3 mice (*P < 0.0001). (F) Axonal spheroids are 3.68-fold
less frequent in the cortex of selenate-treated compared with untreated (control) K3 mice (*P < 0.0001). (G) Bundled axons of cerebellar basket cell form
Pinceau terminals (arrows) around the initial axon segment of the Purkinje cell (PC, asterisks), as visualized by Bielschowsky-silver impregnation and
neurofilament (NF) staining (red) in WT brains. Eight-month-old untreated K3 mice (control) show a pronounced degeneration of Pinceau terminals,
resulting in the absence of NF- and Bielschowsky-positive axons around PCs (open arrows). Treatment with selenate fully prevents this degeneration. PCs are
stained with parvalbumin (PA, green). (Scale bar, 50 μm.)
13890 | www.pnas.org/cgi/doi/10.1073/pnas.1009038107 van Eersel et al.
that is dephosphorylated, upon binding, by distinct isoforms of
PP2A (23, 28). Interestingly, we found that selenate increases
binding of PP2A and tau in cell culture, which may explain its
effects on tau phosphorylation. Other mechanisms of regulating
PP2A activity may include alterations in the methylation of the
core enzyme or dislodging of catalytic metal atoms (29, 30). Here,
the exact molecular effects of selenate on PP2A and other likely,
yet unidentified cellular targets remain to be elucidated. How-
ever, given the role proposed for PP2A in tau pathology in AD
(21, 22, 28, 31), activating PP2A represents an attractive thera-
peutic approach (26). It should be considered, however, that in-
terfering with the activity of PP2A may cause side effects, as the
holoenzyme participates in several signaling cascades in many
tissues other than brain. Hence, a substrate-specific activation of
PP2A rather than a broad inhibition needs to be achieved. Al-
though in K3 and pR5 mice, tau phosphorylation was reduced in
the absence of overt central or peripheral side effects, it cannot
fo