Content uploaded by Eun-Jae Lee
Author content
All content in this area was uploaded by Eun-Jae Lee on Dec 07, 2015
Content may be subject to copyright.
Available via license: CC BY 4.0
Content may be subject to copyright.
ARTICLE
Received 17 Feb 2015 |Accepted 14 Apr 2015 |Published 18 May 2015
Trans-synaptic zinc mobilization improves social
interaction in two mouse models of autism through
NMDAR activation
Eun-Jae Lee1,2, Hyejin Lee2,3, Tzyy-Nan Huang4, Changuk Chung2,3, Wangyong Shin2,3,
Kyungdeok Kim2,3, Jae-Young Koh5,6,7, Yi-Ping Hsueh4& Eunjoon Kim2,3
Genetic aspects of autism spectrum disorders (ASDs) have recently been extensively
explored, but environmental influences that affect ASDs have received considerably less
attention. Zinc (Zn) is a nutritional factor implicated in ASDs, but evidence for a strong
association and linking mechanism is largely lacking. Here we report that trans-synaptic Zn
mobilization rapidly rescues social interaction in two independent mouse models of ASD. In
mice lacking Shank2, an excitatory postsynaptic scaffolding protein, postsynaptic Zn elevation
induced by clioquinol (a Zn chelator and ionophore) improves social interaction. Postsynaptic
Zn is mainly derived from presynaptic pools and activates NMDA receptors (NMDARs)
through postsynaptic activation of the tyrosine kinase Src. Clioquinol also improves social
interaction in mice haploinsufficient for the transcription factor Tbr1, which accompanies
NMDAR activation in the amygdala. These results suggest that trans-synaptic Zn
mobilization induced by clioquinol rescues social deficits in mouse models of ASD through
postsynaptic Src and NMDAR activation.
DOI: 10.1038/ncomms8168 OPEN
1Graduate School of Medical Science and Engineering, Korea Advanced Institute of Science and Technology, Daejeon 305-701, Korea. 2Center for Synaptic
Brain Dysfunctions, Institute for Basic Science (IBS), Daejeon 305-701, Korea. 3Department of Biological Sciences, Korea Advanced Institute of Science and
Technology, Daejeon 305-701, Korea. 4Institute of Molecular Biology, Academia Sinica, Taipei 115, Taiwan. 5Neural Injury Research Lab, University of Ulsan
College of Medicine, Seoul 138-736, Korea. 6Asan Institute for Life Science, University of Ulsan College of Medicine, Seoul 138-736, Korea. 7Department
of Neurology, University of Ulsan College of Medicine, Seoul 138-736, Korea. Correspondence and requests for materials should be addressed to Y.-P.H.
(email: yph@gate.sinica.edu.tw) or to E.K. (email: kime@kaist.ac.kr).
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 1
&2015 Macmillan Publishers Limited. All rights reserved.
Autism spectrum disorders (ASDs) represent a neurodeve-
lopmental disorder characterized by impaired social
interaction and communication, and restricted and
repetitive behaviour, interest and activity. ASDs affect B1% of
the population and are thought to be strongly influenced by
genetic factors. A large number of ASD-associated genetic
variations have recently been identified, indicating that ASDs
represent a genetically heterogeneous family of disorders1–3.
Some of the genetic variations lie along common pathways/
functions, including synaptic transmission, transcriptional
regulation and chromatin remodelling1–3. In addition, studies
using mouse models of ASD carrying these mutations have begun
to suggest possible mechanisms that may underlie the
pathogenesis of ASD, namely glutamatergic dysfunction and an
imbalance between excitatory and inhibitory synapses4–14.
Environmental influences, such as nutrition, toxins and
poisons, drugs, infection and stress, are thought to have a
significant influence on psychiatric disorders. In ASDs, well-
known examples of environmental influences include pre- or
perinatal exposure to viruses or teratogens such as valproic
acid and thalidomide15,16. However, studies on additional
environmental influences and underlying mechanisms are at an
early stage. This contrasts with the rapidly growing evidence
for the contribution of genetic factors to ASDs. Because
environmental factors are highly likely to interact with the
genetic variations of ASD to determine the type, severity and
trajectory of ASD symptoms, a balance between genetic and
environmental causes is required in studies of ASDs.
Zinc (Zn), the second-most abundant trace element with a
critical role in human nutrition and health, regulates a variety of
cellular processes and protein functions. Zn deficiency has been
implicated in diverse neurological and psychiatric disorders,
including Alzheimer’s disease, Parkinson’s disease, ASDs, atten-
tion deficit/hyperactivity disorder, schizophrenia, epilepsy and
mood disorders17. The association of Zn with ASDs has been
suggested based on its deficiency in individuals with ASDs,
including a recent large cohort of 1,967 children16,18, as well as
the phenotypes of Zn-deficient experimental animals19. This
association is further supported by the potential therapeutic value
of Zn supplementation in ASD treatment17,20. However, strong
evidence supporting the association between Zn deficiency and
ASDs is largely unavailable, and the mechanisms underlying the
association remain obscure.
In the synapse, the main pool of Zn ions is presynaptic vesicles
where Zn is in the millimolar range, whereas postsynaptic sites
contain much smaller amounts of Zn (picomolar range)21–24.
Presynaptic free Zn is co-released with glutamate during neuronal
activity and serves to suppress NMDA receptors (NMDARs) in
the synaptic cleft. Some Zn ions enter the postsynaptic sites
through calcium channels, NMDARs and calcium-permeable
AMPA receptors (AMPARs), and regulate target proteins such as
NMDARs and TrkB receptors through mechanisms including
those involving Src family tyrosine kinases (SFKs)25–27. Another
important effector of postsynaptic Zn is Shank (also known as
ProSAP), a family of excitatory postsynaptic scaffolding proteins
with three known members (Shank1/2/3; refs 28,29). Zn binds to
Shank2/3 and enhances their postsynaptic stabilization,
promoting excitatory synapse formation and maturation30.
Shank2/3, members of the Shank family of postsynaptic
scaffolding proteins (also known as ProSAP1/2), have been
implicated in ASDs through human genetic studies31–36 and
mouse model/cultured neuron studies19,30,37–48. Mice carrying
Shank2/3 mutations display diverse dysfunctions at glutamate
synapses40–46,49. One notable change is the reduction in NMDAR
function observed in Shank2 /mice (exons 6 þ7 deletion)45.
In these mice, normalization of NMDAR function with an
NMDAR agonist (D-cycloserine) is associated with the rescue of
impaired social interaction, suggesting that NMDAR
hypofunction might underlie the social deficit in these mice.
Although validation of this hypothesis will require further
analyses, D-cycloserine has also been shown to rescue the
impaired social interaction in mice with a haploinsufficiency of
the transcription factor Tbr1 (T-box brain 1; ref. 50), which
positively regulates the expression of Grin2b (ref. 51), encoding
the GluN2B subunit of NMDARs.
In the present study, we demonstrate that trans-synaptic
Zn mobilization by clioquinol, a Zn chelator and ionophore
(termed CQ hereafter), rescues the social interaction deficits in
Shank2 /and Tbr1 þ/mice. CQ mobilizes Zn from enriched
presynaptic pools to postsynaptic sites, where it enhances
NMDAR function through Src activation. These results indicate
that postsynaptic Zn rescues social interaction deficits in distinct
mouse models of ASDs, and suggest that reduced NMDAR
function is associated with ASDs.
Results
CQ rapidly improves social interaction in Shank2 /mice.
Based on the close associations among Zn, NMDAR, Shank and
ASD mentioned above, we reasoned that Zn delivered to post-
synaptic compartments might rescue the reduced NMDAR
function and ASD-like behaviours observed in Shank2 /mice.
To test this idea, we first intraperitoneally (i.p.) injected
Shank2 /mice with CQ (30 mg kg 1), a lipophilic Zn chelator
(K
d
E10 7) and ionophore that readily crosses the blood–brain
barrier and mobilizes Zn down a concentration gradient52.We
chose systemic administration of CQ because dietary Zn is known
to have poor bioavailability and side effects including gastric
irritation17,53.
Two hours after CQ treatment, the mice were subjected to the
three-chamber social interaction test, which compares the
preference of a mouse for a stranger mouse versus a novel
inanimate object. We found that Shank2 /mice displayed
reduced social interaction compared with wild-type (WT) mice,
as determined by time spent exploring/sniffing the target and the
social preference indices derived from exploration time (see figure
legend for details) (Fig. 1a–c and Supplementary Fig. 1a,d–i),
consistent with previous results from untreated Shank2 /and
WT mice45. This impairment was improved by CQ treatment. In
contrast, social interaction in WT mice was not affected by CQ.
Notably, Shank2 /mice in the chamber with a stranger often
spent time in other activities such as jumping, contrary to WT
mice, as reflected in the relatively low correlation between
exploration time and chamber time (Supplementary Fig. 2), and
showed an apparent lack of CQ-dependent improvement in social
interaction, as determined by the time spent in chamber and the
preference index derived from chamber time (Supplementary
Fig. 1b,c).
When social novelty recognition was determined in the same
three-chamber test, by measuring the preference for a previously
encountered stranger mouse versus a new stranger mouse,
Shank2 /mice showed levels of social novelty recognition
comparable to those in WT mice (Fig. 1d–f and Supplementary
Fig. 1j–m), as reported previously45. In addition, CQ had no effect
on social novelty recognition in both WT and Shank2 /mice.
The CQ-dependent rescue of social interaction in Shank2 /
mice but no effect of CQ on WT mice is unlikely attributable to
differences in the amount of free Zn available for CQ binding in
these mice, because total levels of free Zn measured with the
fluorescent dye TFL-Zn were not different (Supplementary
Fig. 3a,b). In addition, the levels of Zn transporter 3 (ZnT3), a
protein required for Zn transport into presynaptic vesicles54,
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168
2NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications
&2015 Macmillan Publishers Limited. All rights reserved.
which is the main pool of free Zn in the brain, were similar
between genotypes (Supplementary Fig. 3c). Finally, whole-brain
levels of Zn, Cu or Fe were not different between WT and
Shank2 /mice, which is in line with the above-mentioned
TFL-Zn staining result, and, more importantly, 2-hr CQ treat-
ment did not cause an acute reduction in the levels of these
metals in WT or Shank2 /mice (Supplementary Fig. 4a–c),
suggesting that the chelating activity of CQ unlikely contributes
to the observed social rescue.
In repetitive behaviour assays, vehicle-treated Shank2 /
mice showed increased jumping behaviour but normal grooming
in their home cages, relative to vehicle-treated WT mice,
consistent with the previous results45; these behaviours were
unaffected by CQ (Fig. 1g,h and Supplementary Fig. 5a–d).
Similarly, CQ did not affect repetitive behaviours in WT mice.
In the open-field test, Shank2 /mice displayed increased
locomotor activity relative to WT mice, as previously reported.
This hyperactivity was not attenuated by CQ (Fig. 1i,j). Notably,
Shank2 /mice spent less time in the centre region of the
open-field arena, a measure of anxiety-like behaviour. However,
CQ had no effect on the centre-region time in these mice
(Supplementary Fig. 5e). CQ did not affect the repetitive
WT vehicle WT clioquinol
***
***
***
***
WT-V
WT-C
KO-V
KO-C
WT-V
WT-C
KO-V
KO-C
WT-V
WT-C
KO-V
KO-C
WT-V
WT-C
KO-V
KO-C
WT-V
WT-C
KO-V
KO-C
WT-V
WT-C
KO-V
KO-C
WT-V
WT-C
KO-V
KO-C
S1 S2S1 S2S1 S2S1 S2
***
*** *** **
NS
*** **
Exploration time
in 3-chamber (s)
Exploration time
in 3-chamber (s)
250
200
150
100
50
0
150
100
50
0
WT-V
WT-C
KO-V
KO-C
150
100
50
0
NSNSNS
*
150
100
50
0
Min
010 20 30 40 50 60
NS
NS
NS
1,200
1,000
800
600
Total distance moved (m)
Distance moved (m)
Grooming (s)
Jumping (bouts)
400
200
0
800
600
400
200
0
NS
Preference index (S1-O)
from exploration time
Preference index (S2-S1)
from exploration time
100
50
–50
0
100
50
–50
0
OS1 OS1 OS1 OS1
KO clioquinol
Time spent (s)
KO vehicle
OS1O S1
80
40
0
Time spent (s)
80
40
0
OS1O S1
WT vehicle WT clioquinol
KO clioquinol
KO vehicle
S1 S2 S1 S2
S1 S2 S1 S2
Figure 1 | CQ treatment rapidly improves social interaction in Shank2/mice. (a–f) CQ improves social interaction in Shank2 /(KO) mice, but has
no effect on WT mice (a–c). Note that levels of social novelty recognition are similar in WT and Shank2/mice, and that CQ does not affect social
novelty recognition in these mice (d–f). Mice were injected with CQ (30 mg kg 1; i.p.), or vehicle, 2 h before behavioural tests. Heat maps in (a,d)
represent examples of mouse movements. The social preference index from exploration time represents the numerical difference between the times spent
exploring or sniffing the two targets (S1/stranger versus O/object or S2/new stranger versus S1/previous stranger) divided by total time spent100. V,
vehicle; C, CQ. (n¼28 for WT-V and WT-C, and 25 for KO-V and KO-C); **Po0.01, ***Po0.001; Kruskal–Wallis one-way analysis of variance (ANOVA)
with Dunn’s post hoc test). (g,h) CQ has no effect on jumping or grooming behaviour in Shank2/mice. (n¼10 for WT-V and WT-C, and 11 for KO-V
and KO-C, *Po0.05, two-way ANOVA and Kruskal–Wallis one-way ANOVA with Dunn’s post hoc test). (i,j) CQ does not normalize hyperactivity in
Shank2 /mice. (n¼23 for WT-V and WT-C, and 21 for KO-V and KO-C; ***Po0.001, two-way ANOVA and Kruskal–Wallis one-way ANOVA with
Dunn’s post hoc test). Data in all panels with error bars represent mean±s.e.m. NS, not significant.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168 ARTICLE
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 3
&2015 Macmillan Publishers Limited. All rights reserved.
behaviour, locomotor activity or anxiety-like behaviour of WT
mice. Together, these results suggest that CQ improves social
interaction but has no effect on social novelty recognition,
repetitive behaviour, hyperactivity or anxiety-like behaviour in
Shank2 /mice.
NMDAR function at Shank2 /synapses is restored by CQ.
The CQ-dependent rescue of social interaction deficits in
Shank2 /mice could involve normalization of the reported
reduction in NMDAR function in these mice45. Consistent with
this possibility, we found that CQ treatment restored normal
levels of NMDAR function at Shank2 /hippocampal Schaffer
collateral-CA1 pyramidal (SC-CA1) synapses, as determined by
the ratio of NMDAR- to AMPAR-evoked excitatory postsynaptic
currents (NMDA/AMPA ratio of eEPSCs; Fig. 2a). CQ, however,
had no effect at WT synapses. In addition, CQ reversed the
reduced tetanus-induced long-term potentiation (LTP), known to
require NMDAR activity, at Shank2 /SC-CA1 synapses, but
had no effect on LTP at WT synapses (Fig. 2b). These results
indicate that CQ restores NMDAR function at Shank2 /
hippocampal SC-CA1 synapses.
We next measured the time course of CQ-dependent NMDAR
activation at Shank2 /synapses. In these experiments, we
treated Shank2 /hippocampal slices with CQ for 20 min (in
contrast to the continuous bath application for experiments
described above) while monitoring NMDAR-mediated field
excitatory postsynaptic potentials (NMDA-fEPSPs) before,
NMDA/AMPA ratio
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
WT-V
WT-C
KO-C
KO-V
WT-V
WT-C
KO-C
KO-V
WT-V WT-C
KO-V KO-C
50 pA
25 ms
25 ms
25 ms
50 pA
AMPAR EPSCs AMPAR EPSCs
NMDAR EPSCs NMDAR EPSCs
EPSCs (% of baseline)
EPSCs (% of baseline)
200
100
0
200
100
0
200 NS
NS
***
*
100
EPSCs (% of baseline)
0
200
100
0
200
100
0
200
100
0
abc
Min
NMDAR EPSCs
AMPAR EPSCs
CQ D-AP5
50 pA
25 ms
cab
KO
0 10203040
abc
abc
abc
abc
NSNS
**
*
fEPSP slope (% of baseline)
fEPSP slope (% of baseline)
WT-V
WT-V
WT-C
WT-C NS
Last 5 min
*
*
KO-V
KO-V
KO-C
KO-C
0.5 mV
10 ms
250
200
150
100
50
0
250
200
150
100
50
0
0
200
100
0
200
EPSCs (% of baseline)
100
250
200
150
100
50
0
a
ab
c
b
bb
WT
WT
WT
KO
KO
CQ
1 mV
aa
Min
0 20406080100
200
150
100
50
0
****
NS
*
*
0–20 20 40 60
Min
NMDAR EPSCs
0 10203040Min
cba
AMPAR EPSCs
CQ D-AP5
Figure 2 | CQ restores NMDAR function at Shank2 /synapses. (a)CQ(4mM) normalizes the NMDA/AMPA ratio at Shank2/hippocampal
SC-CA1 synapses (P17–25), as measured by NMDA- and AMPA-eEPSCs. Representative eEPSC traces recorded at 70 andþ40 mV. NMDA-eEPSCs
were measured at þ40 mV holding potential, 60 ms after the stimulation. (n¼14 cells (from 10 animals) for WT-V, 11 (7) for WT-C, 12 (8) for KO-V and
12 (7) for KO-C, *Po0.05; one-way analysis of variance (ANOVA) with Tukey’s post hoc test). (b)CQ(4mM) restores LTP, induced by tetanus (100 Hz), at
Shank2 /hippocampal SC-CA1 synapses (3–5 weeks), as measured by fEPSPs. (n¼13 slices (from 8 animals) for WT-V, 12 (5) for WT-C, 13 (7) for KO-
V and 11 (5) for KO-C; *Po0.05; one-way ANOVA with Tukey’s post hoc test). (c)CQ(4mM, 20 min) enhances NMDAR function at WT and Shank2 /
hippocampal SC-CA1 synapses, as measured by NMDA-fEPSPs. (n¼8 slices (from 7 animals) for WT and 8 (7) for KO; **Po0.01; Student’s t-test) The
labels a and b indicate 5-min duration before CQ and the end of recording, respectively. (d,e)CQ(4mM, 20 min) enhances NMDAR function at WT and
Shank2 /hippocampal SC-CA1 synapses, as determined by simultaneous measurements of NMDA- and AMPA-eEPSCs at 40 mV. NMDA-eEPSCs
were measured at 60 ms after stimulation. D-AP5 (50 mM, 10 min) was used to test NMDAR dependence. The labels (a–c) indicate 5-min duration before
and after CQ, and at the end of recording, respectively. (n¼4 cells (three slices) for WT and 5 cells (four slices) for KO; *Po0.05, **Po0.01, ***Po0.001;
Student’s t-test, compared with the 5-min duration before CQ). Data in all panels with error bars represent mean±s.e.m. NS, not significant.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168
4NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications
&2015 Macmillan Publishers Limited. All rights reserved.
during and after CQ treatment. With CQ treatment, the initial
slopes of NMDA-fEPSPs at Shank2 /SC-CA1 synapses
gradually increased to B150% of baseline levels and remained
elevated, a response similar to that observed in WT slices
(Fig. 2c). In contrast to NMDA-fEPSPs, AMPAR-mediated
fEPSPs (AMPA-fEPSPs) were not affected by CQ treatment
(Supplementary Fig. 6a). CQ also had no effect on AMPAR-
related input–output ratio (AMPA-fEPSP slopes plotted against
fibre volleys) or paired-pulse ratio at SC-CA1 synapses
(Supplementary Fig. 6b,c).
To further confirm the CQ-dependent NMDAR activation, we
simultaneously measured NMDA- and AMPA-eEPSCs using
patch-clamp recordings. At a holding potential of 40 mV, CQ
increased NMDA-eEPSCs at Shank2 /SC-CA1 synapses, a
result similar to that observed at WT synapses (Fig. 2d,e). The
NMDAR antagonist D-AP5 significantly reduced NMDA-eEPSCs
but not AMPA-eEPSCs, indicating that these events are NMDAR
dependent. In contrast, AMPA-eEPSCs were not affected by CQ
treatment (Fig. 2d,e). Consistent with this, the NMDA/AMPA
ratios derived from these currents were increased in both
genotypes (Supplementary Fig. 6d,e). Taken together, these
results indicate that CQ enhances NMDAR but not AMPAR
function at Shank2 /and WT synapses.
CQ mobilizes Zn from pre- to postsynaptic sites. Next, we
determined whether CQ-dependent NMDAR activation requires
Zn. To test this, we used two different Zn chelators with much
higher affinities for Zn than CQ: Ca-EDTA (K
d
E10 13), which
is membrane impermeable, and TPEN (K
d
E10 15), which is
membrane permeable. Preincubation of slices with Ca-EDTA
before CQ treatment eliminated the CQ-dependent increase in
NMDAR activity at Shank2 /SC-CA1 synapses, as measured
by NMDA-fEPSPs (Fig. 3a). In a control experiment, Ca-EDTA
by itself had no effect on NMDA-fEPSPs (Supplementary Fig. 7a),
as reported previously24,55. TPEN also blocked CQ-dependent
NMDAR activation (Fig. 3b), although TPEN by itself caused a
small increase in the basal activity of NMDARs (Supplementary
Fig. 7b). Collectively, these findings suggest that CQ requires
Zn for NMDAR activation. In addition, the absence of an
effect of Ca-EDTA alone on NMDA-fEPSPs suggests that
CQ-dependent NMDAR activation is unlikely the result of
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
+ Ca-EDTA WT
WT
1 mV 1 mV
NS NS
NS
+ TPEN
CQ
KO
abab
25 ms 25 ms
25 ms
1 mV
Zn 250 nM
Zn 0 nM
KO
WT
KO
CQ
250
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
Wild-type slices
Zn 0 nM
Zn 250 nM
250
200
300
150
100
50
0
250
200
300
150
100
50
00
Zn (nM)
WT
1 mV
25 ms
*
*
*
250
200
150
100
50
0
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
250
200
300
150
100
50
0
NMDA fEPSP slope
(% of baseline)
250
200
300
150
100
50
0
NMDA fEPSP slope
(% of baseline)
+ Cuprizone
Last 5 min
NS
Zn 0 nM
25 ms
1 mV
Zn 250 nM
CQ
CQ CQ
Zn 0 nM
Zn 250 nM
Shank2 –/– slices
ZnT3 –/– slices
CQ WT
KO
KO
250
200
150
100
50
0
NMDA fEPSP slope
(% of baseline)
250
200
150
100
50
0
NMDA fEPSP slope
(% of baseline)
WT
Last 5 min
NS NS
NS
NS
KO
250
abab
250
200
150
100
50
0
250
200
150
100
50
0
0 20406080
Min
0 20406080
Min
ab
020406080
Min
ab
0 20406080
Min
ab
02040
60 80
Min
b
bbaa
a
0
Zn (nM)
250 0 20406080
Min
1 mV
NS
ab
25 ms
Figure 3 | CQ-dependent NMDAR activation requires Zn mobilization from pre- to postsynaptic sites. (a,b)CQ(4mM, 20 min) fails to enhance
NMDAR function at hippocampal SC-CA1 synapses in the presence of Ca-EDTA or TPEN (Zn chelators more potent than CQ), as measured by NMDA-
fEPSPs. Shank2 /hippocampal slices were bath incubated with Ca-EDTA (2 mM) or TPEN (25mM) throughout recordings. The labels a and b indicate
5-min durations before CQ and the end of recordings, respectively. (Ca-EDTA, n¼6 slices (3 animals) for WT and 6 (3) for KO; Student’s t-test; TPEN,
n¼6 (2) for WT and 4 (2) for KO; Student’s t-test). (c) CQ fails to enhance NMDAR function at ZnT3 /hippocampal SC-CA1 synapses. (n¼7 slices
(3 animals); Student’s t-test). (d,e) Exogenously added Zn (250 nM) enhances NMDAR function at WT but not Shank2 /synapses. Additional Zn
was bath applied throughout the recording. (n¼8 slices (7 animals) for 0nM, 7 (4) for 250 nM in WT, and 8 (5) for 0 nM, 9 (4) for 250 nM in KO;
*Po0.05; Student’s t-test). (f) CQ enhances NMDAR function in the presence of cuprizone (100 mM), a Cu2þ-specific chelator. (n¼5 slices (4 animals)
for WT and 5 (4) for KO; *Po0.05; Student’s t-test). Data in all panels with error bars represent mean±s.e.m. NS, not significant.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168 ARTICLE
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 5
&2015 Macmillan Publishers Limited. All rights reserved.
disinhibition of NMDARs by CQ-mediated chelation of Zn in the
synaptic cleft.
If the Zn-ionophoric activity of CQ is the more likely candidate
mediator of NMDAR activation, this raises the question: what is
the source of Zn for NMDAR activation? One possible candidate
is the Zn pool in presynaptic neurotransmitter vesicles, a major
source of Zn in the brain that requires the ZnT3 transporter for
its maintenance21–24. In tests of this possibility using ZnT3-
deficient (ZnT3 /) mice, we found that CQ had no effect on
NMDAR activity in ZnT3 /SC-CA1 synapses (Fig. 3c),
suggesting that the presynaptic Zn pool is required for the CQ
effect. In a control experiment, we confirmed that Zn signals
are indeed largely absent in the ZnT3 /hippocampus
(Supplementary Fig. 3d).
The results described to this point were obtained in experi-
ments performed without exogenous addition of Zn to brain
slices, relying on the physiological concentrations of free Zn in
the extracellular space. Previous studies have reported free Zn
concentrations in the central nervous system of B20 nM (ref. 56),
although higher concentrations might be possible24. Here, we
tested whether increasing the extracellular Zn concentration to
250 nM affected CQ-dependent NMDAR activation. We found
that 250 nM Zn had no significant effect on CQ-dependent
NMDAR activation at Shank2 /SC-CA1 synapses, as
measured by NMDA-fEPSPs (Fig. 3d,e). This result suggests
that additional Zn is not required for CQ-dependent NMDAR
activation at Shank2 /synapses, implying that the presynaptic
Zn pool under physiological conditions is sufficient for
CQ-dependent NMDAR activation. Interestingly, 250 nM Zn
caused a significant increase in NMDA-fEPSPs at WT synapses
(Fig. 3d,e), a differential effect that warrants further investigation.
We next attempted to visualize CQ-dependent increases in Zn
levels in postsynaptic compartments, using ZnAF-2DA, a
membrane-permeable Zn indicator that, once inside the cell, is
modified and trapped to indicate intracellular Zn levels57,58.
When WT mice were treated with CQ for 2 h, Zn signals
measured by two-photon confocal microscopy were significantly
increased in both dendritic and cell body area of the hippocampal
CA1 region, compared with vehicle-treated controls (Fig. 4a–f;
and Supplementary Movies 1 and 2), consistent with the previous
results obtained using a regular confocal microscope58.
Finally, because CQ can bind Cu2þ(K
d
E10 8.9) in addition
to Zn, we tested whether the Cu2þ-binding activity of CQ also
contributed to its effects on NMDAR function. Application of
cuprizone, a selective Cu2þchelator, to hippocampal slices
before CQ treatment did not inhibit the CQ-induced increase in
NMDAR activity (Fig. 3f), suggesting that the Cu-binding
activity of CQ is not important for NMDAR activation. Taken
together, these results suggest that CQ enhances NMDAR
function through its Zn-ionophoric, but not Zn-chelating or
Cu2þ-binding, activity, and uses mainly the presynaptic Zn pool
for NMDAR activation.
CQ activates NMDARs through postsynaptic Src. The results
described thus far suggest that CQ mobilizes Zn from presynaptic
vesicles into the synaptic cleft and postsynaptic compartments.
Zn in the synaptic cleft is unlikely to contribute to NMDAR
activation because Zn released presynaptically during neuronal
activity is known to inhibit NMDARs24, and we demonstrated
that Ca-EDTA has no effect on NMDA-fEPSPs (Supplementary
Fig. 7a).
If postsynaptic Zn delivery is an important factor, then what
would be the underlying mechanism for NMDAR activation?
Previous studies have shown that Zn binds to and inactivates
C-terminal Src kinase, a negative regulator of SFKs, which
phosphorylate and activate NMDARs27,59. In related
experiments, we found that two independent SFK inhibitors,
PP2 and SU6656, applied to hippocampal slices before CQ
treatment abolished CQ activation of NMDARs at both
Vehicle
Slice depth (5 μm)
Cell body
Cell
body
Dendrite
** ** **
Dendrite
Vehicle
Vehicle
CQ
Vehicle
CQ
CQ
Total
ZnAF-2DA intensity (au)
ZnAF-2DA intensity (au)
ZnAF-2DA intensity (au)
Vehicle
CQ
Ratio
NS
Dendrite / cell body
ZnAF-2DA intensity ratio
8
6
4
2
0
40
30
20
10
0
80
60
40
20
00
50
100
150
Slice depth (50 μm) Slice depth (5 μm) Slice depth (50 μm)
Clioquinol
Figure 4 | CQ treatment increases Zn signals in the dendritic and cell body areas in the hippocampal CA1 region. (a–f) Mice (3–5 weeks) were injected
with CQ (30 mg kg 1; i.p), or vehicle, 2 h before brain slicing, and hippocampal slices (300 mm thick) were incubated with ZnAF-2DA (40 min) followed by
two-photon microscopy. Zn signals (average intensities) were measured in the indicated three square regions in the cell body, or dendritic, area of the
hippocampal CA1 region. Scale bars, 50 mm(a,b). (n¼12 slices (6 animals) for vehicle and 12 slices (5 animals) for CQ; *Po0.05, **Po0.01, Student’s
t-test). Data in all panels with error bars represent mean±s.e.m. Au, arbitrary unit; NS, not significant.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168
6NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications
&2015 Macmillan Publishers Limited. All rights reserved.
Shank2 /and WT SC-CA1 synapses (Fig. 5a,b). In control
experiments, PP3, an inactive PP2 analogue, failed to block the
NMDAR activation (Fig. 5c).
To further confirm that CQ enhances NMDAR function
through SFKs and to test that the subcellular site of SFK action is
indeed postsynaptic, we used the Src-inhibitory peptide
Src(40–58), which contains a sequence corresponding to Src
amino-acid residues 40–58 and selectively blocks endogenous Src
but not other SFK members59–61. Inclusion of this peptide
(0.03 mg ml 1) in the patch pipette during patch-clamp
recordings prevented CQ from increasing NMDAR activity at
Shank2 /and WT synapses, determined by measuring
NMDA-eEPSCs (Fig. 5d,e). In control experiments, a scrambled
Src-peptide variant (sSrc 40–58) had no effect on the
CQ-dependent increase in NMDAR activity.
In contrast to NMDARs, AMPAR function was unaffected by
the Src-inhibitory peptide, as determined from simultaneous
recordings of AMPA-eEPSCs at both WT and Shank2 /
synapses (Fig. 5d,e). Consistent with this, the CQ-dependent
increase in the NMDA/AMPA ratio determined from these
currents was blocked by the Src-inhibitory peptide, but not by the
scrambled peptide (Supplementary Fig. 8).
We also tested whether CQ treatment enhances Src activity, in
addition to NMDAR function, by immunoblot analysis of
CQ-treated hippocampal slices. We found that the levels of
tyrosine phosphorylation of Src at Y416, known to render Src
fully active59, were increased in CQ-treated WT and Shank2 /
slices, which was blocked by PP2 (Fig. 6a,b,d,e). In contrast,
tyrosine phosphorylation of Src at Y527, known to stabilize the
inactive conformation of Src59, was not affected by CQ treatment
(Fig. 6a,c,d,f), suggesting that CQ promotes Src activation
through the phosphorylation of distinct tyrosine residues.
Finally, in order to explore the involvement of other signalling
pathways in the downstream of NMDAR activation, we tested
inhibitors of MAPK kinase/MEK (PD98059) and CaMKIIa
(KN93). We found that suppression of MAPK/Erk by the
NMDA fEPSP slope
(% of baseline)
250
200
150
100
50
0
200
NMDA fEPSP slope
(% of baseline)
100
AMPAR EPSCs
Src(40–58)
sSrc(40–58)
Src(40–58)
sSrc(40–58)
50 pA
25 ms
NMDAR EPSCs
010
ab
20 30 40
Min
D-AP5CQ
250
200
150
100
50
0
0
200
EPSCs (% of baseline)
100
0
0204060
ba Min
80
0204060
b
aMin
80
+ PP2
+ PP3
CQ
CQ
WT
KO
WT
KO
250
NMDA fEPSP slope
(% of baseline)
NMDA fEPSP slope
(% of baseline)
250
200
150
100
50
0
NS NS
NS
NMDA fEPSP slope
(% of baseline)
250
200
150
100
50
0abab
NS NS
NS
250
200
150
100
50
0
b
bab
KOWTKOWT
***
NS
NMDA fEPSP slope
(% of baseline)
EPSCs (% of baseline)
EPSCs (% of baseline)
EPSCs (% of baseline)
250
200
150
100
50
0aba b
KO
WT
AMPAR EPSCs
NMDAR EPSCs
200
NS
NS
NS
*
*
NS
100
0
200
100
0
200
100
0
200
100
0
0 10203040
Min ba
NMDAR EPSCs
AMPAR EPSCs
Src(40–58)
sSrc(40–58)
Src(40–58)
sSrc(40–58)
50 pA
25 pA
CQ D-AP5
abab
ab ab
NMDAR EPSCs
AMPAR EPSCs
200
NS
NS
NS
**
**
NS
100
0
200
100
0
ab ab
ab ab
a
aMin
0 20406080
25 ms
25 ms
25 ms
1 mV
1 mV
1 mV
+ SU6656
CQ
WT
KO
Figure 5 | CQ treatment activates NMDARs through postsynaptic Src activation. (a–c)CQ(4mM, 20 min) fails to enhance NMDAR function at
Shank2 /SC-CA1 synapses in the presence of PP2 or SU6656, specific inhibitors of SFKs, but effectively enhances NMDAR function in the presence of
PP3, an inactive PP2 analogue, as measured by NMDA-fEPSPs. Hippocampal slices were bath incubated with PP2 (10 mM) or SU6656 (10 mM) throughout
recordings. (PP2, n¼7 slices (6 animals) for WT and 7 (5) for KO; SU6656, n¼7 (5) for WT and 7 (4) for KO; PP3, n¼8 (4) for WT and 8 (5) for KO;
*Po0.05, **Po0.01; Student’s t-test). (d,e) CQ fails to enhance NMDAR function at Shank2 /SC-CA1 synapses (e) in the presence of Src(40–58),
a specific peptide inhibitor of Src, but effectively enhances NMDAR function in the presence of sSrc(40–58), a scrambled version of the peptide, as
determined by simultaneous measurements of NMDA- and AMPA-eEPSCs at 40mV. NMDA-eEPSCs were measured at 60 ms after the stimulation. The
labels a and b indicate 5-min duration before and after CQ, respectively. (Src(40–58), n¼5 cells (4 animals) for WT and 6 (4) for KO; sSrc(40–58), n¼7
(5) for WT, 7 (6) for KO; *Po0.05, **Po0.01; Student’s t-test, compared with the 5-min duration before CQ). Data in all panels with error bars represent
mean±s.e.m. NS, not significant.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168 ARTICLE
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 7
&2015 Macmillan Publishers Limited. All rights reserved.
inhibition of MAPKK/MEK in the upstream had no effect on
CQ-dependent NMDAR activation (Supplementary Fig. 9a).
Intriguingly, CaMKIIainhibition caused a small reduction in
the levels of CQ-induced NMDAR activation after but not during
CQ treatment (Supplementary Fig. 9b), suggesting that CaMKIIa
is required for the maintenance of enhanced NMDAR function.
These results collectively suggest that the CQ-induced increase in
NMDAR function is dependent on Src, and imply that the
subcellular site of Src activation is postsynaptic.
CQ treatment improves social interaction in Tbr1 þ/mice.
Finally, we considered whether CQ could rescue social interaction
deficits in other mouse models of ASD in which reduced
NMDAR is associated with autistic-like behaviours. One
such model is the recently developed Tbr1-haploinsufficient
(Tbr1 þ/) mouse, which has been reported to display a
reduction in social interaction that is normalized by the NMDAR
agonist D-cycloserine50. We thus tested whether CQ could also
rescue social interaction in these mice.
Tbr1 þ/mice showed reduced social interaction in the three-
chamber test, compared with WT mice, and acutely injected CQ
(30 mg kg 1), administered 3 h before the three-chamber test,
rescued the social interaction deficits of Tbr1 þ/mice, with no
effect on WT mice, as determined by time spent in exploration/
sniffing, time spent in chamber and the preference index derived
from exploration or chamber time (Fig. 7a–c and Supplementary
Fig. 10a–d). The positive rescue based on chamber time is
supported by the strong correlation between exploration time and
chamber time observed in Tbr1 þ/mice (Supplementary
Fig. 11). CQ, however, did not affect social novelty recognition
in Tbr1 þ/or WT mice, determined based on the preference for
a new stranger mouse relative to a previously encountered
stranger mouse (Fig. 7d–f and Supplementary Fig. 10e–h). These
results suggest that CQ rescues social interaction deficits, but has
no effect on social novelty recognition, in Tbr1 þ/mice. Taken
together with similar results obtained in Shank2 /mice, this
suggests that CQ is capable of rescuing social interaction deficits
in two independent mouse models of ASD characterized by
reduced NMDAR function.
CQ restores NMDAR function at Tbr1 þ/amygdalar synapses.
We hypothesized that the CQ-dependent social rescue in
Tbr1þ/mice might be associated with the restoration of reduced
NMDAR function at Tbr1 þ/synapses. We first examined
synaptic transmission at Tbr1þ/synapses in the hippocampus,
where reduced NMDAR function was observed in Shank2/
mice. However, no significant differences could be observed in the
electrophysiological parameters, including miniature EPSCs
(mEPSCs) in CA1 pyramidal neurons, and input–output ratio,
paired-pulse ratio and NMDA/AMPA ratio at SC-CA1 synapses
(Supplementary Fig. 12).
In contrast, principal neurons in the lateral amygdala (LA), a
brain region also enriched with glutamate- and Zn-releasing
neurons62, showed slightly increased mEPSC amplitude but not
frequency, without a change in the input–output ratio (Fig. 8a,b).
Importantly, thalamic-LA Tbr1 þ/synapses showed a reduction
63
63
63
63
63
63
Active, p-Src(416) Inactive, p-Src(527)
**
*
***
NS
NS
WT-V
WT-C
WT-V + PP2
WT-C + PP2
WT-V
WT-C
WT-V + PP2
WT-C + PP2
KO-V
KO-C
KO-C + PP2
KO-V+ PP2
WT-V
WT-C
WT-V + PP2
WT-C + PP2
NS
2.0
1.5
1.0
0.5
0.0
Normalized ratio of
phospho-Src / total Src
Normalized ratio of
phospho-Src / total Src
2.0
1.5
1.0
0.5
0.0
Active, p-Src(416) Inactive, p-Src(527)
***
*
**
NS
NS
KO-V
KO-C
KO-V + PP2
KO-C + PP2
KO-V
KO-C
KO-V + PP2
KO-C + PP2
NS
2.0
1.5
1.0
0.5
0.0
Normalized ratio of
phospho-Src / total Src
Normalized ratio of
phospho-Src / total Src
2.0
1.5
1.0
0.5
0.0
Src
p-Src(Y416)
p-Src(Y527)
α-Tubulin
α-Tubulin
α-Tubulin
Src
p-Src(Y416)
p-Src(Y527)
α-Tubulin
α-Tubulin
α-Tubulin
Figure 6 | CQ increases Src tyrosine phosphorylation in the hippocampus. (a–f) Treatment of WT (a–c) and Shank2 /(d–f) hippocampal slices
(3–5 weeks) with CQ in ACSF supplemented with 250 nM Zn for 20min in the presence or absence of PP2 (SFK inhibitor) were followed by immunoblot
analysis of total and tyrosine phosphorylated Src (Y416 and Y527). For quantification, levels of Src tyrosine phosphorylation were normalized to total
Src levels. 250 nM Zn was added to maximize the visualization of the changes occurring in Src tyrosine phosphorylation. (n¼6 slices (3 mice) for
WT-V, WT-C, KO-V and KO-C; *Po0.05, **Po0.01, ***Po0.001; one-way analysis of variance with Tukey’s post hoc analysis). Data in all panels with error
bars represent mean±s.e.m. NS, not significant.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168
8NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications
&2015 Macmillan Publishers Limited. All rights reserved.
in the NMDA/AMPA ratio, which was normalized by CQ
treatment (Fig. 8c,d). CQ did not cause a significant increase in
the NMDA/AMPA ratio in WT mice. These results suggest that
Tbr1 heterozygosity causes NMDAR hypofunction selectively
in the amygdala, and its normalization is associated with the
CQ-dependent social rescue in Tbr1 þ/mice.
Discussion
In the present study, we found that trans-synaptic Zn mobiliza-
tion improves social interaction in two distinct mouse models of
ASD through postsynaptic Src and NMDAR activation.
Our study suggests that CQ-dependent mobilization of Zn
from pre- to postsynaptic sites—not Zn removal after chelation—
might be useful in the treatment of ASDs. This unique trans-
synaptic Zn mobilization is supported by the following findings:
(1) CQ failed to enhance NMDAR function in ZnT3 /mice,
which lack the presynaptic Zn pool; and (2) Ca-EDTA, a
membrane-impermeable Zn chelator that should chelate Zn in
the synaptic cleft or extracellular sites, blocked CQ-dependent
NMDAR activation.
CQ can bind Cu and Fe in addition to Zn. However, CQ
appears to exert its effects through Zn interaction. A previous
study has shown that CQ can mobilize Zn and Cu but not Fe into
cytoplasmic sites in neuroblastoma cells63. In addition, our study
shows that cuprizone, a specific Cu chelator, does not block
CQ-dependent NMDAR activation. Regarding the potential
involvement of the ‘chelating’ activity of CQ, as opposed to the
‘ionophoric’ activity, we suspect it is unlikely because CQ
treatment of WT mice for 2 h did not lead to the reduction of
Zn, Cu or Fe in the brain. However, it should be pointed out that
the proposed Zn mobilization by CQ should involve Zn chelation
at presynaptic sites before Zn ions are mobilized to postsynaptic
sites through ionophoric effects.
We propose a specific mechanism that may underlie the
CQ-dependent social rescue, namely NMDAR activation through
postsynaptic Src. In support of this, CQ-dependent NMDAR
activation at Shank2 /hippocampal synapses is blocked by
two independent inhibitors of SFKs (PP2 and SU6656), as well as
the Src-inhibitory peptide Src(40–58), which acts in the
postsynaptic compartments when applied through patch pipettes.
In addition, CQ treatment increases Src phosphorylation at Y416
but not Y527. We initially expected that the phosphorylation at
Src Y527, which keeps Src at an inactive conformation, may be
reduced to activate Src because Y527 is the substrate of Zn-
inhibited C-terminal Src kinase59. On the contrary, we found an
increase in the phosphorylation of Src Y416, which is known to
render Src fully active. Although further details remain to be
studied, CQ appears to promote full activation of Src after its
initial activation by some other mechanisms.
Our study does not exclude the possibility that postsynaptic Zn
acts on targets other than SFKs and NMDARs. For instance,
a previous study has shown that Zn enhances excitatory
synaptic stabilization of Shank2/3 and synapse formation and
maturation30. Because two independent inhibitors of SFKs
and the Src-inhibitory peptide significantly blocked the
WT vehicle
OS1OS1
OS1OS1
S1 S2 S1 S2
S1 S2 S1 S2
HT vehicle
WT vehicle
HT vehicle
WT clioquinol
HT clioquinol
HT clioquinol
HT clioquinol
Exploration time (s)
Time spent (s)
250
200
300
150
100
50
0
OS1 OS1 OS1 OS1
100
50
–50
0
NS
NSNS
NS
NS
NS
***
*** ***
***
*
**
**
Preference index (S2-S1)
from exploration time
Preference index (S1-O)
from exploration time
NS
*
100
50
–50
–100
0
Exploration time (s)
Time spent (s)
80
40
0
80
40
0
250
200
300
150
100
50
0
WT-V
WT-C
HT-C
HT-V
WT-V
WT-C
HT-C
HT-V
S1 S2 S1 S2 S1 S2S1 S2
WT-V
WT-C
HT-V
HT-C
WT-V
WT-C
HT-V
HT-C
Figure 7 | CQ treatment improves social interaction in Tbr1 þ/mice. (a–f) CQ improves social interaction (a–c) but not social novelty recognition
(d–f)inTbr1 þ/(HT) mice, but has no effect in WT mice, as measured by the time spent in exploring targets and the social preference index derived from
these results. Heat maps in aand drepresent examples of mouse movements. (n¼10 for WT-V and WT-C, and 11 for HT-V and HT-C; *Po0.05, **Po0.01,
***Po0.001; two-way analysis of variance (ANOVA) and one-way ANOVA with Tukey’s post hoc test). Data in all panels with error bars represent
mean±s.e.m. NS, not significant.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168 ARTICLE
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 9
&2015 Macmillan Publishers Limited. All rights reserved.
CQ-dependent NMDAR activation (Fig. 5), we did not explore
this possibility. It should be noted, however, that the increase in
CQ-induced NMDA-fEPSPs caused by additional Zn (250 nM)
was greater at WT synapses than at Shank2 /synapses
(Fig. 3d,e). This suggests that postsynaptic Shank2 may
contribute to Zn-dependent NMDAR activation during high
levels of neuronal activity and thus may also contribute to the
aetiology of ASDs.
Previous studies have reported the roles of CQ in the regulation
of synaptic transmission, including those from Dr Takeda’s
group57,58. The latter studies suggest that an increase in
intracellular Zn concentration in the postsynaptic side of CA1
synapses inhibits NMDAR-dependent LTP, which apparently
differ from our results that CQ has no effect on LTP at WT
SC-CA1 synapses, and that CQ enhances LTP at Shank2/
SC-CA1 synapses (Fig. 2b). However, because Shank2/
synapses display substantial alterations in synaptic protein
composition/modification and synaptic transmission/plasticity45,
the CQ-dependent NMDAR activation at Shank2/synapses
cannot be compared with the previous results. Regarding the effect
of CQ on LTP at WT synapses, the difference might stem from
that (1) previous studies used rat hippocampal slices, whereas we
used mouse ones, and (2) previous studies prepared brain slices 2h
after i.p. injection of Zn-CQ (9.8 mmol (B3mg) kg1)and
measured LTP without CQ in the artificial CSF (ACSF), whereas
we introduced CQ (4 mM) directly to slices obtained from CQ-
untreated mice during electrophysiological measurements.
Notably, an independent paper has also reported that bath
application of CQ (4 mM) has no effect on LTP at SC-CA1
synapses in the mouse brain, similar to our data63. To address this
issue directly, we mimicked the method of CQ treatment reported
in Dr Takeda’s group, preparing mouse hippocampal slices 2 h
after i.p. i njection of Zn-CQ (9.8 mmol kg1). However, we found
no difference between treated and untreated groups in LTP
induced by high-frequency stimulation at SC-CA1 synapses (n¼7
slices, 4 mice for vehicle and 7, 5 for CQ; data not shown).
Therefore, the differences may be attributable to the fact that
different animal species were used.
CQ-dependent rescue of social deficits in Shank2 /and
Tbr1 þ/mice was associated with CQ-dependent elevation of
NMDAR function, further supporting the hypothesis that
NMDAR hypofunction may underlie ASDs. This concept was
put forward based on the observations that D-cycloserine
improves ASD symptoms in humans and autistic-like phenotypes
in animals (reviewed in ref. 64), although further studies are
needed to verify this hypothesis.
CQ rescues social interaction in Shank2 /and Tbr1 þ/
mice, but it fails to rescue social novelty recognition, repetitive
behaviour, hyperactivity or anxiety-like behaviour in Shank2 /
mice. This is reminiscent of the previous result that NMDAR
activation by D-cycloserine selectively rescues social interaction in
Shank2 /mice45. This selective rescue might be attributable to
the different nature of the circuits associated with these
behaviours, where some are reversible, or at least treatable,
whereas others are not. In line with this, NMDARs are involved
in the regulation of both neuronal development and synaptic
transmission/plasticity/signalling11,65. In addition, activity-
dependent sculpting of neuronal circuits associated with ASDs
has critical time windows66.
Finally, our study broadens the therapeutic potential of CQ.
CQ has been used as a topical antiseptic or an oral intestinal
amoebicide since 1930s, although the latter use has ceased for its
controversial association with subacute myelo-optic neuropa-
thy52. Recently, however, CQ-dependent chelation of Zn has been
Cumulative probability
NMDA/AMPA ratio
Cumulative probability
1.0
0.8
0.6
0.4
0.2
0.0
0.8
0.6
0.4
0.2
0.0
400
300
200
100
0
1.0
0.8
0.6
0.4
0.2
0.0
AMPA-mEPSC
amplitude (pA)
Inter-event interval (s)
0102030
Stimulus (μA)
0 5 15 2010
WT HT
WT HT
3
2
1
0
0
10
20
30
40 **
1 s
20 pA
HT HT
100 pA
50 ms WT
HT
WTWT
NS
Frequency (Hz)
Amplitude (pA)
Amplitude (pA)
0 20406080
NMDA/AMPA ratio
1.0
1.2
0.8
0.6
0.4
0.2
0.0
WT-V
WT-C
HT-C
HT-V
100 pA
NS *
25 ms
HT
HT
*
WT
WT
HT-V
WT-V WT-C
HT-C
100 pA
25 ms
Figure 8 | CQ restores NMDAR function at Tbr1 þ/amygdalar thalamic-LA (T-LA) synapses. (a) CQ increases mEPSC amplitude but not frequency in
Tbr1 þ/principal neurons in the LA (4–6 weeks). (n¼17 cells (3 animals) for WT and HT, **Po0.01, Student’s t-test). (b) CQ has no effect on the input–
output ratio at Tbr1 þ/T-LA synapses (4–6 weeks), as indicated by plots of fEPSP slopes against stimulus intensities. Representative current traces are an
average of three consecutive responses with input stimulations of 25 mA. (n¼8 cells (3 animals) for WT and HT; Student’s t-test). (c) Reduced NMDA/
AMPA ratio at Tbr1þ/T-LA synapses (4–6 weeks). (n¼8 cells (5 animals) for WT and 9 (5) for HT; *Po0.05; Student’s t-test). (d) CQ restores the
NMDA/AMPA ratio at Tbr1 þ/T-LA synapses but has no effect on WT synapses (4–6 weeks). (n¼10 (5) for WT-V, 7 (4) for WT-C, 10 (6) for HT-V and
9 (4) for HT-C; *Po0.05; two-way analysis of variance (ANOVA) and one-way ANOVA with Tukey’s post hoc test). Data in all panels with error bars
represent mean±s.e.m. NS, not significant.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168
10 NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications
&2015 Macmillan Publishers Limited. All rights reserved.
suggested for the treatment of neurological disorders including
Alzheimer’s disease67, Parkinson’s disease68 and Huntingtons’
disease69. Moreover, PBT2, a second-generation CQ-related
compound under clinical trials, seems to be safe and improve
cognitive deficits in patients with Alzheimer’s disease67.
Therefore, our study is the first to demonstrate the possibility
of repositioning of the FDA-approved antibiotic, CQ, to ASDs
based on a novel mechanism distinct from chelation. In addition,
CQ-dependent trans-synaptic Zn mobilization might also be
useful in other psychiatric disorders that are notable for being
caused by a decrease in NMDAR function70.
In conclusion, our study suggests that trans-synaptic Zn
mobilization rapidly improves social interaction in two indepen-
dent mouse models of ASD through Src and NMDAR activation,
and a new therapeutic potential of CQ in the treatment of ASDs.
Methods
Mice.Shank2 /mice and Tbr1 þ/mice have been reported45,50. All mice
were backcrossed to a C57BL/6 background for more than 20 generations, and
housed and bred in a mouse vivarium at the Korea Advanced Institute of Science
and Technology (KAIST; Shank2 /mice) and the Academia Sinica (Tbr1 þ/
mice). For breeding of Shank2 mice, we used a scheme of heterozygous (HT) HT
to produce littermate pairs of WT and KO mice. To breed Tbr1 mice, we used
offspring from HT malesWT females. Other combinations did not yield any
differences in breeding efficiency or behavioural phenotypes. Pups were kept with
the dam until weaning at postnatal day 21. After weaning, animals were housed in
mixed-genotype groups of 3–5 mice per cages, and randomly subjected to
electrophysiological and behavioural experiments. Animals at 3–5 weeks of age
were used for electrophysiological experiments and two-photon imaging; male
animals at 2–4 months of age were used for behavioural assays. For TFL-Zn
staining, male animals at 8 weeks (for Shank2 /mice) were used. WT
littermates were used as controls.
ZnT3 /mice, reported previously54, were maintained in the KAIST animal
facility. These mice also were backcrossed to the C57BL/6 background for more
than 10 generations. Both male and female animals at 3–5 weeks were used for
electrophysiological experiments and TFL-Zn staining (P23).
All mice were bred and maintained according to the KAIST and Academia Sinica
Animal Research Requirements, and all procedures were approved by the
Committees of Animal Research at KAIST, and at Academia Sinica. Mice were fed
ad libitum by standard rodent chow and tap water, and housed under 12 h light/dark
cycle (lights off at 1900 hours in KAIST and at 2000 hours in Academia Sinica).
Clioquinol.CQ (Calbiochem) was dissolved in dimethylsulfoxide (DMSO; Sigma)
and polyethylene glycol (Aldrich; DMSO:polyethylene glycol ¼1: 9) to a final
concentration of 20 g l 1. WT and Shank2 /mice (or Tbr1 þ/mice) received
acute i.p. injection of CQ (30 mg kg 1) or the same volume of DMSO-polyethylene
glycol mixture. The injection was performed 2 h before (Shank2/mice and WT
littermates) or 3 h before (Tbr1 þ/mice and WT littermates) behavioural assays
at the discretion of the facility.
Drug treatment scheme.We devised a within-subjects design with a 1-week
washout period (Supplementary Fig. 1a), and divided animals into two groups, vehicle-
first and CQ-first group, to rule out carryover effects. Each mouse received a single
acute dose of vehicle or CQ, and underwent a single behavioural task, one task per
week. Testing was conducted in dedicated behavioural test rooms during the light phase
(three-chamber test) and the dark phase (repetitive behaviours and open-field test).
Three-chamber social interaction assay.The three-chamber social interaction
assay for Shank2 mice (Shank2 /and WT littermates) and Tbr1 mice (Tbr1 þ/
and WT littermates) were performed45,50. In short, the assay consisted of three
phases of 10 min duration: habituation, social interaction (stranger 1 versus object)
and social novelty recognition (stranger 1 versus stranger 2). Exploration was
defined as instances in which WT or mutant mouse tries to sniff object/stranger, or
orients its nose towards and come close to object/stranger. Individual movement
tracks were analysed by Ethovision 10.0 (Noldus) and modified by custom-
designed software MatLab (MathWorks) to generate heat maps. Time spent in
exploration was analysed by the researcher who was blinded to the subject
genotype (in Shank2 mice), or by using the Smart Video Tracking System (Panlab,
in Tbr1 mice). In addition to exploration time, we used the preference index, which
represents a numerical difference between time spent exploring the targets
(stranger 1 versus object or stranger 2 versus stranger 1) divided by total time spent
exploring both targets45.
Repetitive behaviours.Shank2 /mice and their WT littermates in their home
cages without bedding were used to measure times spent in repetitive behaviours,
including jumping and grooming during 10 min. Jumping was defined as the
behaviour of a mouse where it rears on its hind legs at the corner of the cage, or
along the side walls, and jumps so that the two hind legs are simultaneously off the
ground. Grooming behaviour was defined as stroking or scratching of face, head or
body with the two forelimbs, or licking body parts45. The experiments and analyses
were performed independently in a blind manner.
Open-field test.The size of the open-field box was 40 40 40 cm, and the
centre zone line was 13.3 cm apart from the edge. Mice were placed in the centre in
the beginning of the test, and mouse movements were recorded with a video
camera for 60 min, and were analysed by Ethovision 10.0 (Noldus).
Electrophysiology.For hippocampal electrophysiological experiments, sagittal
hippocampal slices (400 mm thick for extracellular and 300 mm thick for
intracellular recordings) of the mutant mice (Shank2 /mice, Tbr1 þ/mice or
ZnT3 /mice) and their WT littermates were prepared using a vibratome
(Leica VT1200) in ice-cold dissection buffer containing (in mM) 212 sucrose,
25 NaHCO
3
, 5 KCl, 1.25 NaH
2
PO
4
, 0.5 CaCl
2
, 3.5 MgSO
4
,10D-glucose,
1.25 L-ascorbic acid and 2 Na-pyruvate bubbled with 95% O
2
/5% CO
2
. CA3 was
removed to prevent epileptiform activity. For amygdalar electrophysiological
experiments, coronal slices (300 mm) including the LA of Tbr1 þ/mice and their
WT littermates were cut. The slices were recovered at 32 °C for 1 h in normal ACSF
(in mM: 124 NaCl, 2.5 KCl, 1 NaH
2
PO
4
, 25 NaHCO
3
, 10 glucose, 2 CaCl
2
and
2 MgSO
4
oxygenated with 95% O
2
/5% CO
2
). For the recording, a single slice was
moved to and maintained in submerged-type chamber at 28 °C, continuously
perfused with ACSF (2ml min 1) saturated with 95% O
2
/5% CO
2
. Stimulation
and recording pipettes were pulled from borosilicate glass capillaries (Harvard
Apparatus) using a micropipette electrode puller (Narishege).
For extracellular recordings, mouse hippocampal slices at the age of postnatal day
21–35 were used. fEPSPs were recorded in the stratum radiatum of the hippocampal
CA1 region using pipettes filled with ACSF (1 MO). fEPSP was amplified
(Multiclamp 700B, Molecular Devices) and digitized (Digidata 1440A, Molecular
Devices) for measurements. The Schaffer collateral pathway was stimulated every
20 s with pipettes filled with ACSF (0.3–0.5 MO). The stimulation intensity was
adjusted to yield a half-maximal response, and three successive responses were
averaged and expressed relative to the normalized baseline. To induce LTP, high-
frequency stimulation (100Hz, 1 s) was applied after a stable baseline was acquired.
CQ (4 mM) was bath applied before and after LTP induction during the whole
experimental processes. To isolate NMDAR-mediated fEPSPs, we used ACSF
containing 2 mM calcium, 0.1 mM magnesium and 6,7-dinitroquinoxaline-2,3-dione
(10 mM, DNQX, Tocris), which inhibits AMPAR-mediated EPSPs.
Whole-cell patch-clamp recordings of hippocampal CA1 pyramidal neurons,
and of LA principal neurons in the dorsolateral division were made using a
MultiClamp 700B amplifier (Molecular Devices) and Digidata 1440A (Molecular
Devices). During whole-cell patch-clamp recordings, series resistance was
monitored each sweep by measuring the peak amplitude of the capacitance
currents in response to sho rt hyperpolarizing step pulse (5mV, 40ms); only cells
with a change in o20% were included in the analysis. For afferent stimulation of
hippocampal pyramidal neurons, the Schaffer collateral pathway was selected,
while for that of LA, the thalamic afferent pathway was stimulated. For LA
electrophysiology, brain slices were selected based on the presence of a well-
isolated, sharply defined trunk (containing thalamic afferents) crossing the
dorsolateral division of the LA, which is a site of convergence of somatosensory and
auditory inputs. For NMDA/AMPA ratio experiments, mouse hippocampal slices
(P17–P25) and LA slices (4–6 weeks old) were used. The recording pipettes
(2.5–3.5 MO) were filled with an internal solution containing the following (in
mM): 100 CsMeSO
4
, 10 TEA-Cl, 8 NaCl, 10 HEPES, 5 QX-314-Cl, 2 Mg-ATP, 0.3
Na-GTP and 10 EGTA, with pH 7.25, 295 mOsm). CA1 pyramidal neurons and LA
principal neurons were voltage clamped at 70 mV, and EPSCs were evoked at
every 15 s. AMPAR-mediated EPSCs were recorded at 70 mV, and 30
consecutive responses were recorded after stable baseline. After recording AMPAR-
mediated EPSCs, holding potential was changed to þ40 mV to re cord NMDAR-
mediated EPSCs. NMDA component was measured at 60 ms after the stimulation.
The NMDA/AMPA ratio was determined by dividing the mean value of 30 NMDA
components of EPSCs by the mean value of 30 AMPAR-mediated EPSC peak
amplitudes. Somatic whole-cell recording of mEPSCs were obtained in amygdalar
principal neurons at a holding potential of 70 mV. TTX (1mM) and picrotoxin
(100 mM) were added to ACSF to inhibit spontaneous action potential-mediated
synaptic currents and IPSCs, re spectively. CQ (4 mM) was bath applied from
20 min before and during the whole period of NMDA/AMPA ratio recording. For
measuring AMPAR-mediated and NMDAR-mediated EPSCs together upon CQ
treatment, pyramidal neurons were voltage clamped at 40 mV, and EPSCs were
evoked at every 15 s. AMPAR-mediated EPSC was determined as a peak amplitude
of EPSC, and NMDAR-mediated EPSC as a component at 60 ms after stimulation.
NMDA/AMPA ratio at 40 mV was calculated by using both values, and
monitored during the experimental process. Src-inhibiting peptide, Src(40–58), and
its analogue, scrambled Src(40–58; ref. 61), were purchased from Peptron, and
introduced into the internal solution at a concentration of 0.03 mg ml1to observe
whether Src-inhibition affects the action of CQ . Picrotox in (100 mM) was always
added to ACSF to block GABAA receptor-mediated currents.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168 ARTICLE
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 11
&2015 Macmillan Publishers Limited. All rights reserved.
Data were acquired by Clampex 10.2 (Molecular Devices) and analysed by
Clampfit 10 (Molecular Devices). Drugs were purchased from Tocris (DNQX,
TPEN, PP2, PP3 and D-AP5), Abcam (PD98059) and Sigma (KN93, picrotoxin,
Ca-EDTA, cuprizone and SU6656).
ZnAF-2DA imaging.For intracellular Zn staining, two groups of WT mice (3–5
weeks) were injected with vehicle (DMSO) or CQ (30 mg kg1) 2 h before imaging.
For two-photon imaging, sagittal hippocamp al slices (300 mm thick) were
immersed in 10 mM ZnAF-2DA (Enzo Life Sciences) in ACSF for 40 min, followed
by 1 h washout with ACSF. Zn signals were measured by using a multiphoton laser
scanning microscope system LSM 7 MP (Carl Zeiss). A total of B200 Z stack
images (0.7 mm interval; B130–180 mm total depth) were captured from each slice.
For quantification of Zn signals, a total of 11 images at every 4.5 mm depth from the
surface (fromB5to50mm depth) were selected. Three regions of interest (ROIs;
squares of 33 mm2) in the CA1 cell body area, or in the CA1 dendritic field, were
analysed of Zn fluorescence signals using MetaMorph (Molecular Devices).
TFL-Zn staining.Without fixation, brain sections (10 mm thick) were stained with
the Zn-specific dye TFL-Zn (N-(6-methoxy-8-quinolyl)-p-carbox-
ybenzoylsulfonamide (0.1 mM), Calb iochem) dissolved in phosphate-buffered
saline (pH 7.2), and photographed with a digital camera linked to a fluorescence
microscope (Olympus IX71; excitation, 330–385 nm; dichromatic, 400 nm; and
barrier, 420 nm). Fluorescence signals were obtained using an image prog ramme
(Image-Pro Insight, Media Cybernetics, Silver Spring, MD). Images of five con-
secutive hippocampal slices from an individual brain were quantified using
MetaMorph (Molecular Devices). ROIs were defined as three to five squares
(50 50 mm2) in WT and Shank2 /hippocampal DG, CA3 and CA1 regions (5,
3 and 5 squares, respectively). For quantification, the total fluorescence from a
Shank2 /ROI was normalized to that from an equivalent WT ROI.
Src immunoblot analysis.For immunoblotting of Src proteins, WT and
Shank2 /sagittal hippocampal slices (400mm thick; 3–5 weeks) were prepared
using a vibratome (Leica VT1200). After recovery at 37 °C for 1 h in normal ACSF
supplemented with 250 nM ZnCl
2
, slices were treated with CQ, or vehicle (DMSO),
for 20 min, in the presence or absence of PP2. After treatments, the slices were
homogenized in 100 ml ice-cold homogenization buffer (0.32 M sucrose, 10 mM
HEPES, pH 7.4, 2 mM EDTA, protease inhibitors and phosphatase inhibitors) per
each slice, and subjected to immunoblot analysis and quantification using Odyssey
Fc Imaging System (LI-COR). The following antibodies were purchased: Src,
phosphor-Src (Tyr-416; 1:1,000 dilution) and phosphor-Src (Tyr-527; Cell Sig-
naling; 1:1,000 dilution). Full-size immunoblot images for Fig. 6 are shown in
Supplementary Fig. 13.
Crude synaptosomes.Crude synaptosomes from Shank2 /mice were pre-
pared as described45. Briefly, mouse brains (2 months old) were homogenized in
ice-cold homogenization buffer (0.32 M sucrose, 10 mM HEPES, pH 7.4, 2mM
EDTA, protease inhibitors and phosphatase inhibitors). The homogenates were
centrifuged at 900gfor 10 min. The resulting supernatant was centrifuged again at
12,000gfor 15 min. The pellet was resuspended in homogenization buffer and
centrifuged at 13,000gfor 15 min (the resulting pellet is P2; crude synaptosomes).
This sample was immunoblotted with ZnT3 antibodies (SYSY).
Metal analysis.WT and Shank2 /mice (2–3 months) were treated with CQ
(30 mg kg 1), or DMSO, by i.p. injection 2 h before brain preparation and sub-
sequent whole-brain metal analysis by inductively coupled plasma mass
spectrometry.
Statistical analysis.We randomly performed all the behaviour experiments,
ZnAF-2DA and TFL-Zn imaging, and metal analysis by researchers blind to the
identity of the animals, and analysed the data in a blind manner. Data collection
and analysis for slice electrophysiology and immunoblotting were performed
randomly, but not blind to the conditions of the experiments. Statistical analyses
were performed using Prism GraphPad (version 6.05) software, and details of the
results are described Supplementary Data 1. No statistical methods were used to
predetermine sample sizes, but our sample sizes are similar to those reported in
previous related publications45,50. An outlier was defined as a value outside the
mean±3 s.d.
References
1. Huguet, G., Ey, E. & Bourgeron, T. The genetic landscapes of autism spectrum
disorders. Annu. Rev. Genomics Hum. Genet. 14, 191–213 (2013).
2. Jeste, S. S. & Geschwind, D. H. Disentangling the heterogeneity of autism
spectrum disorder through genetic findings. Nat. Rev. Neurol. 10, 74–81 (2014).
3. Krumm, N., O’Roak, B. J., Shendure, J. & Eichler, E. E. A de novo convergence
of autism genetics and molecular neuroscience. Trends Neurosci. 37, 95–105
(2014).
4. Sudhof, T. C. Neuroligins and neurexins link synaptic function to cognitive
disease. Nature 455, 903–911 (2008).
5. Ting, J. T., Peca, J. & Feng, G. Functional consequences of mutations in
postsynaptic scaffolding proteins and relevance to psychiatric disorders. Annu.
Rev. Neurosci. 35, 49–71 (2012).
6. Jiang, Y. H. & Ehlers, M. D. Modeling autism by SHANK gene mutations in
mice. Neuron 78, 8–27 (2013).
7. Silverman, J. L., Yang, M., Lord, C. & Crawley, J. N. Behavioural phenotyping
assays for mouse models of autism. Nat. Rev. Neurosci. 11, 490–502 (2010).
8. Bagni, C. & Oostra, B. A. Fragile X syndrome: from protein function to therapy.
Am. J. Med. Genet. A 161A, 2809–2821 (2013).
9. Kleijer, K. T. et al. Neurobiology of autism gene products: towards pathogenesis
and drug targets. Psychopharmacology (Berl) 231, 1037–1062 (2014).
10. Darnell, J. C. & Klann, E. The translation of translational control by FMRP:
therapeutic targets for FXS. Nat. Neurosci. 16, 1530–1536 (2013).
11. Ebert, D. H. & Greenberg, M. E. Activity-dependent neuronal signalling and
autism spectrum disorder. Nature 493, 327–337 (2013).
12. Zoghbi, H. Y. & Bear, M. F. Synaptic dysfunction in neurodevelopmental
disorders associated with autism and intellectual disabilities. Cold Spring Harb.
Perspect. Biol. 4 (2012).
13. Ehninger, D. & Silva, A. J. Rapamycin for treating tuberous sclerosis and autism
spectrum disorders. Trends Mol. Med. 17, 78–87 (2011).
14. Ghosh, A., Michalon, A., Lindemann, L., Fontoura, P. & Santarelli, L. Drug
discovery for autism spectrum disorder: challenges and opportunities. Nat. Rev.
Drug Disc. 12, 777–790 (2013).
15. Persico, A. M. & Bourgeron, T. Searching for ways out of the autism maze:
genetic, epigenetic and environmental clues. Trends Neurosci. 29, 349–358
(2006).
16. Grabrucker, A. M. Environmental factors in autism. Front. Psychiatry 3, 118
(2012).
17. Grabrucker, A. M., Rowan, M. & Garner, C. C. Brain-delivery of zinc-ions as
potential treatment for neurological diseases: mini review. Drug Deliv. Lett. 1,
13–23 (2011).
18. Yasuda, H., Yoshida, K., Yasuda, Y. & Tsutsui, T. Infantile zinc deficiency:
association with autism spectrum disorders. Sci. Rep. 1, 129 (2011).
19. Grabrucker, S. et al. Zinc deficiency dysregulates the synaptic ProSAP/Shank
scaffold and might contribute to autism spectrum disorders. Brain 137,
137–152 (2014).
20. Russo, A. J. & Devito, R. Analysis of copper and zinc plasma concentration and
the efficacy of zinc therapy in individuals with Asperger’s syndrome, pervasive
developmental disorder not otherwise specified (pdd-nos) and autism. Biomark
Insights 6, 127–133 (2011).
21. Sensi, S. L., Paoletti, P., Bush, A. I. & Sekler, I. Zinc in the physiology and
pathology of the CNS. Nat. Rev. Neurosci. 10, 780–791 (2009).
22. Gundelfinger, E. D., Boeckers, T. M., Baron, M. K. & Bowie, J. U. A role for
zinc in postsynaptic density asSAMbly and plasticity? Trends Biochem. Sci. 31,
366–373 (2006).
23. Sensi, S. L. et al. The neurophysiology and pathology of brain zinc. J. Neurosci.
31, 16076–16085 (2011).
24. Vergnano, A. M. et al. Zinc dynamics and action at excitatory synapses. Neuron
82, 1101–1114 (2014).
25. Kim, T. Y., Hwang, J. J., Yun, S. H., Jung, M. W. & Koh, J. Y. Augmentation by
zinc of NMDA receptor-mediated synaptic responses in CA1 of rat
hippocampal slices: mediation by Src family tyrosine kinases. Synapse 46, 49–56
(2002).
26. Huang, Y. Z., Pan, E., Xiong, Z. Q. & McNamara, J. O. Zinc-mediated
transactivation of TrkB potentiates the hippocampal mossy fiber-CA3 pyramid
synapse. Neuron 57, 546–558 (2008).
27. Manzerra, P. et al. Zinc induces a Src family kinase-mediated up-regulation of
NMDA receptor activity and excitotoxicity. Proc. Natl Acad. Sci. USA 98,
11055–11061 (2001).
28. Sheng, M. & Kim, E. The shank family of scaffold proteins. J. Cell Sci. 113,
1851–1856 (2000).
29. Boeckers, T. M., Bockmann, J., Kreutz, M. R. & Gundelfinger, E. D. ProSAP/
Shank proteins—a family of higher order organizing molecules of the
postsynaptic density with an emerging role in human neurological disease.
J. Neurochem. 81, 903–910 (2002).
30. Grabrucker, A. M. et al. Concerted action of zinc and ProSAP/Shank in
synaptogenesis and synapse maturation. EMBO J. 30, 569–581 (2011).
31. Leblond, C. S. et al. Genetic and functional analyses of SHANK2 mutations
suggest a multiple hit model of autism spectrum disorders. PLoS Genet. 8,
e1002521 (2012).
32. Berkel, S. et al. Mutations in the SHANK2 synaptic scaffolding gene in
autism spectrum disorder and mental retardation. Nat. Genet. 42, 489–491
(2010).
33. Durand, C. M. et al. Mutations in the gene encoding the synaptic scaffolding
protein SHANK3 are associated with autism spectrum disorders. Nat. Genet.
39, 25–27 (2007).
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168
12 NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications
&2015 Macmillan Publishers Limited. All rights reserved.
34. Pinto, D. et al. Functional impact of global rare copy number variation in
autism spectrum disorders. Nature 466, 368–372 (2010).
35. Leblond, C. S. et al. Meta-analysis of SHANK mutations in autism spectrum
disorders: a gradient of severity in cognitive impairments. PLoS Genet. 10,
e1004580 (2014).
36. Moessner, R. et al. Contribution of SHANK3 mutations to autism spectrum
disorder. Am. J. Hum. Genet. 81, 1289–1297 (2007).
37. Verpelli, C. et al. Importance of Shank3 protein in regulating metabotropic
glutamate receptor 5 (mGluR5) expression and signaling at synapses. J. Biol.
Chem. 286, 34839–34850 (2011).
38. Berkel, S. et al. Inherited and de novo SHANK2 variants associated with autism
spectrum disorder impair neuronal morphogenesis and physiology. Hum. Mol.
Genet. 21, 344–357 (2011).
39. Arons, M. H. et al. Autism-associated mutations in ProSAP2/Shank3 impair
synaptic transmission and neurexin-neuroligin-mediated transsynaptic
signaling. J. Neurosci. 32, 14966–14978 (2012).
40. Bozdagi, O. et al. Haploinsufficiency of the autism-associated Shank3 gene
leads to deficits in synaptic function, social interaction, and social
communication. Mol. Autism 1, 15 (2010).
41. Kouser, M. et al. Loss of predominant Shank3 isoforms results in hippocampus-
dependent impairments in behavior and synaptic transmission. J. Neurosci. 33,
18448–18468 (2013).
42. Peca, J. et al. Shank3 mutant mice display autistic-like behaviours and striatal
dysfunction. Nature 472, 437–442 (2011).
43. Schmeisser, M. J. et al. Autistic-like behaviours and hyperactivity in mice
lacking ProSAP1/Shank2. Nature 486, 256–260 (2012).
44. Wang, X. et al. Synaptic dysfunction and abnormal behaviors in mice lacking
major isoforms of Shank3. Hum. Mol. Genet. 20, 3093–3108 (2011).
45. Won, H. et al. Autistic-like social behaviour in Shank2-mutant mice improved
by restoring NMDA receptor function. Nature 486, 261–265 (2012).
46. Yang, M. et al. Reduced excitatory neurotransmission and mild autism-relevant
phenotypes in adolescent Shank3 null mutant mice. J. Neurosci. 32, 6525–6541
(2012).
47. Wang, X., Xu, Q., Bey, A. L., Lee, Y. & Jiang, Y. H. Transcriptional and
functional complexity of Shank3 provides a molecular framework to
understand the phenotypic heterogeneity of SHANK3 causing autism and
Shank3 mutant mice. Mol. Autism 5, 30 (2014).
48. Zhu, L. et al. Epigenetic dysregulation of SHANK3 in brain tissues from
individuals with autism spectrum disorders. Hum. Mol. Genet. 23, 1563–1578
(2014).
49. Han, K. et al. SHANK3 overexpression causes manic-like behaviour with
unique pharmacogenetic properties. Nature 503, 72–77 (2013).
50. Huang, T. N. et al. Tbr1 haploinsufficiency impairs amygdalar axonal
projections and results in cognitive abnormality. Nat. Neurosci. 17, 240–247
(2014).
51. Chuang, H. C., Huang, T. N. & Hsueh, Y. P. Neuronal excitation upregulates
Tbr1, a high-confidence risk gene of autism, mediating Grin2b expression in
the adult brain. Front. Cell. Neurosci. 8, 280 (2014).
52. Bareggi, S. R. & Cornelli, U. Clioquinol: review of its mechanisms of action and
clinical uses in neurodegenerative disorders. CNS Neurosci. Ther. 18, 41–46
(2012).
53. Grabrucker, A. M. A role for synaptic zinc in ProSAP/Shank PSD scaffold
malformation in autism spectrum disorders. Dev. Neurobiol. 74, 136–146 (2014).
54. Cole, T. B., Wenzel, H. J., Kafer, K. E., Schwartzkroin, P. A. & Palmiter, R. D.
Elimination of zinc from synaptic vesicles in the intact mouse brain by
disruption of the ZnT3 gene. Proc. Natl Acad. Sci. USA 96, 1716–1721 (1999).
55. Izumi, Y., Auberson, Y. P. & Zorumski, C. F. Zinc modulates bidirectional
hippocampal plasticity by effects on NMDA receptors. J. Neurosci. 26,
7181–7188 (2006).
56. Frederickson, C. J. et al. Concentrations of extracellular free zinc (pZn)e in the
central nervous system during simple anesthetization, ischemia and
reperfusion. Exp. Neurol. 198, 285–293 (2006).
57. Takeda, A. Zinc signaling in the hippocampus and its relation to pathogenesis
of depression. J. Trace Elem. Med. Biol. 26, 80–84 (2012).
58. Takeda, A. et al. Transien t increase in Zn2 þin hippocampal CA1 pyramidal
neurons causes reversible memory deficit. PloS ONE 6, e28615 (2011).
59. Salter, M. W. & Kalia, L. V. Src kinases: a hub for NMDA receptor regulation.
Nat. Rev. Neurosci. 5, 317–328 (2004).
60. Yu, X. M., Askalan, R., Keil, 2nd G. J. & Salter, M. W. NMDA channel
regulation by channel-associated protein tyrosine kinase Src. Science 275,
674–678 (1997).
61. Lu, Y. M., Roder, J. C., Davidow, J. & Salter, M. W. Src activation in the
induction of long-term potentiation in CA1 hippocampal neurons. Science 279,
1363–1367 (1998).
62. Frederickson, C. J., Koh, J. Y. & Bush, A. I. The neurobiology of zinc in health
and disease. Nat. Rev. Neurosci. 6, 449–462 (2005).
63. Adlard, P. A. et al. Rapid restoration of cognition in Alzheimer’s transgenic
mice with 8-hydroxy quinoline analogs is associated with decreased interstitial
Abeta. Neuron 59, 43–55 (2008).
64. Lee, E. J., Choi, S. Y. & Kim, E. NMDA receptor dysfunction in autism
spectrum disorders. Curr. Opin. Pharmacol. 20C, 8–13 (2015).
65. Paoletti, P., Bellone, C. & Zhou, Q. NMDA receptor subunit diversity: impact
on receptor properties, synaptic plasticity and disease. Nat. Rev. Neurosci. 14,
383–400 (2013).
66. LeBlanc, J. J. & Fagiolini, M. Autism: a ‘critical period’ disorder? Neural. Plast.
2011, 921680 (2011).
67. Bush, A. I. The metal theory of Alzheimer’s disease. J. Alzheimers Dis.
33(Suppl 1): S277–S281 (2013).
68. Lei, P. et al. Tau deficiency induces parkinsonism with dementia by impairing
APP-mediated iron export. Nat. Med. 18, 291–295 (2012).
69. Nguyen, T., Hamby, A. & Massa, S. M. Clioquinol down-regulates mutant
huntingtin expression in vitro and mitigates pathology in a Huntington’s
disease mouse model. Proc. Natl Acad. Sci. USA 102, 11840–11845 (2005).
70. Coyle, J. T., Basu, A., Benneyworth, M., Balu, D. & Konopaske, G.
Glutamatergic synaptic dysregulation in schizophrenia: therapeutic
implications. Handb. Exp. Pharmacol. 267–295 (2012).
Acknowledgements
The work was supported by Academia Sinica and Ministry of Science and Technology,
Taiwan (MOST 103-2321-B-001-002 to Y.-P.H. and 102-2811-B-001-060 to T.-N.E.H.),
the NRF (National Research Foundation of Korea) grant funded by the Korean Gov-
ernment (NRF-2013-Fostering Core Leaders of the Future Basic Science Program to
C.C.) and the Institute for Basic Science (IBS-R002-D1 to E.K.).
Author contributions
H.L. performed immunoblot experiments and imaging analyses; C.C. performed NMDA-
fEPSP experiments; W.S. performed behavioural and Zn staining experiments; K.K.
performed behavioural experiments; T.-N.E.H. performed behavioural experiments for
Tbr1 mice. E.-J.L. performed the majority of electrophysiological analyses and all the
other experiments; and J.-Y.K., Y.-P.H. and E.K. supervised the project and wrote the
manuscript.
Additional information
Supplementary Information accompanies this paper at http://www.nature.com/
naturecommunications
Competing financial interests: The authors declare no competing financial interests.
Reprints and permission information is available online at http://npg.nature.com/
reprintsandpermissions/
How to cite this article: Lee, E. -J. et al. Trans-synaptic zinc mobilization improves social
interaction in two mouse models of autism through NMDAR activation. Nat. Commun.
6:7168 doi: 10.1038/ncomms8168 (2015).
This work is licensed under a Creative Commons Attribution 4.0
International License. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise
in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material.
To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8168 ARTICLE
NATURE COMMUNICATIONS | 6:7168 | DOI: 10.1038/ncomms8168 | www.nature.com/naturecommunications 13
&2015 Macmillan Publishers Limited. All rights reserved.
