Molecular Adaptations Underlying
Susceptibility and Resistance to
Social Defeat in Brain Reward Regions
Vaishnav Krishnan,1,5Ming-Hu Han,1,5Danielle L. Graham,1Olivier Berton,1William Renthal,1
Scott J. Russo,1Quincey LaPlant,1Ami Graham,1Michael Lutter,1Diane C. Lagace,1Subroto Ghose,1
Robin Reister,1Paul Tannous,2Thomas A. Green,1Rachael L. Neve,3Sumana Chakravarty,1Arvind Kumar,1
Amelia J. Eisch,1David W. Self,1Francis S. Lee,4Carol A. Tamminga,1Donald C. Cooper,1
Howard K. Gershenfeld,1and Eric J. Nestler1,*
1Departments of Psychiatry and Basic Neuroscience
2Department of Molecular Biology
The University of Texas Southwestern Medical Center (UTSWMC), 5323 Harry Hines Boulevard, Dallas, TX 75390-9070, USA
3Department of Genetics and McLean Hospital, Harvard University, Cambridge, MA 02478, USA
4Departments of Psychiatry and Pharmacology, Weill Medical College of Cornell University, New York, NY 10021, USA
5These authors contributed equally.
While stressful life events are an important
cause of psychopathology, most individuals
exposed to adversity maintain normal psycho-
logical functioning. The molecular mechanisms
underlying such resilience are poorly under-
stood. Here, we demonstrate that an inbred
population of mice subjected to social defeat
can be separated into susceptible and unsus-
ceptible subpopulations that differ along sev-
eral behavioral and physiological domains. By
a combination of molecular and electrophysio-
logical techniques, we identify signature adap-
tations within the mesolimbic dopamine circuit
that are uniquely associated with vulnerability
or insusceptibility. We show that molecular
recapitulations of three prototypical adapta-
tions associated with the unsusceptible pheno-
type are each sufficient to promote resistant
mechanisms of variations in stress resistance,
and illustrate the importance of plasticity within
the brain’s reward circuits in actively maintain-
ing an emotional homeostasis.
Anindividual’s emotional responseto severe, acute stress
(e.g., trauma, terrorist acts) or to more prolonged chronic
stress (e.g., divorce, war-time torture) is determined by
genetic and environmental elements that interact in com-
plex and poorly understood ways (Charney and Manji,
the effects of several kinds of acute and chronic stress on
an individual’s physiology and behavior, much less is
known about the biological basis of individual differences
in stress responses (Yehuda et al., 2006). A majority of
humans exposed to stressful events do not show signs
(PTSD) or depression (Charney, 2004; Kessler et al., 1995;
Yehuda, 2004). These ‘‘resilient’’ individuals (Hoge et al.,
2007) display traits such as cognitive flexibility (Yehuda
et al., 2006) and optimism (Charney, 2004). However, the
neural substrates and molecular mechanisms that medi-
ate resistance to the deleterious effects of stress remain
Insight into the biology of variations in susceptibility can
be gained by understanding models of individual differ-
ences in response to stress (Rutter, 2006). One such
rodent model is social defeat, which has the ethological
relevance of examining social subordination (Malatynska
and Knapp, 2005), as well as face validity in its ability to
model the symptomatology of stress-related disorders
like PTSD and depression (Avgustinovich et al., 2005;
Martinez et al., 1998). The development of social defeat
in mice has also enabled the examination of the effects
of specific genetic manipulations (McLaughlin et al.,
2006). By employing a novel measure of social interaction,
we recently showed that socially defeated mice demon-
strate a long-lasting social avoidance that is reversed by
chronic (but not acute) treatment with antidepressants
(Berton et al., 2006; Tsankova et al., 2006). Social avoid-
ance induced by chronic social defeat was dependent
on brain-derived neurotrophic factor (BDNF) signaling in
the mesolimbic dopamine circuit, which is composed of
dopamine neurons in the ventral tegmental area (VTA)
and their forebrain projection regions, in particular the
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 391
nucleus accumbens (NAc). This VTA-NAc circuit plays
a critical integrative role in reward- and emotion-related
behaviors (Nestler and Carlezon, 2006).
Here, we take advantage of a large variance in behav-
ioral outcomes after social defeat in inbred c57bl/6 mice
to study the molecular basis of susceptibility and resis-
tance to emotional stress. We show that resistance to
social defeat is latent, long lasting, extends across several
behavioral and physiological domains, and is mediated by
specific molecular neuroadaptations within the brain’s
mesolimbic dopamine reward circuit. We propose that
our findings may model resilience, operationally defined
as ‘‘the process of adapting well in the face of adversity’’
Segregation of Defeated Mice into Susceptible
and Unsusceptible Populations
An episode of social defeat is accomplished by forcing
a mouse to intrude into the space territorialized by a larger
mouse of a more aggressive genetic strain, leading to an
agonistic encounter that ultimately results in intruder sub-
ordination. We have previously shown that c57bl/6 mice
subjected to chronic social defeat (10 such defeats over
10 days) display a long-lasting reduction in social interac-
tion (Berton et al., 2006; Tsankova et al., 2006), which is
measured by comparing the time a mouse spends in an
interaction zone with a social target to the time in that
zone in the absence of a social target. By analyzing a large
tribution of responses: when examined 24 hr after the last
defeat (‘‘day 11’’), 40%–50% of defeated mice displayed
interaction scores similar to nondefeated controls. Be-
cause the vast majority of control mice spend more time
interacting with a social target than an empty target enclo-
sure, an interaction ratio of 100 (equal times in the pres-
mice with scores <100 were labeled ‘‘Susceptible’’ and
those with scores R100 were labeled ‘‘Unsusceptible’’
(Figure 1A). This latter group displayed median and vari-
ance values similar to controls (see Table S1 in the Sup-
plemental Data available with this article online).
Several further analyses support the validity of distinct
Susceptible and Unsusceptible subpopulations. A fre-
quency distribution histogram of absolute time spent
interacting with a social target (Figure 1B) and a two-di-
mensional scatterplot comparing interaction times to the
time spent in the corner zones of the arena (Figure S1B)
also revealed a segregation of Susceptible mice from
Unsusceptible and control mice. Figure 1C shows day
11 data from a representative experiment illustrating
how only Susceptible mice actively avoid the target by
spending more time in the corner zones. An awake behav-
ing social target is necessary for social avoidance, as we
observed a lack of avoidance to an anesthetized target
(Figure 1D). Differences in susceptibility cannot be
explained by locomotor behavior (equivalent in both sub-
groups; Figures S1C and S1D) or variations in aggression
during defeat (both groups sustained the same degree of
minor injuries). This avoidance phenotype was found to be
long-lasting: when Susceptible and Unsusceptible mice
were retested 4 weeks later, we observed a significant
correlation between day 11 and day 39 interaction ratios
(r = +0.61, p < 0.0001, n = 44).
Age, genotypic, or vendor differences cannot explain
this type of variance, because only 9-week-old c57bl/6
and Unsusceptible mice displayed similar predefeat inter-
action ratios and body weights (Figures S1E and S1F).
Because ‘‘risk-seeking’’ individuals are prone to stressful
life events (Charney and Manji, 2004), we examined a
between control, Susceptible, and Unsusceptible mice on
avariety ofopen-fieldmeasures, includingcenterduration
and total distance traveled (Figures S1G and S1H). Thus,
unsusceptibility appears to be a ‘‘latent’’ trait.
Susceptible and Unsusceptible Mice Display
To examine whether resistance to defeat-induced social
avoidance generalizes to other behavioral measures, we
performed an extensive phenotypic characterization of
Susceptible and Unsusceptible mice (summarized in
Table 1 and Table S2). On day 11, only Susceptible mice
displayed a significant decrease in body weight (Fig-
ure S2B) and sucrose preference (F2,33= 5.70, p < 0.01;
Figure 1E), both consistent with increased depression-
like behavior. In contrast, both Susceptible and Unsus-
spending significantly less time in the open arms of the
elevated plus maze (F2,76=5.23, p <0.01; Figure 1F). Sim-
ilarly, both subgroups of mice demonstrated a sensitized
corticosterone (CORT) response to a 6 min swim stress
(F2,35= 12.34, p < 0.0001; Figure 1G). To evaluate auto-
set of mice with subcutaneous temperature transponders
(Liu et al., 2003). Both Susceptible and Unsusceptible
mice showed an anticipatory form of autonomic arousal
during the course of social defeat: a significant elevation
of body temperature in the 30 min prior to the onset of
an expected defeat episode (Figure S2C). In contrast,
only Susceptible mice demonstrated a significant reduc-
tion in the circadian amplitude of temperature fluctuations
(F2,79= 3.21, p < 0.05; Figure 1H) and a significantly ele-
vated hyperthermic response to the social avoidance
test (F2,86= 5.30, p < 0.01; Figure 1I). Interestingly, only
Susceptible mice displayed significant conditioned place
preference to a low dose of cocaine (Figure S2D), demon-
strating sensitization to psychostimulant reward. Collec-
tively, these data show that the development of social
avoidance in Susceptible mice is associated with a syn-
drome of hedonic changes, weight loss, and circadian
abnormalities. In contrast, increases in anxiety and corti-
costerone reactivity are seen in both subgroups of mice.
392 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.
Figure 1. Identification of Susceptible and Unsusceptible Subgroups
(A) Horizontal scatterplot depicting the distribution of interaction ratios for control, Susceptible, and Unsusceptible mice over multiple social defeat
experiments (error bars represent mean ± interquartile range). (B) Frequency distribution histogram for the absolute time spent in interaction zone. (C)
In response to an aversive CD1 social target, only Susceptible mice avoid the interaction zone (F2,54= 33.20, p < 0.0001) and prefer the corner zones
(F2,54= 35.27, p < 0.0001). (D) Interaction zone times for the three groups of mice in conditions of no target, an awake behaving CD1 target, and an
anesthetized CD1 target (group 3 repeated-measure target interaction effect, F4,54= 2168.6, p < 0.0001). (E) Only Susceptible mice display anhe-
donia as measured by a reduction in 1% sucrose preference, whereas both Susceptible and Unsusceptible mice display decreased exploration
on the elevatedplus maze test (F) and enhanced CORT response to a6 min swim stress (G). Only Susceptible mice display blunted circadian rhythms
(H) and significantly enhanced social hyperthermia ([I], an elevated hyperthermic response to the CD1 target). Bars represent mean + SE (standard
error) with n = 10–20, * indicates significant post hoc differences with respect to nondefeated control mice, *p < 0.05, **p < 0.01, ***p < 0.001.
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 393
Increased BDNF Signaling within the NAc
Because chronic social defeat increases BDNF protein
levels in the NAc on days 11 and 39 (Berton et al., 2006),
we tested whether this response differs between Suscep-
tible and Unsusceptible mice. Western blot analysis of
NAc tissue 24 hr after avoidance testing revealed that
only Susceptible mice demonstrated this BDNF increase,
namely a 90% elevation in BDNF levels over controls
(F2,28= 3.88, p < 0.05), with no change in BDNF seen in
the NAc of Unsusceptible mice (Figure 2A). We found no
changes in levels of the full-length or truncated isoforms
of the BDNF receptor, tropomyosin-related kinase B
(TrkB.F or TrkB.T) or in levels of phospho-TrkB (Figure 2B).
Consistent with an increase in NAc BDNF protein, we also
observed a robust activation of signaling molecules down-
stream of TrkB (Chao et al., 2006). Susceptible mice dis-
played increased levels of phosphorylated akt thymoma
viral oncogene (Akt), glycogen synthase kinase 3b (Gsk-
ure 2C), with no significant changes in total levels of these
proteins. Unsusceptible mice didnot show thesechanges,
although there was a strong trend for increased phospho-
ERK levels, suggesting the possibility that ERK activation
could stem partly from nonneurotrophic pathways.
We next tested the involvement of increased BDNF sig-
naling in the NAc in the development of the Susceptible
versus Unsusceptible phenotype. Bilateral intra-NAc infu-
sions of BDNF enhanced susceptibility in response to
a submaximal exposure to defeat stress (Figure 2D), with-
out modifying locomotor activity (Figure S2E). Conversely,
a blockade of increased ERK signaling in the NAc in Sus-
ceptible mice, via overexpression of a dominant-negative
form of ERK2 using a herpes simplex virus (HSV-dnERK),
promoted insusceptibility (Figure 2E), again with no effect
on general locomotor activity (Figure S5D). These data
strongly implicate BDNF induction and downstream sig-
naling within the NAc as a mediator of defeat-induced
To explore mechanisms by which chronic social defeat
increases BDNF levels in the NAc, we first measured
BDNF mRNA levels in this region by qPCR. Control, Sus-
ceptible, and Unsusceptible mice displayed equivalent
levels of BDNF mRNA (p > 0.5), suggesting that the
increased NAc BDNF protein associated with social
avoidance is not dependent on local transcriptional regu-
lation. To test this prediction, we examined the behavioral
effects of an established method to locally delete the bdnf
gene from the NAc by stereotaxically infusing adeno-
associated virus (AAV) that expresses Cre-recombinase
into the NAc of floxed BDNF mice (Berton et al., 2006;
Graham et al., 2007). When AAV-CreGFP- and AAV-
GFP-infected mice were subjected to the social defeat
paradigm, Bdnf gene knockdown within the NAc did not
alleviate defeat-induced avoidance (Figure 3B). This is in
striking contrast to a knockdown of Bdnf within the VTA,
which we have recently shown to prevent defeat-induced
avoidance (Berton et al., 2006). To further characterize the
Table 1. Susceptible and Unsusceptible Mice Display Distinct Syndromes
Day 11 Day 39
Anxiety-like behavior (time in closed arms)
Despair behavior (immobility on TST)
Despair behavior (immobility on FST)
Anhedonia (change in sucrose preference)
Cocaine-conditioned place preference
Stress-induced polydipsia (increased fluid intake)
Locomotor activity (ambulatory beam breaks)
AM serum corticosterone
Swim-stress-induced corticosterone levels
AM serum DHEA-S
Cardiac hypertrophy (heart wt/body wt ratio)N/AN/A
This table illustrates the phenotypic differences between Susceptible and Unsusceptible mice on day 11 and also shows which of
those phenotypes persist 4 weeks later (day 39). TST, tail suspension test; FST, forced swim test; DHEA-S, dehydroepiandroster-
one-sulfate; Wt, weight; 4, [, and Y (no change, significantly greater than, or less than nondefeated control group [p < 0.05],
respectively); N/A, not available.
394 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.
relative contributions of BDNF within these structures to
the behavioral sequelae of social defeat, we defeated
floxed BDNF mice that had been infused with AAV-
CreGFP or AAV-GFP into either the VTA or the NAc. As
shown previously, a VTA-specific Bdnf knockdown led
to an increase in the proportion of Unsusceptible mice af-
terdefeat (11%inAAV-GFP- versus34%inAAV-CreGFP-
injected mice), an effect not seen after a NAc Bdnf knock-
down (data not shown). Likewise, Bdnf knockdown in VTA
and sucrose preference deficit (t18 = 2.46, p < 0.05;
Figure 3D) associated with the Susceptible phenotype.
Coincident with these behavioral findings, we observed
a substantial reduction in the ability of chronic social de-
featto increase BDNF levels inthe NAcin micewitha local
VTA Bdnf knockdown (t13= 3.53, p < 0.01; Figure 3E).
These results strongly implicate the VTA as a crucial
source of BDNF to the NAc during defeat.
Genome-Wide Expression Analyses Reveal
the VTA as a Key Substrate for Resistance
to Social Defeat
Our behavioral and molecular findings indicate that, when
compared to Susceptible mice, Unsusceptible mice dis-
play a prominent lack of phenotype (i.e., changes in social
avoidance, sucrose preference, and BDNF signaling are
observed in Susceptible mice only). While these data
shed light on features associated with vulnerability, they
offer little insight into molecular mechanisms underlying
Unsusceptibility. In the absence of candidate ‘‘resistance
genes,’’ we designed a microarray experiment to explore
global patterns of gene expression in the NAc and VTA
of control, Susceptible, and Unsusceptible mice on day
11. Our goal was to describe two main categories of
genes: (1) genes regulated similarly in Susceptible and
Unsusceptible groups (as a result of exposure to stress)
and (2) genes regulated differentially in Susceptible and
Unsusceptible mice (which may mediate differences in
behavior). Our results, summarized as Venn diagrams in
Figure 4A, revealed that the Unsusceptible phenotype
was associated with the regulation of far more genes.
While the NAc showed a substantially larger list of regu-
lated genes, expression patterns in the VTA were particu-
larly notable in the virtual absence of genes regulated
similarly in Susceptible versus Unsusceptible mice. An
expression-based dendrogram (Figure S4A) confirmed
that VTA gene expression patterns more strongly corre-
lated with our behavioral observations. Figure 4B displays
Figure 2. Increased NAc BDNF: A Molec-
ular Signature of Susceptibility
(A) Results from immunoblotting experiments
showinga90%increase inBDNF levelsinSus-
ceptible mice compared to controls, without
any changes in Unsusceptible mice and with-
out changes in levels of phospho-TrkB.F
(pTrkB.F) or total TrkB.F or TrkB.T (B). (C) In-
creases in NAc BDNF levels in Susceptible
mice are accompanied by the significant acti-
vation of downstream signaling molecules,
including Akt, Gsk-3b, and p44/42 MAPK/
ERK1,2. Mammalian target
kinase-1 (PDK1) proteins were not significantly
activated. (D) A single intra-NAc infusion of
recombinant BDNF (1.5 mg/side) decreases
social interaction (promotes susceptibility)
following a submaximal exposure to defeat
(vehicle: t20= 3.96, p < 0.0001 and BDNF:
t20= 1.44, p > 0.1). (E) While the NAc-specific
overexpression of HSV-GFP in Susceptible
mice did not alleviate social avoidance (t8=
3.49, p < 0.01), HSV-dnERK promoted an Un-
susceptible phenotype (t8 = 0.96, p > 0.3);
groups were matched for day 11 interaction
times (12.2 ± 4.6 and 16.9 ± 7.2 s). Bars repre-
sent mean + SE (n = 5–11), * indicates signifi-
cant post hoc comparisons to respective con-
trol groups, *p < 0.05, **p < 0.01, ***p < 0.0001.
C, Controls; S, Susceptible; U, Unsusceptible.
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 395
a summary of regulated genes in heatmap form, which
emphasizes the unique regulation of gene expression in
amples whose products have previously been implicated
in depressive behaviors, such as histone deacetylase-2
(Hdac2) and adenylyl cyclase 7 (Adcy7) (Hines et al.,
2006; Schroeder et al., 2006). Similarly, only Susceptible
VTA showed a significant upregulation in mRNA levels of
galanin (Gal), which creates a prodepressant phenotype
when infused directly into the VTA (Weiss et al., 1998), fur-
ther validating our microarray results (see Supplemental
Microarray Gene Lists).
Augmented Firing of VTA Dopamine Neurons
Mediates Adaptations to Social Defeat
Among the genes that were significantly upregulated in
the VTA of Unsusceptible mice only, we identified three
voltage-gated potassium (K+) channels (Kcnf1, Kcnh3,
and Kcnq3). Because the induction of these proteins
would be expected to reduce neuronal excitability, we
hypothesized that their unique induction in Unsusceptible
mice could provide a mechanism of insusceptibility, per-
haps by counteracting a defeat-induced excitation of
VTA dopamine neurons. To test this hypothesis, we stud-
ied the effect of social defeat on spontaneous firing rates
of VTA dopamine neurons. We first obtained extracellular
single-unit recordings from VTA dopamine neurons in
slices obtained from control or defeated mice on day 11
(without classifying mice based on susceptibility). At this
time point, chronic social defeat caused a 36% increase
in the firing rate of VTA dopamine neurons (n = 5, t78=
2.15, p < 0.05; Figure 5A). Nondopaminergic cells showed
no change in firing frequency (n = 5, t22= 0.11, p > 0.5).
One defeat experience (n = 4, t138= 1.04, p > 0.3) or
a 10-week-long social isolation stress (n = 4, t109= 0.90,
p > 0.3) both failed to alter VTA firing rates, suggesting
that this change is specific for chronic social defeat.
Next, mice were classified as either Susceptible orUnsus-
ceptible on day 11, and single-unit recordings were
Figure 3. Effects of Region Specific
(A) Schematic coronal sections (Paxinos and
Franklin, 2001) from NAc (left) and VTA (right),
with insets showing representative high-power
micrographs of viral GFP expression.
(B) Floxed BDNF mice were injected with either
AAV-CreGFP or AAV-GFP into the NAc and
subsequently subjected to the social defeat
paradigm. In the presence of a significant
effect of defeat (F1,33= 14.00, p < 0.001), local
knockdown of BDNF in NAc did not attenuate
social avoidance (F1,33= 0.01, p > 0.5).
(C–E) As compared to local BDNF knockdown
within the NAc, an analogous VTA knockdown
ameliorated the effects of social defeat on
weight loss (C) and (D) sucrose preference.
Immunoblotting NAc samples from these two
groups revealed that VTA-injected mice dis-
tein (E). Bars represent mean + SE (n = 7–12),
*p < 0.05, **p < 0.01.
396 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.
Figure 4. Gene Expression Analysis after Social Defeat
DNA microarrays were performed on VTA and NAc of control, Susceptible, and Unsusceptible mice on day 11 after chronic social defeat. (A) Venn
lap depicting genes that were identically regulated by both conditions. Upregulated (red) and downregulated (blue) genes are shown separately
(criteria for significance: R1.5-fold change compared to respective anatomical control group at p < 0.05). (B) Heatmaps illustrating the regulation
of genes for each condition and anatomical structure, with red to blue gradient depicting an up to downregulation (R2-fold increase / %2-fold
decrease). For example, the upper left panel displays significantly regulated genes in Susceptible NAc (top row) and how each of those genes is
regulated in Unsusceptible NAc (bottom row). (C) Summary table showing examples of genes significantly upregulated ([) or downregulated (Y)
ascompared tothenondefeatedcontrolgroupforeachbrain region.TF, transcriptionfactor; SRY,sex-determining region-Y; TRAAK,TWICK-related
amino acid-sensitive K+channel; NEL, neural epidermal growth factor-like; MMTV, mouse mammary tumor virus.
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 397
performed 2weeks later. At this time point, VTA dopamine
5,F2,381=14.37, p<0.0001), withnoeffect seenin Unsus-
ceptible mice (Figures 5B and 5C). VTA firing rates were
significantly correlated with the interaction ratio measured
on day 11 (r = ?0.67, p < 0.01, n = 15; Figure 5C, inset).
To establish a causal link between changes in VTA
excitability and social avoidance, we overexpressed an
inward rectifying K+channel (Kir2.1 or Kcnj2) in the VTA
of Susceptible mice to examine whether this manipulation
would promote resistance to avoidance. We chose Kir2.1
because it has been shown to reliably suppress the excit-
ability of several types of neurons (Burrone et al., 2002;
Nitabach et al., 2002; Dong et al., 2006); indeed, we found
that HSV-mediated Kir2.1 overexpression robustly de-
creased the firing of VTA dopamine neurons (Figure 5D).
We next took two groups of Susceptible mice, matched
for day 11 interaction times (43.0 ± 5.3 and 43.9 ± 6.6 s),
and injected one with HSV-GFP and the other with HSV-
Kir2.1 into the VTA. When assayed 3 days later, HSV-
Kir2.1-injected mice displayed an Unsusceptible pheno-
type as compared to their GFP-infected counterparts
(Figure 5E), despite comparable levels of locomotor activ-
ity (Figure S5D). Also, HSV-Kir2.1-infected mice displayed
significantly reduced NAc levels of BDNF (t20= 1.96, p <
0.05; Figure 5F). The Unsusceptible phenotype induced
by HSV-Kir2.1 was absent after transgene expression had
degraded (8 days following HSV infusions; Figure S5E).
Intra-VTA infusions of HSV-Kir2.1 had no effect on the
behavior of Unsusceptible mice (Figure S5F). Converse
effects were seen upon overexpressing a K+channel
(Kcnab2) rendered dominant-negative (HSV-dnK). HSV-
dnK increased the firing rate of VTA dopamine neurons
(Figure 5D), promoted the development of a Susceptible
phenotype on a submaximal exposure to defeat (Fig-
ure 5G) without affecting general locomotor activity
Figure 5. Increased VTA Dopamine Neuron Firing Mediates Susceptibility
histogram. (B) On day 25, dopamine neurons from only Susceptible mice display significantly enhanced firing rates. (C) Sample traces and spikes;
inset shows that averaged VTA firing rates for each mouse are significantly correlated with interaction ratios measured on day 11. (D) Single-unit
recordings of GFP-positive or GFP-negative VTA dopamine neurons in slice culture, showing that HSV-Kir2.1 and HSV-dnK are able to significantly
modulate the spontaneous activity of VTA neurons in vitro (n = 3–5 mice/group, 20–30 neurons/group). (E) While the VTA-specific overexpression of
HSV-GFP in Susceptible mice did not alleviate social avoidance (t11= 2.98, p < 0.05), HSV-Kir2.1 promoted resilient behavior (t14= 0.30, p > 0.3), and
(F) resulted in a significant reduction in NAc BDNF levels (one-tailed t test). (G) An intra-VTA infusion of HSV-dnK decreased social interaction
(produced a Susceptible phenotype) following a submaximal social defeat regimen (HSV-GFP: t10= 4.52, p < 0.001; and HSV-dnK: t10= 0.31,
p > 0.5) and (H) resulted in a significant increase in BDNF levels (one-tailed t test). Bars represent mean + SE (n = 5–11), * indicates significant
post hoc comparisons to respective control groups, *p < 0.05, **p < 0.01, ***p < 0.0001.
398 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.
(Figure S5D), and enhanced stress-induced increases in
NAcBDNF levels (t9=2.10,p <0.05;Figure5H). Together,
these findings provide direct support for our hypothesis
that a defeat-induced increase in K+channel activity in
the VTA represents a molecular mechanism for resistance
trical activity may be a mechanism for increased BDNF
release into the NAc.
Deficits in Activity-Dependent BDNF Release
Promote an Unsusceptible Phenotype
Thus far, our data are consistent with a model wherein
susceptibility to an avoidant phenotype is caused by up-
regulation of VTA neuronal activity, which results in in-
creased BDNF signaling within the NAc. To test this
hypothesis, we examined the consequence of a naturally
occurring human single-nucleotide polymorphism (SNP)
in the BDNF prodomain (G196A, Val66Met), which impairs
activity-dependent BDNF secretion (Chen et al., 2004;
Egan et al., 2003). Val/Val and Met/Met mice (Chen
et al., 2006) showed comparable responses in the forced
swim and sucrose preference tests (Figures S6A and
S6B), suggesting that Met-BDNF does not affect baseline
responses to stress or natural rewards. However, a dra-
matic phenotype emerged when mice were subjected to
chronic social defeat: while Val/Val mice demonstrated
a significant reduction in social interaction after defeat
(t14= 3.9, p < 0.01; Figure 6A), Met/Met mice displayed
an Unsusceptible phenotype (t14= 0.2, p > 0.5). Nonde-
feated Val/Val and Met/Met mice showed similar interac-
tion scores (p > 0.5). When NAc samples from both de-
feated groups were analyzed for BDNF levels, Met/Met
mice showed 50% lower levels of BDNF protein com-
pared to Val/Val mice (t15= 1.78, p < 0.05; Figure 6B).
While this polymorphism impairs BDNF release, it did
not modify VTA neuronal activity: extracellular recordings
of VTA dopamine neurons from defeated Val/Val and Met/
Met mice showed similar levels of firing (n = 3/group, t72=
0.14, p > 0.5; Figure 6C). These findings further support
our model of how BDNF signaling within the VTA-NAc cir-
cuit relates to vulnerability and resistance to social defeat
(Figure 6E) and indicate that preventing BDNF signaling to
Depressed Humans Display Increased Levels
of BDNF in NAc
To examine the clinical relevance of our findings, we
obtained postmortem samples of human NAc from de-
pressed patients and unaffected controls (Table S3A).
Only samples from males were examined, and groups
were matched for age, postmortem interval, RNA integrity
number (RIN), and tissue pH (Table S3B). We observed
a 40% increase in levels of BDNF protein in the NAc of de-
pressed samples as compared to controls (n = 10, t18=
2.95, p < 0.01; Figure 6D), with no changes in levels of
TrkB protein (data not shown) or BDNF mRNA levels
(Figure S6C). Because most of the depressed patients
had been chronically treated with antidepressants (Table
S3A), we tested whether this BDNF increase could reflect
a drug effect. Chronic treatment (28 days) with imipramine
or TrkB isoforms in NAc from naive mice (Figure S6D).
Upon exposure to psychological stress, why do some
individuals succumb to debilitating psychiatric disease
whereas others progress normally? The goal of the pres-
ent study was to identify molecular mechanisms underly-
ing vulnerability to stress-induced psychopathology, as
well as molecular adaptations that promote resistance to
those changes. We utilized the social defeat paradigm
and segregated socially defeated c57bl/6 mice into Sus-
ceptible and Unsusceptible subgroups: resistance to de-
feat-induced avoidance was found to be long-lasting
and latent. While Unsusceptible mice were immune to
several depression-like changes (e.g., anhedonia and
weight loss), they did display other signs consistent with
exposure to chronic stress (e.g., elevated anxiety and
mice. Interestingly, on day 39, only Unsusceptible mice
developed a significant increase in relative cardiac mass,
suggesting that the persistence of the Unsusceptible phe-
notype may be associated with the potential tradeoff of
prolonged b-adrenergic stimulation (Bonanno et al., 2003)
and possibly its subsequent adverse consequences.
This study shows that genetically identical (inbred) mice
can display phenotypic differences after exposure to
chronic stress; analogous findings have been observed
in the chronic mild stress model (Strekalova et al., 2004).
Such examples of phenotypic variability in inbred mice
have always been attributed to environmental influences
in prenatal and postnatal development and early domi-
nance hierarchies (Peaston and Whitelaw, 2006; Wong
et al., 2005). However, experiments performed on inbred
mice raised in strictly defined environments have shown
that up to 80% of random variability in quantitative traits
(e.g., body weight) are unrelated to genetic and environ-
mental influences (Gartner, 1990). This third component
to natural variation is thought to maintain Gaussian distri-
butions of biological variables independent of environ-
mental influences and sequence constraints and is now
acetylation or methylation (Wong et al., 2005). While these
modifications play an important role in an organism’s
stress responses (Tsankova et al., 2006), future work is
needed todelineate the relativecontribution ofepigenetic,
genetic, and environmental factors that may together
explain variations in susceptibility.
A Role for Neural Reward Substrates
In our paradigm, Susceptible mice showed a deficit in
natural reward (sucrose preference), coincident with an
increase in drug reward (cocaine place conditioning),
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 399
while Unsusceptible mice showed neither of these
changes. This aspect of our paradigm may serve to model
iments revealed that susceptibility to avoidance is marked
by significantly increased levels of BDNF, a molecular ad-
aptation that has also been shown to occur in rodent
models of cocaine withdrawal (Grimm et al., 2003) and
which promotes behavioral responses to cocaine (Gra-
ham et al., 2007; Horger et al., 1999). Postmortem NAc
Figure 6. Human Correlates: Variant BDNF (Val66Met) and Postmortem NAc Studies
(A) Socially defeated Val/Val mice demonstrated a significant reduction in social interaction upon exposure to a CD1 target mouse, while Met/Met
mice behaved comparably in both trials. (B) Day 11 western blot analysis of defeated Val/Val and Met/Met mice showing 50% lower BDNF levels
in NAc of Met/Met mice (one-tailed t test). (C) VTA dopamine neurons from socially defeated Val/Val and Met/Met mice displayed comparable firing
rates. (D) ELISA comparing levels of total BDNF in NAc lysates from depressed humans and unaffected controls (n = 10) (inset: scatterplot). Bars
represent mean + SE (n= 7–8), *indicates significant changes compared to controls, *p< 0.05, **p < 0.01. (E)Schematic: inSusceptible mice, chronic
social defeat increases the firing rate of VTA dopamine neurons, which subsequently gives rise to heightened BDNF signaling within the NAc.
Unsusceptible mice display a resistance to this adverse cascade of events by upregulating various K+channels within the VTA.
400 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.
samples from human depressed patients also showed
increased BDNF levels, indicating that our social defeat
paradigm is a useful method to understand the molecular
neurobiology of human depression.
Results of gene-expression profiling studies revealed
that the Unsusceptible phenotype was associated with
larger number of genes were regulated in this subgroup,
suggesting that the expression of an Unsusceptible
phenotype is an active neurobiological process that is
gulation of several voltage-gated K+channel subunits in
the VTA of Unsusceptible mice encouraged us to explore
the electrophysiological correlates of Susceptible and Un-
susceptible behavior. We discovered that Susceptible
mice show a long-lasting upregulation in the firing rate of
VTA dopamine neurons, an effect not seen in Unsuscepti-
ble mice. With the aid of HSV-encoded K+channels, we
demonstrated that selective and specific modulations of
ity. These data support a causative role for increased VTA
VTA firing rates are normal in Unsusceptible mice despite
mechanism requires further study, as our gene-array
experiments did not reveal alterations in ion channels
that could explain this phenomenon. Further work is also
needed to assess whether social stressors alter the
tonic-phasic pattern of VTA neuronal firing in vivo (Grace,
We also observed a strong association between VTA
firing rates and NAc BDNF levels. This increase in NAc
BDNF protein levels (in both Susceptible mice and de-
pressed humans) was not accompanied by Bdnf mRNA
changes, and a region-specific Bdnf gene deletion from
the VTA, but not from the NAc, blocked the defeat-
induced increase in NAc BDNF protein levels and led to
an Unsusceptible phenotype. We propose that increases
in NAc levels of BDNF protein and downstream signaling
observed in Susceptible mice are due to enhanced activ-
ity-dependent BDNF release from VTA dopamine neu-
conditions (Figure 6E). This model is consistent with the
phenomenon of anterograde axonal BDNF transport that
is reported to occur within this circuit (Altar and DiStefano,
1998), and the observation that VTA neurons are strongly
activated during the threat of social subordination (Tidey
and Miczek, 1996). Increased dopamine release in the
NAc in the context of social defeat may promote alertness
during a potentially harmful situation, and simultaneous
BDNF release may promote neuroplastic changes neces-
sary for survival. However, excessive and prolonged VTA
activation and BDNF signaling to the NAc may be malad-
aptive, as it could produce long-lasting inflexibility and
overgeneralization, causing harmless social cues to be-
A Naturally Occurring Polymorphism Promotes
Insusceptibility: BDNF G196A/Val66Met
The G196A SNP of the Bdnf gene results in the substitu-
tion of Met in place of Val in the prodomain of BDNF (Lu
et al., 2005). Humans possessing this polymorphism
display a selective impairment in episodic memory and
abnormal hippocampal activation (Eganetal.,2003).Con-
sistent with a role for the BDNF prodomain in intracellular
trafficking and secretion, the Met-BDNF variant causes
defective intracellular localization and impaired activity-
dependent BDNF release (Chen et al., 2004, 2006). Mice
homozygous for the Met/Met variant have recently been
shown to display impaired learning and reduced hippo-
campal volumes (Chen et al., 2006). We show here that
while Val/Val and Met/Met mice showed comparable be-
havior on baseline measures of emotionality, Met/Met
mice displayed a striking Unsusceptible phenotype in
the social defeat paradigm, which was associated with
an ?50% reduction in levels of BDNF protein in the NAc.
These data show that a naturally occurring impairment in
activity-dependent BDNF release promotes resistance
to social defeat and further strengthen our working
hypothesis. Genetic-association studies report that the
BDNF Met allele is associated with more favorable antide-
pressant responses (Choi et al., 2006; Yoshida et al.,
2006), but other studies have displayed little consensus
as to whether the BDNF Met allele alters rates of depres-
2006; Strauss et al., 2005; Surtees et al., 2007). Clearly,
further studies are required to examine the influence of
the BDNF Val/Met polymorphism on human psychopa-
thology in light of our findings. It is also particularly inter-
esting that a variant of BDNF that impairs contextual
learning (Chen et al., 2006) also leads to resistance to de-
feat-induced social avoidance. This finding may reflect an
evolutionary tradeoff to ensure species survival: while
a certain degree of memory impairment may adversely
affect an organism’s fitness, it may serve a positive func-
tion by preserving motivation for social interaction in the
face of chronic stress.
In conclusion, we propose that resistance to social
phenomenon of resilience in humans. The resistance
described here constitutes a latent trait, which accurately
recapitulates the human condition, in that resiliency is
often only identifiable after exposure to stressful events
(Hoge et al., 2007; Yehuda et al., 2006). Resilient individ-
uals display a striking ability to preserve optimism in the
reward substrates that are either especially plastic or in-
sensitive to change (Charney, 2004; Curtis and Cicchetti,
2003). Either way, these features could be mediated by
neuroplastic events within the VTA-NAc circuit. Our ability
to enhance resistance to stress by reducing BDNF release
from the VTA to NAc, or by blocking BDNF signaling in the
NAc, provides novel insight into the development of thera-
peutic agents to promote human resilience in the context
of severe stress. Furthermore, our studies on the BDNF
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 401
Met allele suggest that while this variant may confer unfa-
Egan et al., 2003), its effects on mesolimbic BDNF signal-
ing may represent a compensatory biological advantage
under adverse conditions.
Subjects and Drugs
Male 7-week-old c57bl/6 (Jackson), CD1 retired breeders (Charles
River), 9- to 13-week-old floxed BDNF mice (Berton et al., 2006),
and 10- to 14-week-old BDNF Met/Met and Val/Val mice (Chen
et al., 2006) were used. Cocaine (5 mg/kg) and imipramine (20 mg/kg)
were given ip. All experiments were performed in accordance with the
UTSWMC Institutional Animal Care and Use Committee and the Insti-
tutional Review Board.
Social Defeat and Behavioral Testing
lished protocols (Berton et al., 2006; Tsankova et al., 2006). During
each defeat episode, intruder mice were allowed to interact for
10 min with an aggressive CD1 mouse, during which they were at-
tacked and displayed subordinate posturing. For the social interaction
test, we measured the time spent in the interaction zone (Figure S1A)
during the first (target absent) and second (target present) trials; the
interaction ratio was calculated as 100 3 (interaction time, target pres-
ent)/(interaction time, target absent). IPTT-300 temperature transpon-
ders (Bio Medic Data Systems) were implanted in the dorsal interscap-
ular region under isoflurane anesthesia. Behavioral phenotyping on
days 11 and 39 were performed using standard protocols extensively
described in Supplemental Experimental Procedures.
For sucrose-preference testing, a solution of 1% or 2% sucrose or dil-
uent alone (drinking water) was filled in 50 ml tubes with stoppers fitted
with ball-point sipper tubes (Ancare). All animals were acclimatized to
two-bottle choice conditions prior to testing conditions. Daily, at
?1600 hr, the fluid level was noted, and the position of the tubes
were interchanged. Sucrose preference was calculated as a percent-
age [100 3 volume of sucrose consumed (in bottle A)/total volume
consumed (bottles A and B)] and was averaged over at least 3 days
Mice were anesthetized with a cocktail of ketamine (100 mg/kg) and
xylazine (10 mg/kg), positioned in a small-animal stereotaxic instru-
ment, and the skull surface was exposed. 33 gauge needles were
used to bilaterally infuse 0.5 ml of virus (or BDNF) into NAc or VTA at
a rate of ?0.1 ml/min (Berton et al., 2006).
Immunoblotting and Immunoassays
NAc tissue punches (core and shell) were lysed, sonicated, and centri-
fuged in an EMSA buffer, following which 40 mg of supernatant protein
was electrophoresed on precast 4%–20% SDS gradient gels. Follow-
ing transfer, PVDF membraneswere washed in13 Tris-bufferedsaline
with 0.1% Tween-20 (TBS-T), and blocked in 5% w/v milk for 1 hr at
25?C. The membrane was then incubated in a solution of the appropri-
ate primary antibody overnight at 4?C, peroxidase-labeled secondary
antibody at 25?C for 1 hr, and bands were visualized by enhanced
chemiluminescence. Serum for enzyme immunoassays (EIAs) was
obtained from centrifuged trunk blood, and steroid levels were as-
sayed with commercially available EIA kits per manufacturer’s instruc-
tions. Similarly, BDNF levels from human NAc were assayed from
homogenized and lysed protein extracts using previously published
protocols (Chen et al., 2006).
Human Postmortem Study
Human specimens were obtained from the Dallas Brain Collection
(Stan et al., 2006). After obtaining next of kin permission, tissue was
collected from cases at the Dallas County Medical Examiners Office
and The Transplant Service Center at UTSWMC. Blood toxicology
screens were conducted in each case, and subjects with a recent or
past history of drug abuse, neurological disorders, or head injury
were excluded. Clinical records and collateral information from tele-
phone interviews with a primary caregiver was obtained for each
case. Two psychiatrists carried out an extensive review of the clinical
records and made independent diagnoses followed by a consensus
diagnosis using DSM IV criteria. To obtain specimens of human
nucleus accumbens, cerebral hemispheres were cut coronally into
1–2 cm blocks. Dissected NAc was immediately placed in a mixture
of dry ice and isopentane (1:1, v:v). The frozen tissue was then pulver-
ized on dry ice and stored at ?80?C. For measurements of tissue pH,
a 150 mg cerebellar punch was homogenized in 5 ml of ddH2O (pH
adjusted to 7.00) and centrifuged for 3 min at 8000 3 g at 4?C. The
pH of this supernatant was measured in duplicate. Each sample’s
RNA integrity number was determined by isolating total RNA using
Trizol (Invitrogen) followed by analysis with an Agilent 2100 Bioana-
lyzer. For protein studies, ?100 mg of NAc tissue was homogenized
in 1 ml of lysis buffer (100 mg/ml PMSF, 2 mg/ml aprotinin of leupeptin,
aprotinin, and pepstain in PBS) with a Polytron homogenizer (900
rpm312times). Samples werethensonicated, and byusingthe Brad-
ford assays, protein concentrations were found to be between 2 and
PCR and Gene Expression Microarrays
RNA from NAc, VTA, and hypothalamus was prepared using the
RNAeasy Micro Kit (QIAGEN). cDNA was obtained using a first-strand
synthesis kit (Invitrogen). All PCR experiments were conducted in trip-
licate, and the data were analyzed by using the DDCt method (Tsan-
kova et al., 2006) and were normalized to measures of Gapdh
mRNA. For microarrays, NAc and VTA tissue was obtained from a sin-
gle experiment where 50% of stressed mice were Unsusceptible. To
reduce variability and increase statistical power, we simultaneously
performed three biological replicates for each group, each consisting
of pools of mRNA from four mice (Peng et al., 2003). RNA quality
was verified by an Agilent Bioanalyzer prior to labeling and hybridiza-
tion (performed by the UTSWMC Microarray Core) onto Illumina
Mouse V6-1.1 full genome arrays (Illumina). Raw expression values
were subjected to a cubic spline normalization and averaged across
triplicates. Genes were considered to be significantly regulated if
they displayed a >1.5-fold change in expression compared to their
respective anatomical control group (at p < 0.05).
Mice were perfused with cold artificial cerebrospinal fluid (aCSF),
following which 250 mm VTA slices were placed in an aCSF-filled hold-
ing chamber. Slices were transferred into a recording chamber fitted
with a constant flow rate of aCSF (2.5 ml/min). Glass microelectrodes
filled with 2.0 M NaCl were used to record single-unit extracellular
potentials that were monitored through a high-input impedance ampli-
fier (Axon Instruments). Dopamine neurons were identified by their
location and electrophysiological criteria: regular and spontaneous
action potentials with triphasic waveforms (Ungless et al., 2004). Firing
rate was recorded in the amplifier’s bridge mode, and data acquisition
and on-line analysis of firing rate were collected using a Digidata
1322A digitizer and pClamp 8.2 (Axon Instruments). Cell culture
recordings were performed in a cell-attached configuration.
Unless otherwise noted, we used two-tailed unpaired Student’s t tests
(for comparison of two groups), one-way ANOVAs followed by the
Dunnett’s Multiple Comparison (for three groups), and one-way
402 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.
effects). Two-way ANOVAs were performed when more than one fac-
tor was examined simultaneously, followed by Fisher’s Least Signifi-
cant Difference post hocs.
The Supplemental Data for this article can be found online at http://
We would like to thank V. Iyer for statistical advice; M. Cobb for ERK
plasmids; G. Xiao, S. Laali, and H. Truong for technical assistance.
This was work was funded by grants from The National Institute of
Mental Health and The National Institute on Drug Abuse (E.J.N.)
and National Alliance for Research in Schizophrenia and Depression
Received: March 23, 2007
Revised: July 23, 2007
Accepted: September 14, 2007
Published: October 18, 2007
Altar, C.A., and DiStefano, P.S. (1998). Neurotrophin trafficking by
anterograde transport. Trends Neurosci. 21, 433–437.
Avgustinovich, D.F., Kovalenko, I.L., and Kudryavtseva, N.N. (2005). A
model of anxious depression: persistence of behavioral pathology.
Neurosci. Behav. Physiol. 35, 917–924.
Berton, O., McClung, C.A., Dileone, R.J., Krishnan, V., Renthal, W.,
Russo, S.J., Graham, D., Tsankova, N.M., Bolanos, C.A., Rios, M.,
et al. (2006). Essential role of BDNF in the mesolimbic dopamine path-
way in social defeat stress. Science 311, 864–868.
Bonanno, G.A., Noll, J.G., Putnam, F.W., O’Neill, M., and Trickett, P.K.
(2003). Predicting the willingness to disclose childhood sexual abuse
Maltreat. 8, 302–318.
Brady, K.T., and Sinha, R. (2005). Co-occurring mental and substance
use disorders: the neurobiological effects of chronic stress. Am. J.
Psychiatry 162, 1483–1493.
Burrone, J., O’Byrne, M., and Murthy, V.N. (2002). Multiple forms of
synaptic plasticity triggered by selective suppression of activity in
individual neurons. Nature 420, 414–418.
Chao, M.V., Rajagopal, R., and Lee, F.S. (2006). Neurotrophin signal-
ling in health and disease. Clin. Sci. (Lond.) 110, 167–173.
Charney, D.S. (2004). Psychobiological mechanisms of resilience and
vulnerability: implications for successful adaptation to extreme stress.
Am. J. Psychiatry 161, 195–216.
Charney, D.S., and Manji, H.K. (2004). Life stress, genes, and depres-
sion: multiple pathways lead to increased risk and new opportunities
for intervention. Sci. STKE 2004, re5.
Chen, Z.Y., Patel, P.D., Sant, G., Meng, C.X., Teng, K.K., Hempstead,
B.L., and Lee, F.S. (2004). Variant brain-derived neurotrophic factor
(BDNF) (Met66) alters the intracellular trafficking and activity-depen-
dent secretion of wild-type BDNF in neurosecretory cells and cortical
neurons. J. Neurosci. 24, 4401–4411.
Chen, Z.Y., Jing,D., Bath, K.G., Ieraci, A.,Khan, T.,Siao, C.J., Herrera,
D.G., Toth, M., Yang, C., McEwen, B.S., et al. (2006). Genetic variant
BDNF (Val66Met) polymorphism alters anxiety-related behavior.
Science 314, 140–143.
Choi, M.J., Kang, R.H., Lim, S.W., Oh, K.S., and Lee, M.S. (2006).
Brain-derived neurotrophic factor gene polymorphism (Val66Met)
and citalopram response in major depressive disorder. Brain Res.
Curtis, W.J., and Cicchetti, D. (2003). Moving research on resilience
into the 21st century: theoretical and methodological considerations
in examining the biological contributors to resilience. Dev. Psychopa-
thol. 15, 773–810.
Dong, Y., Green, T., Saal, D., Marie, H., Neve, R., Nestler, E.J., and
Malenka, R.C. (2006). CREB modulates excitability of nucleus accum-
bens neurons. Nat. Neurosci. 9, 475–477.
Egan, M.F., Kojima, M., Callicott, J.H., Goldberg, T.E., Kolachana,
B.S., Bertolino, A., Zaitsev, E., Gold, B., Goldman, D., Dean, M.,
et al. (2003). The BDNF val66met polymorphism affects activity-
dependent secretion of BDNF and human memory and hippocampal
function. Cell 112, 257–269.
Gartner, K. (1990). A third component causing random variability be-
side environment and genotype. A reason for the limited success of
a 30 year long effort to standardize laboratory animals? Lab. Anim.
Grace, A.A. (2000). The tonic/phasic model of dopamine system regu-
lation and its implications for understanding alcohol and psychostimu-
lant craving. Addiction 95 (Suppl 2), S119–S128.
Graham, D.L., Edwards, S., Bachtell, R.K., Dileone, R.J., Rios, M., and
Self, D. (2007). Dynamic BDNF activity in nucleus accumbens with
cocaine use increases self administration and relapse. Nat. Neurosci.
Published online July 8, 2007. 10.1038/nn1929.
Gratacos, M., Gonzalez, J.R., Mercader, J.M., de Cid, R., Urretaviz-
caya, M., and Estivill, X. (2007). Brain-derived neurotrophic factor
Val66Met and psychiatric disorders: Meta-analysis of case-control
studies confirm association to substance-related disorders, eating
disorders, and schizophrenia. Biol. Psychiatry 61, 911–922.
Grimm, J.W., Lu, L., Hayashi, T., Hope, B.T., Su, T.P., and Shaham, Y.
(2003). Time-dependent increasesinbrain-derived neurotrophicfactor
from cocaine: implications for incubation of cocaine craving. J. Neuro-
sci. 23, 742–747.
Hines, L.M., Hoffman, P.L., Bhave, S., Saba, L., Kaiser, A., Snell, L.,
Goncharov, I., LeGault, L., Dongier, M., Grant, B., et al. (2006). A
sex-specific role of type VII adenylyl cyclase in depression. J. Neuro-
sci. 26, 12609–12619.
evidence and conceptual considerations for posttraumatic stress
disorder. Depress. Anxiety 24, 139–152.
Hong, C.J., Huo, S.J., Yen, F.C., Tung, C.L., Pan, G.M., and Tsai, S.J.
(2003). Association study of a brain-derived neurotrophic-factor
genetic polymorphism and mood disorders, age of onset and suicidal
behavior. Neuropsychobiology 48, 186–189.
Horger, B.A., Iyasere, C.A., Berhow, M.T., Messer, C.J., Nestler, E.J.,
and Taylor, J.R. (1999). Enhancement of locomotor activity and condi-
tioned reward to cocaine by brain-derived neurotrophic factor. J. Neu-
rosci. 19, 4110–4122.
Hwang, J.P., Tsai, S.J., Hong, C.J., Yang, C.H., Lirng, J.F., and Yang,
Y.M. (2006). The Val66Met polymorphism of the brain-derived neuro-
trophic-factor gene is associated with geriatric depression. Neurobiol.
Aging 27, 1834–1837.
Kessler, R.C., Sonnega, A., Bromet, E., Hughes, M., and Nelson, C.B.
(1995). Posttraumatic stress disorder in the National Comorbidity Sur-
vey. Arch. Gen. Psychiatry 52, 1048–1060.
Liu, X., Peprah, D., and Gershenfeld, H.K. (2003). Tail-suspension
Res. 37, 249–259.
Lu, B., Pang, P.T., and Woo,N.H. (2005). The yin and yang of neurotro-
phin action. Nat. Rev. Neurosci. 6, 603–614.
Malatynska, E., and Knapp, R.J. (2005). Dominant-submissive behav-
ior as models of mania and depression. Neurosci. Biobehav. Rev. 29,
Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc. 403
Martinez, M., Calvo-Torrent, A., and Pico-Alfonso, M.A. (1998). Social
defeat and subordination as models of social stress in laboratory
rodents: A review. Aggress. Behav. 24, 241–256.
McLaughlin, J.P., Li, S., Valdez, J., Chavkin, T.A., and Chavkin, C.
(2006). Social defeat stress-induced behavioral responses are medi-
ated by the endogenous kappa opioid system. Neuropsychopharma-
cology 31, 1241–1248.
reward circuit in depression. Biol. Psychiatry 59, 1151–1159.
Nestler, E.J., Barrot, M., DiLeone, R.J., Eisch, A.J., Gold, S.J., and
Monteggia, L.M. (2002). Neurobiology of depression. Neuron 34, 13–
Nitabach, M.N., Blau, J., and Holmes, T.C. (2002). Electrical silencing
of Drosophila pacemaker neurons stops the free-running circadian
clock. Cell 109, 485–495.
Paxinos, G., and Franklin, K.B.J. (2001). The Mouse Brain in Stereo-
taxic Coordinates (New York: Academic Press).
Peaston, A.E., and Whitelaw, E. (2006). Epigenetics and phenotypic
variation in mammals. Mamm. Genome 17, 365–374.
Peng, X., Wood, C.L., Blalock, E.M., Chen, K.C., Landfield, P.W., and
Stromberg, A.J. (2003). Statistical implications of pooling RNA sam-
ples for microarray experiments. BMC Bioinformatics 4, 26.
Rutter, M. (2006). Implications of resilience concepts for scientific
understanding. Ann. N Y Acad. Sci. 1094, 1–12.
Schroeder, F.A., Lin, C.L., Crusio, W.E., and Akbarian, S. (2006). Anti-
depressant-like effects of the histone deacetylase inhibitor, sodium
butyrate, in the mouse. Biol. Psychiatry. 62, 55–64.
Stan, A.D., Ghose, S., Gao, X.M., Roberts, R.C., Lewis-Amezcua, K.,
Hatanpaa, K.J., and Tamminga, C.A. (2006). Human postmortem tis-
sue: what quality markers matter? Brain Res. 1123, 1–11.
Strauss, J., Barr, C.L., George, C.J., Devlin, B., Vetro, A., Kiss, E., Baji,
I., King, N., Shaikh, S., Lanktree, M., et al. (2005). Brain-derived neuro-
trophic factor variants are associated with childhood-onset mood dis-
order: confirmation in a Hungarian sample. Mol. Psychiatry 10, 861–
Strekalova, T., Spanagel, R., Bartsch, D., Henn, F.A., and Gass, P.
(2004). Stress-induced anhedonia in mice is associated with deficits
in forced swimming and exploration. Neuropsychopharmacology 29,
Surtees, P.G., Wainwright, N.W., Willis-Owen, S.A., Sandhu, M.S.,
Luben, R., Day, N.E., and Flint, J. (2007). No association between
the BDNF Val66Met polymorphism and mood status in a non-clinical
community sample of 7389 older adults. J. Psychiatr. Res. 41, 404–
Tidey, J.W., and Miczek, K.A. (1996). Social defeat stress selectively
alters mesocorticolimbic dopamine release: an in vivo microdialysis
study. Brain Res. 721, 140–149.
Tsankova, N.M., Berton, O., Renthal, W., Kumar, A., Neve, R.L., and
Nestler, E.J. (2006). Sustained hippocampal chromatin regulation in
a mouse model of depression and antidepressant action. Nat. Neuro-
sci. 9, 519–525.
Ungless, M.A., Magill, P.J., and Bolam, J.P. (2004). Uniform inhibition
of dopamine neurons in the ventral tegmental area by aversive stimuli.
Science 303, 2040–2042.
Weiss, J.M., Bonsall, R.W., Demetrikopoulos, M.K., Emery, M.S., and
West, C.H. (1998). Galanin: a significant role in depression? Ann. N Y
Acad. Sci. 863, 364–382.
Wong, A.H., Gottesman,I.I., and Petronis, A. (2005). Phenotypic differ-
ences in genetically identical organisms: the epigenetic perspective.
Hum. Mol. Genet. 14(Spec No 1), R11–R18.
J. Clin. Psychiatry 65 (Suppl 1), 29–36.
Yehuda, R., Flory, J.D., Southwick, S., and Charney, D.S. (2006).
ability following trauma exposure. Ann. N Y Acad. Sci. 1071, 379–396.
Yoshida, K., Higuchi, H., Kamata, M., Takahashi, H., Inoue, K., Suzuki,
T., Itoh, K., and Ozaki, N. (2006). The G196A polymorphism of the
brain-derived neurotrophic factor gene and the antidepressant effect
of milnacipran and fluvoxamine. J. Psychopharmacol., in press. Pub-
lished online November 8, 2006. 10.1177/0269881106072192.
Microarray data can be accessed through NCBI’s Gene Expression
Omnibus Webpage at http://www.ncbi.nlm.nih.gov/geo (Accession
404 Cell 131, 391–404, October 19, 2007 ª2007 Elsevier Inc.