, 75 (2012);
et al.Olivier Berton
Are We Getting Closer to Valid Translational Models for Major
This copy is for your personal, non-commercial use only.
clicking here. colleagues, clients, or customers by
, you can order high-quality copies for your
If you wish to distribute this article to others
The following resources related to this article are available online at
here. following the guidelines
can be obtained by
Permission to republish or repurpose articles or portions of articles
Updated information and services,
): January 23, 2013 www.sciencemag.org (this information is current as of
version of this article at:
including high-resolution figures, can be found in the online
related to this article
A list of selected additional articles on the Science Web sites
, 15 of which can be accessed free:
cites 59 articles
This article appears in the following
registered trademark of AAAS.
is a Science 2012 by the American Association for the Advancement of Science; all rights reserved. The title
Copyright American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on January 23, 2013
the human hippocampus (12). Aside from niche
components, stem and progenitor cells should
also be considered for their contribution to neu-
rogenic control and mood regulation. They do
not make or receive traditional neuronal synapses
or connections and have previously been consid-
ered passive participants in the neurogenic niche.
cells provide structural, biochemical, and meta-
bolic signals to the niche that can regulate neu-
rogenesis (3). Thus, we have to investigate how
each individual component of the niche is influ-
enced by potential antidepressants and consider
components of the niche themselves as specific
targets for novel antidepressant therapeutics.
mechanisms, which may lead to novel avenues
for treatments. However, more work needs to be
done. The typically parallel lines of research on
for example, whether enhanced pattern separa-
tion enhances stress coping and levels of neuro-
genesis in laboratory animals (19, 23). Other
preclinical experiments are needed to fine-tune
the neurogenic interactome, assessing whether
stimulation of the amygdala enhances neuro-
genesis and whether other brain regions involved
in depression, like the nucleus accumbens, are
also influenced by levels of neurogenesis. Clin-
regulationof human neurogenesis,particularly in
the brains of depressed humans (12). Interesting-
ly, neurogenesis decreases with age in humans
and animals (13, 25), whereas depression prev-
alence increases with age. More research is war-
ranted to examine to what extent the age-induced
increase in depression is due to life experience,
age-induced increase in medical burden, or pos-
sibly age-induced decrease in neurogenesis
(4, 5, 12, 25). Also, in vivo imaging of correlates
of human neurogenesis is feasible (25, 26), but
greater technical advances are needed before we
can conclude what aspects of neurogenesis struc-
a destination on the road to remission for people
with mood disorders may eventually be custom-
ized diagnostic evaluation that takes into account
their individual levels of adult neurogenesis.
References and Notes
1. A. Tanti, C. Belzung, Br. J. Pharmacol. 159, 1187 (2010).
2. D. Petrik, D. C. Lagace, A. J. Eisch, Neuropharmacology
62, 21 (2012).
3. G. L. Ming, H. Song, Neuron 70, 687 (2011).
4. S. A. Small, S. A. Schobel, R. B. Buxton, M. P. Witter,
C. A. Barnes, Nat. Rev. Neurosci. 12, 585 (2011).
5. M. Fotuhi, D. Do, C. Jack, Nat Rev Neurol 8, 189 (2012).
6. J. E. Malberg, A. J. Eisch, E. J. Nestler, R. S. Duman,
J. Neurosci. 20, 9104 (2000).
7. B. Czéh et al., Proc. Natl. Acad. Sci. U.S.A. 98, 12796
8. L. Santarelli et al., Science 301, 805 (2003).
9. M. N. Jayatissa, K. Henningsen, G. Nikolajsen, M. J. West,
O. Wiborg, Stress 13, 95 (2010).
10. D. C. Lagace et al., Proc. Natl. Acad. Sci. U.S.A. 107,
11. D. J. David et al., Neuron 62, 479 (2009).
12. M. Boldrini et al., Biol. Psychiatry 72, 562 (2012).
13. R. Knoth et al., PLoS One 5, e8809 (2010).
14. J. S. Snyder, A. Soumier, M. Brewer, J. Pickel,
H. A. Cameron, Nature 476, 458 (2011).
15. T. D. Perera et al., PLoS One 6, e17600 (2011).
16. D. M. Lyons et al., Proc. Natl. Acad. Sci. U.S.A. 107,
17. E. D. Kirby et al., Mol. Psychiatry 17, 527 (2012).
18. V. K. Parihar, B. Hattiangady, R. Kuruba, B. Shuai,
A. K. Shetty, Mol. Psychiatry 16, 171 (2011).
19. D. D. Pollak et al., Neuron 60, 149 (2008).
20. T. Christensen, C. F. Bisgaard, O. Wiborg, Neuroscience
196, 66 (2011).
21. V. Krishnan et al., Cell 131, 391 (2007).
22. C. D. Clelland et al., Science 325, 210 (2009).
23. T. Nakashiba et al., Cell 149, 188 (2012).
24. C. Anacker, C. M. Pariante, Mol. Psychiatry 17, 9 (2012).
25. L. N. Manganas et al., Science 318, 980 (2007).
26. A. C. Pereira et al., Proc. Natl. Acad. Sci. U.S.A. 104,
Acknowledgments: This work was supported by grants from
the NIH, National Alliance for Research on Schizophrenia and
Depression and NASA. The authors thank J. Hsieh, S. Padrick,
and C. Tamminga for feedback and helpful discussions. We
apologize to all the authors whose work we could not cite and
discuss because of limitations of space and scope.
Are We Getting Closer to
Valid Translational Models for
Olivier Berton,1* Chang-Gyu Hahn,1,2Michael E. Thase2
Advances in characterizing the neuropathology and functional dysconnectivity of depression and
promising trials with emerging circuit-targeted and fast-onset therapeutics are providing unprecedented
opportunities to gain deeper insight into the neurobiology of this devastating and pervasive disorder.
Because of practical and ethical limitations to dissecting these mechanisms in humans, continued
progress will critically depend on our ability to emulate aspects of depressive symptomatology and
treatment response in nonhuman organisms. Although various experimental models are currently
available, they often draw skepticism from both clinicians and basic research scientists. We review
recent progress and highlight some of the best leads to diversify and improve discovery end points for
preclinical depression research.
or physiological symptoms in animals. To be
considered valid for hypothesis testing and ther-
apeutic development, modeled symptoms are ex-
of human disease and should respond to clin-
ically effective treatments,with clinicallyrelevant
time courses. Modeling major depressive disor-
odels of psychiatric illness rely on ma-
nipulation of known etiological or risk
factors to induce tractable behavioral
der (MDD) along these lines poses a number of
Lost in Translation: Why Animal Models
Struggle with Major Depression
A first obstacle is our limited understanding of
the multifactorial pathogenic processes that un-
derlie affective, cognitive, and homeostatic ab-
normalities in MDD. Given the subjective nature
of most core symptoms and the lack of valid
biomarkers, establishing links between clinical
variables and animal end points cannot currently
be done without a certain degree of interpretation
Further complication derives from the ex-
treme heterogeneity of the illness and from the
difficult delineation of its boundaries with health
and other psychiatric conditions (1). A good ex-
ample is the symptomatic facet of anhedonia (or
loss of pleasure and interest) that is considered a
cardinal feature of MDD but is truly present only
in up to 50% of MDD patients and is very prom-
inent in other conditions, such as schizophrenia
(2). Our growing appreciation of individual pat-
terns of functional disconnections and genetic
variability associated with MDD suggests that
the current diagnostic approach collapses a myr-
iad of distinct pathophysiological states that are
unlikely to be captured comprehensively by any
single experimental model (3).
Not surprisingly, given this clinical and eti-
ological heterogeneity, the therapeutic effects of
current antidepressants, which all share mono-
1Department of Psychiatry, Center for Neurobiology and Be-
havior, University of Pennsylvania, Philadelphia, PA 19104,
USA.2Department of Psychiatry, Mood and Anxiety Disorders
Treatment and Research Program, University of Pennsylvania,
Philadelphia, PA 19104, USA.
*To whom correspondence should be addressed. E-mail:
VOL 338 5 OCTOBER 2012
on January 23, 2013
target, are often inconsistent (about 50% of pa-
placebo-expectancy effects are typically larger
than specific drug effects) (1). This therapeutic
inconsistency, too, has important implications
for modeling, as it impedes the use of reference
drugs in animals as a clear-cut criterion for vali-
dation or invalidation.
Despite these problems, we have many heu-
and several have proven undeniably useful for
the rational development of monoaminergic anti-
the epidemiological evidence that stress and ad-
verse psychosocial experiences (such as interper-
sonal violence, neglect, or unwanted separation)
often precede the onset, or predictthe recurrence,
of depressive episodes. Most of these paradigms
were developed 30 years ago and have been ap-
plied since with little modifications. Given the
failure of virtually all clinical development pro-
grams with candidate therapeutics based on non-
monoaminergic mechanisms during the past
decades, the capability of current models to de-
tect antidepressants with truly novel mechanisms
sive reviews have been published in recent years
discussing the pros and cons of each model and
a strong consensus on the need to better incor-
porate etiological factors in the design of novel
paradigms (4, 5).
During the past 5 years, we have witnessed
a number of key conceptual and technical ad-
vances that hold the promise of transforming the
light some of these recent developments and at-
tempt a prospective analysis of how they may
contribute to improve and diversify discovery
end points for depression. We focus on two areas
of need: (i) the need to translate our improved
understanding of depression circuitry into behav-
neuropathology into models of MDD.
Toward Models of Circuit-Centered
Symptomatic Dimensions in MDD
To address the problem of MDD’s clinical het-
erogeneity, an approach that has emerged over
the last decade is the recasting of MDD symp-
types defined using quantifiable and objective
behavioral or neural end points (3). Examples of
symptoms include measures of whole-brain con-
nectivity under resting state or in response to he-
donic or aversive stimuli, as well as standardized
behavioral measures of hedonic capacity, emo-
tional reactivity,and nonverbal social attunement.
Although integration of these neurobehavioral
variables seems to afford a discriminative power
for pathology approaching that of psychometric
constructs (2, 6–9), large-scale trials necessary
still lacking. Nevertheless, efforts to validate and
diversify this type of translatable variable, prox-
imal to circuit function, are critical to bridge the
divide between MDD clinical phenomena and
preclinical models. A complementary aspect of
this effort, on the preclinical side, consists in dis-
secting circuit modules that underpin aligned
phenotypic domains in animals. Emergent tech-
of circuit activity (10) and real-time monitoring
of interregional connectivity (11) in behaving an-
(12, 13) are facilitating this task. We highlight
below a number of recent studies using these ap-
proaches to explore behavioral domains related
to negative affect, positive affect, socioaffective
functioning, and cognitive function.
Negative affect. The behavioral end point re-
lated to “negative affect” that the majority of ro-
dent studies examines using simple behavioral
tasks (like the forced swimming test) or more
elaborate paradigms (like the learned helpless-
ness test) is a form of behavioral passivity and
quiescence that develops in many species upon
exposure to uncontrollable stress. Stress-induced
can be delayed or normalized by antidepressants.
Although this class of test is often labeled as be-
that cannot be explained through direct mono-
it may seem necessary to update certain of our
views about these paradigms and to improve our
understanding of underlying circuitry.
ing of motor responses in freely moving rats and
simultaneous recording of medial prefrontal cor-
tex (mPFC) neurons, Warden et al. (16) were
able to identify a cluster of cells that predict ac-
tive escape attempts in the forced swimming test.
neurons, they demonstrated their causal role in
the animal’s “decisions” to struggle or stay pas-
sive in the test. Not all pyramidal neurons in the
escape behaviors but only a specific distributed
subset defined by their axonal connections with
the raphe nuclei, the region in the brainstem that
contains the bulk of serotonin neurons. These re-
sults converge with those of several other groups
to suggest that, as drug screens, the so-called
by which the prefrontal cortex exerts top-down
control over ancestral circuits implicated in the
generation of aversive states (such as panic and
dysphoria) and comprising serotonergic neurons.
These results reinforce the construct validity of
this class of tests and may facilitate the develop-
ment of reverse-translational tasks to interrogate
The lateral habenula (LHb) is another brain
region heavily implicated in the descending con-
trol of monoaminergic circuits. It has also recent-
ly been implicated in the generation of aversion
learned helplessness test (17, 18). Here again,
virally mediated track tracing and optogenetics
were used to identify specific neuronal subpopu-
lations in the LHb that affect behavior and to
of the habenula is highly conserved, and its role
in the encoding of negative affective states like
fish to primates. DBS of the LHb is considered a
possible therapeutic approach for MDD after a
number of promising case reports.
and hedonic capacity are commonly observed in
MDD and contribute to the complex construct of
anhedonia (19). Evidence for fast antianhedonic
effects of DBS in the nucleus accumbens in
treatment-resistant MDD patients implicates this
highly conserved component of the mesolimbic
motivation and addictive processes. A number of
recent studies using genetic approaches to trace,
identify, and silence neurons have started pro-
anisms through which chronic stress in rodents
represses preference for highly palatable foods,
an anhedonia-like symptom reversedby chronic
antidepressants. By implicating synaptic changes
affecting specific neuronal subtypes in this het-
erogeneous brain region, such as dopamine me-
dium spiny neurons expressing the dopamine D1
studies have started tracing a wiring diagram that
may prove instrumental in refining electrode-
based strategies for treatment of anhedonia.
Socioaffective function. Deficits of interper-
sonal functioning are another important compo-
nent of MDD symptomatology. These deficits
have been linked to endophenotypes such as de-
serotonin reuptake inhibitor (SSRI)–sensitive re-
smaller primates (such as the tree shrews) have
provided the best evidence for the occurrence of
of socioaffective alterations in nonhuman pri-
mates exposed to “ethological” stressors (such as
periods of isolation during early development or
forced subordination in adulthood) include ex-
5 OCTOBER 2012 VOL 338
on January 23, 2013
unresponsive posture (Fig. 1) and the exacerba-
toms, reported both in captive and free-ranging
primates (23), appear partly reversible by admin-
istering chronic SSRIs and co-occur with cardio-
hippocampal volume (24) and decreased neuro-
models of socioaffective deficits have a strong
face-validity with MDD, our capability to inves-
tigate their underlying circuitry is limited by tech-
nical and ethical considerations.
Rodent social stress models such as the so-
cialdefeatparadigm(26)or testsofsocial domi-
(27)provide a number of valid alternatives to ex-
amine neurobiological underpinnings of socio-
affective behaviors.The relatively long-lived and
antidepressant-sensitive form of social avoidance
that develops in a subset of mice subjected to
with an array of other motivational and metabolic
changes) has provided a reliable end point to
with social aversion or resilience to social stress
(26). Electrophysiological studies in this model
point to alterations in several subcortical circuits
(partly overlapping with those involved in anhe-
donia (28, 29) and negative affect (30) high-
lighted above) that discriminate between socially
avoidant animals and their resilient counterparts.
In vivo multiregion electrophysiological tech-
niques have also revealed altered patterns of
corticolimbic synchrony in defeated mice (31),
which suggests that stress-induced alterations in
the connections between the mPFC and down-
stream targets in the limbic system (such as the
of socioaffective phenotypes relevant to MDD.
This hypothesis is in line with the recent demon-
stration that high-frequency photostimulation of
or genetic manipulations affecting synaptic effi-
cacy in the same circuits induce antidepressant-
like responses in tests of social aversion (32) and
social competition (33).
Cognition. Cellular and molecular correlates
of cognitive deficits in MDD (such as deficits
in attention and executive function) are under-
behavioral models to address these mechanisms,
and it is good news that a number of such para-
digms are starting to emerge, in primates (indis-
pensable to approach mechanisms underlying
higher cognition) as well as in rodents. A core
concept of cognitive theories of depression is the
Neurogenic, volumetric, and functional correlates
of socioaffective function and higher cognition
Molecular, cellular, and circuit mechanisms
underlying stress-induced behavioral deficits
Large scale behavior-based genetic or
Fig. 1. Preclinical paradigms used in the study of MDD-related neurobe-
havioral traits in nonhuman species. No etiologically valid depression-like
syndrome that reproduces the complexity and heterogeneity of the human
syndrome has been established in nonhuman animals. Current paradigms cap-
italize primarily on exposure to acute or chronic stressors as a depressogenic
trigger and examine species-specific behavioral repertoires in an attempt to
capture distinct behavioral or psychological dimensions of the illness. Ex-
amples of animals studied include the zebrafish, various laboratory rodents
the etiology of MDD by replicating environmental risk factors, circuit ab-
normalities, and genetic mutations affecting candidate biological pathways.
Various models also replicate responses to pharmacotherapies or somatic
therapeutic interventions, such as electroconvulsive therapy and DBS. The
zebrafish is a model organism commonly used to dissect behaviorally relevant
based on automatic movement tracking (40, 41). This approach is illustrated
here with a hear map depicting movement during a social interaction task in
an adult zebrafish homozygous for the mutation grs357, which disrupts tran-
scriptional regulation by glucocorticoid receptor and leads to abnormal socio-
affective behaviors sensitive to antidepressants (42). A range of behavioral
motivational, affective, cognitive, and social deficits associated with MDD (5).
Spectral plots derived from multicircuit field recordings depict changes in the
oscillatory synchrony of the basal amygdala and nucleus accumbens during
depression-related behaviors in the mouse (11). Nonhuman primates have
provided the most immediate evidence for the occurrence of “pathological”
and are also providing key insights about higher cognitive dysfunctions asso-
ciated with depression (35). A form of depression-like behavior, observed in
both free-ranging and captive nonhuman primates, is a behavioral posture
strikingly reminiscent of human prostrated sadness and social detachment,
first described in macaques deprived of secure maternal attachment during
associated with pathophysiology [such a hippocampal volume (61) and treat-
ment response such as regulation of hippocampal neurogenesis (25)] in the
context of social stress.
VOL 3385 OCTOBER 2012
on January 23, 2013
notion of cognitive-affective bias, which derives
emphasize negatively valenced information and
or memory away from negative material (34).
A recent study relying on a novel cost-benefit
decision task in macaques recently identified a
subregion of the anterior cingulate cortex that
encodes negative motivational value and whose
simistic predictions of outcomes (35). The fact
that this subregion is contiguous and intercon-
nected with cg25 and is similarly regulated by
various classes of antidepressants reinforces its
possible role in MDD. As the use of analogous
(albeit more rudimentary) behavioral tasks is be-
ginning to be reported in rodents, it will be im-
cortex in these species plays a similar role in de-
termining “pessimistic” decision biases (36, 37).
Discovering new behavioral end points. As
our understanding of circuit modules underlying
MDD symptoms and therapeutic responses in-
on optogenetics (10), pharmacogenetics (38) and
to reach a wider range of species including pri-
“tunable symptoms” in animals that can support
antidepressant therapeutic screens. In combina-
ing circadian monitoring of complex home-cage
behaviors in groups of rodents (39) or high-
throughput behavioral-based screening in zebrafish
(40,41)—these approaches may offer “unbiased”
means to discover novel depression-related be-
taxonomy based on the genetic conservation of
underlying circuits rather than on simple behav-
Toward Translational Models of
Genetic Susceptibility and Cellular
Dysfunction in Depression
In the search for etiologically valid animal mod-
els of MDD, much hope has been placed in re-
cent years in our ability to harness the power of
genomics to resolve the genetic architecture of
MDD heritability (estimated to 40%) and to iden-
tify causal mutations that can be reproduced in
animals to emulate (in interaction with stress
inducers discussed above) various aspect of be-
havioral and neural MDD phenotypes.
els have been so far developed in the context
the “common variant” hypothesis, whereby the
genetic risk for MDD is hypothesized to reflect
the additive or multiplicative influence of several
alleles with small effect. Several depression-like
phenotypes have been reported in animals car-
rying mutations replicating naturally occurring,
date genes, such as BDNF (42), TPH2 (43), or
5-HTT (44). Other mouse lines have targeted
potential risk genes that emerged from unbiased
genome-wide association studies (GWAS), such
as the presynaptic protein Piccolo (45).
Humanized animals that faithfully reproduce
variants naturally occurring in patients have al-
lowed the development of pioneering translation-
analogous behavioral tasks [for review, see (46)].
Because they allow examination of the effect of
human alleles in the context of a homogenous
genetic background,in animalsraisedunder con-
the BDNF Val66Met knock-in mice represent a
powerful approach to identify functional effects
that may be difficult to replicate or even detect
in clinical setting. Studies with these mice have
iations in affective traits that predispose to depres-
sion (such as threat perception or fear extinction)
are influenced by variations inthe trafficking and
release of BDNF resulting from the substitution
in a single codon (47). However, the epidemio-
logical evidence that this variant significantly in-
creases the MDD’s risk remains contentious for
reasons that, as discussed above, may reflect the
heterogeneous stratification produced by MDD
diagnosis as much as the complexity of depres-
As genotyping methods progress and the
sizes of the cohorts included in GWAS for MDD
continue to grow and facilitate the replication
of small effect sizes, more stringently validated
MDD risk genes [such as the recently identified
SLC6A15 gene that encodes a neuron-specific
neutral amino acid transporter (48)] may become
more frequent, ultimately opening the way to ex-
of depression-related phenotypes is moderated
through epistatic interactions in synergy with po-
tent environmental constraints applied at specific
ber of technical barriers (such as the ability to
point mutations in the same animal), we can
look ahead optimistically given the rapid pace
as TALENs or zinc finger nucleases) are evolv-
ing. Because these approaches not only hold the
promise of facilitating the generation of mutant
as a model species for preclinical MDD studies.
The lack of a highly penetrant mutation as-
sociated with MDD is often presented as one of
the primary causes for the lack of valid MDD
ing evidence of a higher burden of copy number
variants in depressed patients (49) (i.e., rare per-
mutations in large portions of gene or chromo-
somes) and the early characterization of one
such rare mutation with strong penetrance [a
duplication recently reported in the SLIT3 gene
that encodes a protein involved in axon guid-
Another important line of data in this context
involves the well-replicated, albeit not systemat-
ic, depression-related phenotypic features (i.e.,
in anhedonia and “despair” tests) reported in the
several different lines of mutant mice now avail-
able for the DISC1 gene (51). Translocation of
DISC1 in the Scottish kindred leads to a var-
with MDD versus 24% with schizophrenia or
bipolar disorder) and the link between DISC1
and MDD has been reinforced by several recent
lines of evidence for an association of DISC1
with symptom severity and age of onset in MDD
(52). The question of how various or even iden-
tical mutations of DISC1 could increase risk to
develop MDD versus other psychiatric disorders
converges with the current debate that surrounds
ahead, the increasing availability of induced
pluripotent stem cells derived from patients carry-
ing DISC1 mutations (53) may help tease apart
relevantcellular mechanismsand facilitatereverse
suchasolfactory neuroepithelialcelllines that can
be biopsied both in patients and animals for lon-
gitudinalstudies of various aspectsof neuralfunc-
tion and signaling—also offer unique translational
opportunities to dissect the respective impact of
genetics, epigenetics, and environmental factors
on intracellular signaling mechanisms relevant to
stress vulnerability and risk for MDD (54).
In part thanks to the development of tissue-
bank initiatives, postmortem transcriptomics and
proteomicshave helpedlink regionalevidence of
neuropathology in MDD with abnormalities in a
number of candidate signaling pathways that are
continue to involve neurotrophic signaling cas-
cades (55) and neurogenic processes, two inter-
related mechanisms mediating the therapeutic
activity of monoaminergic antidepressants. Reg-
ulation of monoaminergic and amino acid neu-
rotransmission, neurohormonal signaling, and
inflammatory cascades continue to be other active
areas of investigation for MDD cellular patho-
genesis (56). Progress of conditional and inter-
sectional genetic strategies and virally mediated
manipulations have allowed dissection of these
pathways in animals with greater degrees of pre-
ificity. Advances in these strategies have helped
pinpoint critical periods for the developmental
programming of emotional functions relevant
to MDD (57) and have allowed identification
specific cell-types by which important modula-
tors of stress vulnerability such as CRHR1 (12),
P11 (58), and HDAC6 (30) exert their effects on
5 OCTOBER 2012 VOL 338
on January 23, 2013
Despite these important new leads, there is Download full-text
genetic mouse model of depression, with a well-
used for the study neurodegenerative disorders.
etiology and phenomenology, but may also de-
efforts to homogenize behavioral paradigms and
methodologies across laboratories. This is a like-
ly cause for the frequent lack of replication of
Gene × Environment effects and an issue that
will become increasingly important as studies of
epigenetic mechanisms in models of depression
of consensus is related to the time course of anti-
a delayed response requiring weeks of treatment
might simulate the clinical effects of current anti-
depressants, this question needs reconsideration
lasting antidepressant effects of ketamine and de-
rivatives (14). There is a need to determine how
the doses of drugs administered to animals relate
to their clinical effects for future discoveries, be-
the engagement of target effects are rarely re-
of patients remain resistant to current treatments,
models for these subpopulations.
Neuroscience may not hold all the keys to a pub-
lic health issue as complex as depression. How-
ever, the tremendous advances made over the
past few decades continue to hold the promise
that a better understanding may ultimately ease
suffering and erase stigma. Animal models are
pivotal in this effort to translate basic progress
suggests that, although it seems unlikely that any
one model will ever recapitulate this heteroge-
neous illness in its entirety, many current para-
digms are yielding key neurobiological insights
relevant to behavioral dimensions and affective
constructs in humans. A challenge remains to un-
derstand how these dimensions integrate in the
context of pathology. A second challenge will be
with those assessed in genetic studies or during
the various phases of development of novel anti-
depressants. Translational medicine is a two-way
bridge. Preclinical research needs to inform clin-
ical trials and diagnosis, but the reverse is also
true. Without a consensus about what depression
isand how to more reliablyand specificallymea-
sure it, rapid progress seems unlikely. Thus, it is
critical to keep in mind the conceptual, method-
ological, and organizational factors that cause the
field of depression therapeutics to remain at least
1. M. E. Thase, in Handbook of Depression, I. H. Gotlib and
C. L. Hammen, Eds. (Guilford Press, New York, 2008),
2. R. Bogdan, Y. S. Nikolova, D. A. Pizzagalli, Neurobiol. Dis.
5. E. J. Nestler, S. E. Hyman, Nat. Neurosci. 13, 1161 (2010).
6. L. L. Zeng et al., Brain 135, 1498 (2012).
7. T. Hahn et al., Arch. Gen. Psychiatry 68, 361 (2011).
8. J. K. Cohn et al., presentation at the Third Conference on
Affective Computing and Intelligent Interaction and
Workshops, 2009 (ACII 2009), 10 to 12 September 2009
(IEEE, Piscataway, NJ, 2010).
9. X. Zhang, Z. S. Yaseen, I. I. Galynker II, J. Hirsch,
A. Winston, PLoS ONE 6, e27253 (2011).
10. M. K. Lobo, E. J. Nestler, H. E. Covington 3rd,
Biol. Psychiatry 71, 1068 (2012).
11. K. Dzirasa, R. Fuentes, S. Kumar, J. M. Potes,
M. A. Nicolelis, J. Neurosci. Methods 195, 36 (2011).
12. D. Refojo et al., Science 333, 1903 (2011).
13. R. D. Airan et al., Science 317, 819 (2007).
14. A. E. Autry et al., Nature 475, 91 (2011).
15. C. Hamani, J. N. Nóbrega, Eur. J. Neurosci. 32, 1109 (2010).
16. M. R. Warden et al., presentation no. 306.15/YY26, 2011
Neuroscience Meeting Planner (Society for Neuroscience,
Washington, DC, 2011).
17. S. J. Shabel, C. D. Proulx, A. Trias, R. T. Murphy,
R. Malinow, Neuron 74, 475 (2012).
18. B. Li et al., Nature 470, 535 (2011).
19. A. Der-Avakian, A. Markou, Trends Neurosci. 35, 68 (2012).
20. B. K. Lim, K. W. Huang, B. A. Grueter, P. E. Rothwell,
R. C. Malenka, Nature 487, 183 (2012).
21. J. L. Warner-Schmidt et al., Proc. Natl. Acad. Sci. U.S.A.
109, 11360 (2012).
22. A. M. Cusi, A. Nazarov, K. Holshausen, G. M. Macqueen,
M. C. McKinnon, J. Psychiatry Neurosci. 37, 154 (2012).
23. H. R. Ferdowsian et al., PLoS ONE 6, e19855 (2011).
24. C. A. Shively, S. L. Willard, Exp. Neurol. 233, 87 (2012).
25. T. D. Perera et al., PLoS ONE 6, e17600 (2011).
26. O. Berton et al., Science 311, 864 (2006).
27. E. Malatynska, R. Rapp, D. Harrawood, G. Tunicliff,
Pharmacol. Biochem. Behav. 82, 306 (2005).
28. J. L. Cao et al., J. Neurosci. 30, 16453 (2010).
29. D. J. Christoffel et al., J. Neurosci. 31, 314 (2011).
30. J. Espallergues et al., J. Neurosci. 32, 4400 (2012).
31. V. Narayanan et al., PLoS ONE 6, e22600 (2011).
32. H. E. Covington 3rd et al., J. Neurosci. 30, 16082 (2010).
33. F. Wang et al., Science 334, 693 (2011).
34. S. G. Disner, C. G. Beevers, E. A. Haigh, A. T. Beck,
Nat. Rev. Neurosci. 12, 467 (2011).
35. K. Amemori, A. M. Graybiel, Nat. Neurosci. 15, 776 (2012).
36. E. J. Harding, E. S. Paul, M. Mendl, Nature 427, 312 (2004).
38. S. Dong, S. C. Rogan, B. L. Roth, Nat. Protoc. 5, 561 (2010).
39. L. H. Tecott, E. J. Nestler, Nat. Neurosci. 7, 462 (2004).
40. D. Kokel et al., Nat. Chem. Biol. 6, 231 (2010).
41. L. Ziv et al., Mol. Psychiatry (2012).
42. H. Yu et al., J. Neurosci. 32, 4092 (2012).
43. J. M. Beaulieu et al., Proc. Natl. Acad. Sci. U.S.A. 105,
44. A. M. Carneiro et al., Proc. Natl. Acad. Sci. U.S.A. 106,
45. Y. Furukawa-Hibi et al., Neuroreport 21, 1177 (2010).
46. A. R. Hariri, Sci. Transl. Med. 2, 18ps6 (2010).
47. F. Soliman et al., Science 327, 863 (2010).
48. M. A. Kohli et al., Neuron 70, 252 (2011).
49. J. J. Rucker et al., Mol. Psychiatry (2011).
50. J. T. Glessner et al., PLoS ONE 5, e15463 (2010).
51. D. J. Porteous, J. K. Millar, N. J. Brandon, A. Sawa,
Trends Mol. Med. 17, 699 (2011).
52. P. A. Thomson et al., Mol. Psychiatry (2012).
53. C. H. Chiang et al., Mol. Psychiatry 16, 358 (2011).
55. H. J. Kang et al., Nat. Med. 18, 1413 (2012).
57. E. S. Deneris, S. C. Wyler, Nat. Neurosci. 15, 519 (2012).
58. E. F. Schmidt et al., Cell 149, 1152 (2012).
59. H. Sun, P. J. Kennedy, E. J. Nestler, Neuropsychopharmacology
60. W. T. McKinney Jr., S. J. Suomi, H. F. Harlow,
Am. J. Psychiatry 127, 1313 (1971).
61. S. L. Willard, J. B. Daunais, J. M. Cline, C. A. Shively,
Menopause 18, 582 (2011).
Acknowledgments: Preparation of this Review was supported
by grant MH087581 from the U.S. National Institute of Mental
Health, NIH, and by an award from the International Mental Health
ResearchOrganization. Theauthors thankI. Luckiand B. Brookshire
for their comments during preparation of the manuscript.
The Science of Resilience:
Implications for the Prevention
and Treatment of Depression
Steven M. Southwick1and Dennis S. Charney2*
Human responses to stress and trauma vary widely. Some people develop trauma-related
psychological disorders, such as posttraumatic stress disorder (PTSD) and depression; others
develop mild to moderate psychological symptoms that resolve rapidly; still others report no new
psychological symptoms in response to traumatic stress. Individual variability in how animals and
humans respond to stress and trauma depends on numerous genetic, developmental, cognitive,
psychological, and neurobiological risk and protective factors.
resilience, it is generally understood as the abil-
ity to bounce back from hardship and trauma.
The American Psychological Association defines
esilience to stress is a complex multidi-
mensional construct. Although there is
no one universally accepted definition of
resilience as “the process of adapting well in the
face of adversity, trauma, tragedy, threats or even
significant sources of threat” (1).
Genetic factors play an important role in an
individual’s response to stress and trauma (2).
Twin studies have estimated an overall heritability
VOL 3385 OCTOBER 2012
on January 23, 2013