se hypothesis
kal1,3, Mark D. Kvarta1,3,4,and Xiang Cai1,5
edicine, 655 West Baltimore Street, Baltimore, MD 21201, USAedicine, 655 West Baltimore Street, Baltimore, MD 21201, USA
of Maryland School of Medicine, 655 West Baltimore Street,
hool of Medicine, 655 West Baltimore Street, Baltimore, MD 21201,
ndale, IL 62901, USA
in the brain driving the range of behavioral symptoms thatcharacterize depression is essential for developing moreeffective treatments for this disorder and preventing suicide.
In the search for ways to prevent, treat, and cure diseases,a strong hypothesis can prove invaluable if it accounts forknown etiologies, is consistent with existing human andpreclinical data gathered across a range of modalities,makes clearly testable predictions, explains the effective-ness of existing therapies mechanistically, and offers guid-ance for the development of novel prophylactic and
Feature Reviewlifetime prevalence of 712% in men and 2025% in wom-en, and a multibillion-dollar annual economic burden inthe USA [1,2].
The most tragic consequence of untreated depression issuicide, attempted by as many as 8% of severely depressedpatients. According to the Centers for Disease Control andPrevention, nearly half a million patients receive emergencycare for suicide attempts each year in the USA and over38 000 individuals die by intentional self-inflicted injuries,twice as many lives as are lost to homicide. Shockingly,23% of suicide victims were being treated with antidepres-sants at the time [3]. In fact, only half of patients with majordepression respond to standard-of-care antidepressants(ADs), such as selective serotonin reuptake inhibitors(SSRIs) [4], with 70% failing to achieve full remission[5]. A better understanding of the nature of the changes
mild stressors twice per day for 3 weeks.
Forced swim test: animals are placed in a tank of water until they cease
struggling. Nave, unstressed animals have longer latencies to cease struggling
compared with stressed animals.
Learned helplessness: animals receive a single session with repeated
unpredictable and uncontrollable stressors (foot shocks).
Learned helplessness test: animals are placed in a chamber where they
previously received inescapable foot shocks paired with a conditioned
stimulus. Latency to escape from the same chamber is later measured in
response to the conditioned stimulus.
Novelty suppressed feeding test: animals are food deprived then placed in a
brightly lit arena with food in the center. The latency until they venture into the
center and eat is measured. Nave, unstressed animals display a shorter
latency compared with stressed animals. This test may reveal more about
anxiety than about hedonic state.
Social interaction test: animals are placed in an arena with a novel juvenile
animal held in a small cage. Time spent in the vicinity of the cage is compared
in the presence and absence of the juvenile. Nave, unstressed animals spend
more time in the vicinity of the cage when the juvenile is present, whereas
chronically stressed animals do not, presumably because the social interaction
is no longer rewarding.
Sucrose preference test: a two-bottle choice between plain water and dilute
(1%) sucrose solution. Nave, unstressed animals drink approximately 8090%
of their liquid from the sucrose solution, whereas chronically stressed animals
drink nearly equally from both bottles, presumably because the sucrose
solution is no longer rewarding.
Tail suspension test: mice are dangled by their tails until they cease to
struggle. Nave, unstressed animals have longer latencies to immobility
compared with stressed animals.
0166-2236/
2015 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tins.2015.03.003
Corresponding author: Thompson, S.M. ([email protected]).Keywords: glutamate; stress; ketamine; reward; hippocampus; nucleus accumbens.An excitatory synapof depressionScott M. Thompson1,2,3, Angy J. KallaracAdam M. Van Dyke1,3, Tara A. LeGates1, 1Department of Physiology, University of Maryland School of M2Department of Psychiatry, University of Maryland School of M3Programs in Neuroscience and Membrane Biology, UniversityBaltimore, MD 21201, USA4Medical Scientist Training Program, University of Maryland ScUSA5Department of Physiology, Southern Illinois University, Carbo
Depression is a common cause of mortality and morbid-ity, but the biological bases of the deficits in emotionaland cognitive processing remain incompletely under-stood. Current antidepressant therapies are effectivein only some patients and act slowly. Here, we proposean excitatory synapse hypothesis of depression in whichchronic stress and genetic susceptibility cause changesin the strength of subsets of glutamatergic synapses atmultiple locations, including the prefrontal cortex (PFC),hippocampus, and nucleus accumbens (NAc), leading toa dysfunction of corticomesolimbic reward circuitry thatunderlies many of the symptoms of depression. Thishypothesis accounts for current depression treatmentsand suggests an updated framework for the develop-ment of better therapeutic compounds.
Major depressive disorder: from a symptom-baseddescription to biological phenotypesMajor depressive disorder (MDD) is one of the most com-mon and costly of neuropsychiatric syndromes, with aGlossary
Allostatic load/overload: allostasis is the concept that homeostatic set-points
can be differentially regulated to meet different demands in the internal and
external environment (e.g., physical or psychological stress). Repeated
allostatic adaptation exacts a cost on the brain.
Chronic restraint stress: animals are placed in restraint tubes for several hours
daily, repeated over several days (e.g., 4 hours/day for 1014 days).
Chronic social defeat: animals are typically placed in the home cage of a novel
aggressor for 30 min, once daily for 3 weeks.
Chronic unpredictable stress: animals experience two bouts of one of manyTrends in Neurosciences, May 2015, Vol. 38, No. 5 279
therapeutic approaches. The serotonin hypothesis of depres-sion, formulated during the early 1960s, provides an excel-lent example, driving the field of biological psychiatryforward and leading to the development of SSRIs (Box 1).More recently, considerable evidence of the involvement ofneurotrophins in depression and antidepressant action hasaccumulated, leading to a neurotrophin hypothesis of de-pression (Box 2). Nevertheless, these hypotheses leavemany questions about both the causes and treatment ofdepression.
In this review, we highlight emerging evidence of dys-functions of excitatory synaptic transmission [6], and sug-gest how this defect correlates with the genesis of depressionand the evidence that serotonin and neurotrophins have arole in depression. We are now moving beyond a symptom-based description of depression as a single disease entityand beginning to identify biological phenotypes that can becompared across species and across classically definedlabels, as encouraged by the Research Domain Criteria ofthe National Institute of Mental Health (NIMH). Thus, asignificant goal of basic research is to identify the neurobio-logical consequences of disease-promoting conditions at the
involves a combination of genetic and epigenetic suscepti-bility together with environmental risk factors, such asstress, emotional trauma, or traumatic head injury [7], withheritable factors contributing slightly less than half of therisk. Many SNPs and epigenetic differences are linked toincreased risk for depression, but no single gene candidateproduces a strong enough effect to provide convincing mech-anistic hypotheses [8]. This reinforces the fact that MDD is acomplex and heterogeneous collection of symptoms causedby variations in multiple genes, each responsible for a smalleffect on risk, that ultimately converge onto common circuit,cellular, and molecular pathways.
Stress and depressionStressful life events are a key environmental risk factor fordepressive disorders in genetically susceptible individuals[9] and are suspected to be causal in many patients.Patients who are depressed report more stressful lifeevents and have fewer social resources compared withnondepressed subjects [10]. Personality traits, in particu-lar neuroticism and lack of a confidence, have also beenlinked to the etiology of depression and may increase risk
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 5level of alterations in gene products, synapses, cells, andcircuits, and then, in a bottom-up manner, map these dis-coveries to the specific behavioral deficits characterizinghuman neuropsychiatric conditions. Here, we focus on thecircuits mediating reward behavior, which underlies manyof the symptoms of human depression, such as anhedoniaand aberrant reward-associated perception and memory.We also highlight how this excitatory synapse hypothesis ofdepression offers a new framework with which to approachthe treatment of patients with depression.
The etiology of depressionDespite its high incidence and its socioeconomic impact, thecauses of depression remain poorly understood. Depression
Box 1. The serotonin hypothesis of depression
The monoamine neurotransmitter serotonin (5-hydroxytryptamine,5-HT) is synthesized by neurons in the DR nucleus. These neuronsintegrate inputs from multiple brain regions, including the NAc,amygdala, LHb, and PFC, and send projections throughout the brain,including to the hippocampus, PFC, substantia nigra, and NAc.
Chance observations of mood-altering compounds more than50 years ago led to the first coherent theory of depression andopened the door for current first-line treatments. Evidence that theseagents alter the concentration of monoamine neurotransmitters, suchas serotonin, dopamine, and norepinephrine (e.g., [136]), led to thehypothesis that depression is caused by a deficiency of monoamines[137139]. Given that the pharmacological profile of many AD drugs ismore consistent with changes in serotonin levels than of othermonoamine neurotransmitters, it is now more common to speak of aserotonin hypothesis of depression, although inhibitors of norepi-nephrine uptake are also effective ADs, including the tricyclics. Moststatements of the serotonin hypothesis fail to offer mechanisticexplanations; nevertheless, this theory became a foundation forresearch in biological psychiatry, and the central dogma in the field ofmajor depression for many decades.
Although deficits in serotonin levels or release are a centralprediction of this hypothesis, the relation between human depression
and the levels of serotonin or its metabolite 5-hydroxyindoleaceticacid (5-HIAA) remains unclear, with early enthusiasm giving way tomixed results [140]. Experimental manipulations of serotonin levels in
280in response to stress [11].McEwen, Sapolsky, and others emphasized the impor-
tance of allostatic overload (see Glossary) as an explana-tion for many chronic illnesses of modern human life,including depression [12,13]. These stressors may bephysical or psychosocial, the latter including low self-esteem, loneliness, deficient social skills, excessive anxi-ety, rumination, and negative thinking. All stressorsactivate the sympathetic nervous system and the hypo-thalamuspituitaryadrenal (HPA) axis, causing eleva-tion of glucocorticoids (GC) and other stress hormones.Normally, the physiological stress response is self-termi-nating, due to negative feedback, directly via the hypo-thalamus and pituitary and indirectly via several brain
humans by depleting or enhancing the precursor for its synthesis,tryptophan, also remain inconclusive [141144]. In general, acutemanipulation of tryptophan levels is more likely to affect mood inpatients who are depressed than in otherwise healthy individuals,perhaps because it impairs the beneficial actions of their SSRIs.
One prediction of this hypothesis is that elevation of serotoninlevels would relieve the symptoms of depression, and testing thisprediction led directly to the development of SSRIs, which produce arapid elevation of extracellular serotonin concentration [145]. Withfewer adverse effects than the less-specific tricyclics, SSRIs wereprescribed to tens of millions during their first decade.
SSRI-induced elevation of serotonin increases activation of seroto-nin receptors, of which there are 14 subtypes, each having a uniquepattern of expression and localization. Alterations in a singlepopulation, or subpopulation, of serotonin receptors might also beresponsible for the symptoms of depression, rather than a globaldysfunction of the neurotransmitter system, but evidence of suchchanges remains inconclusive [146148].
Fifty years after the initial observations, considerable debate aboutthe validity of the serotonin hypothesis remains, with only incon-sistent and inconclusive evidence supporting it. It is unlikely thatdepression is caused by a simple decrease in serotonin synthesis,
release, or receptor expression. Nevertheless, there is clear evidencethat elevated levels of monoamines can relieve the symptoms ofdepression.
areas enriched in glucocorticoid receptors (GRs), includ-ing the PFC and hippocampus.
A subset of patients with depression display abnormal
Box 2. The neurotrophin hypothesis of depression
Depression has been hypothesized to result when environmentalfactors, chronic stress, and genetic susceptibility interact to impairneurotrophin signaling [15,98,149]. The consequences of inadequateneurotrophin signaling that might contribute to the genesis ofdepression include impaired neurogenesis in the dentate gyrus,atrophy of distal dendrites, and impaired neuronal plasticity. It hasbeen further postulated that ADs alleviate the symptoms of depres-sion by reversing or blocking this effect, so as to restore neurotrophinsignaling. Given their predominance in the neocortex and hippocam-pus, the most data concern BDNF and its receptor, TrkB [150,151].
This hypothesis predicts that BDNF signaling is decreased inpatients with depression and restored with successful AD treatment,and there is some evidence of decreased BDNF-TrkB signaling inhuman depression. First, serum BDNF levels are decreased in patientswith depression [152] and recover with AD treatment [153]. Addition-ally, BDNF and TrkB mRNA levels are decreased in postmortem tissuefrom suicide victims [154]. Patients treated with ADs have a higherlevel of BDNF compared with untreated patients, specifically in thedentate gyrus, hilus, and supragranular regions of the hippocampus[155], and increased hippocampal neurogenesis [156]. Furthermore, apolymorphism (Val66Met) in the BDNF gene is correlated with variousaspects of depression and AD action (e.g., [157,158]), includingreduced hippocampal volume, depression, and suicidality [159161].
The hypothesis also predicts that impairing BDNF signaling shouldinduce a depressive-like phenotype and interfere with AD responsesin animal models. Some mice with decreased BDNFTrkB signalingdisplay a depression-like phenotype, including decreased neuralproliferation in the dentate gyrus [94] and reduced dendritic spine
Feature ReviewHPA axis activity, resulting in increased basal levels ofGCs and a blunted circadian rhythm [14]. GCs regulateneuronal survival, excitability, proliferation, and metabo-lism, and allostatic overload may promote depression byimpairing these processes [15]. Given their role in negativefeedback, a dysfunction of GRs in areas such as the hippo-campus can cause HPA dysregulation and failure to limitthe stress response [16]. Indeed, some patients with de-pression display an inability to suppress the axis in re-sponse to a challenge [17]. However, HPA challenges lacksensitivity and specificity as a diagnostic biomarker. In-terestingly, patients treated chronically with corticoste-roids and patients with Cushings disease, bothhypercortisolemic states, are more likely to have depres-sion and experience a range of depression-related cognitiveand memory deficits [18,19].
The structure of the depressed brain is different fromthe healthy brain, likely representing one visible manifes-tation of pathophysiology. Postmortem studies of suicidecompleters and patients with severe depression haverevealed a reduction in the volume of the PFC and hippo-campus, regions thought to have a role in the cognitiveaspects of depression [20]. An impairment of normal, on-going neurogenesis in the dentate gyrus of the hippocam-pus, as observed in several models of chronic stress [21],and atrophy of dendrites [22] might underlie this decreasein hippocampal volume. Successful AD treatment reversesthese decreases in hippocampal volume (i.e., [23]). Asreviewed elsewhere, the size of other brain regions in-volved in regulation of reward and emotion may also beaffected in patients with depression, including the amyg-dala [24] and striatum [25].
density in CA1 cells [162], as well as reduced responses to acute ADs[163]. By contrast, heterozygous BDNF-knockout mice do not differfrom wild types in anxiety and behavioral despair measures in some[164,165], but not all [166], studies. Monteggia et al. reported that onlyfemale conditional forebrain BDNF-knockout mice had a depressedphenotype and failed to respond acutely to ADs [167]. Pyramidal cellsin the PFC of transgenic mice with the BDNF Met allele knocked indisplay an impairment of synaptic plasticity, but no change inbaseline synaptic function [168]. Unfortunately, depression-relatedbehavioral assays with construct and face validity, such as thesucrose preference and novelty suppressed feeding tests, have notyet been tested in these mice.
Such experiments are complicated for two reasons. First, BDNFTrkB signaling is crucial for brain development, rendering interpreta-tion of results from knockout animals difficult. Second, BDNFTrkBsignaling does not exert the same actions in all brain regions. In thehippocampus and cortex, BDNFTrkB signaling promotes resilienceto stress, whereas in the NAc BDNF promotes susceptibility [38]. Forexample, disrupted BDNF signaling in the NAc renders mice moreresistant to chronic social defeat [169].
Altogether, it appears that the neurotrophin hypothesis of depres-sion is incomplete. In addition to the conflicting data in the literature,there is no established mechanism linking serotonin elevation tosubsequent increases in BDNF transcription and, therefore, noexplanation as to how ADs, such as SSRIs, activate this pathway.Nevertheless, these findings highlight the possibility that changes inneuronal strength and plasticity underlie depression and the action ofADs.
Trends in Neurosciences May 2015, Vol. 38, No. 5Changes in excitatory synapses in depressionA common element linking stress, serotonin, and neuro-trophins is their effects on excitatory synaptic transmis-sion [26]. In preclinical studies, there is increasingevidence that chronic stress exerts deleterious effects onexcitatory synaptic structure and function in multiplebrain regions associated with cognitive and emotionalcontrol of reward behaviors that resemble those seen inhuman depression. Conversely, serotonin and neurotro-phins exert an opposing action, generally promoting excit-atory synaptic transmission in the same brain regions.Many symptoms of depression seem to result from a dys-function in the valuation of stimuli, with decreases inpositive valuation (i.e., anhedonia) occurring in parallelwith increases in negative valuation (disappointment, fear,or expectation of punishment). Changes in the strength ofvarious excitatory synapses in the distinct circuits medi-ating both of these processes have been described.
Nucleus accumbensThe NAc is critical for integrating cortical and hippocam-pal inputs and regulating the firing of neurons in theventral tegmental area (VTA) [27], thereby influencingthe motivation to seek rewarding stimuli [28], as well asmotor behavior. Lim et al. [29] made a particularly signifi-cant advance in our understanding of the neurobiology ofdepression with their study of the effects of chronic stresson excitatory synapses in the NAc (Figure 1). They ob-served that excitatory synapses in medium spiny neurons(MSNs) in the NAc that express the D1 dopamine receptor
281
(D1R) displayed a selective decrease in AMPA receptor(AMPAR)-mediated excitation after 5 days of chronic re-straint stress. The decrease in excitation occurred in par-allel with anhedonia in the sucrose preference test anddepressive-like changes in the tail suspension and forcedswim tests. Given that D1R-expressing MSNs promoteactivation of dopaminergic cells [30], decreased excitationof D1R cells should lower dopamine release, a key regula-tor of reward-seeking and motivated behavior.
Experimental manipulations that decreased AMPAR-mediated excitation in D1R-expressing cells mimicked thestress-induced behavioral changes in unstressed animals,and manipulations that prevented stress-induceddecreases in AMPAR-mediated excitation blocked somebehavioral changes (sucrose preference test) in stressedanimals [29]. Interestingly, other behaviors that are widely
in the electrophysiological responses of dopaminergic VTAneurons in chronic stress models remain controversial. Inone study, optogenetic activation of dopamine-releasingVTA neurons relieved depression-like symptoms in chron-ically stressed mice and, conversely, inhibition of VTAneurons triggered a depression-like behavioral phenotype[34]. Thus, the decrease in the ability of cortical andhippocampal inputs to excite D1R cells in the NAc afterchronic stress, described above, could underlie the reducedfiring of VTA neurons [35,36], reduced dopamine release[37], and the loss of the rewarding properties of variousstimuli in these models. Changes in synaptic strengthwithin the VTA itself could also contribute [34]. We predictthat restoration of normal behavior by ADs should beaccompanied by a restoration of synaptic strength, but thishas not yet been tested. However, other studies report that
00
80
60
40
Su
ine
reco
r (NM
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 5assayed in depression studies, such as the tail suspensiontest and the forced swim test, were not rescued by pre-venting AMPAR downregulation in the NAc. The authorsconclude that the stress-induced decrease in excitation issufficient and necessary for many behavioral changes thatresemble human depression symptoms, specifically anhe-donia, highlighting the central importance of these synap-tic circuits to the genesis of anhedonia. The source of theaxons forming these excitatory synapses was not identi-fied, but is likely to include afferents from the hippocam-pus, PFC, and amygdala. In addition to these postsynapticchanges in excitatory synaptic strength, decreases in thefrequency of miniature excitatory postsynaptic currents(mEPSCs), indicative of presynaptic changes in releaseprobability, have also been reported in D1R MSNs[31,32]. Accompanying these physiological changes, mod-est increases in the density of stubby dendritic spines areobserved in susceptible mice subjected to chronic socialdefeat [31,33], but the identity of the MSNs was notdetermined.
Ventral tegmental areaThe cell bodies of dopamine-releasing cells enervating theforebrain are located in the VTA, intermingled with apopulation of local GABAergic interneurons, both of whichreceive input from NAc MSNs and the PFC [27]. Changes
Pref
eren
ce (%
)
1
AMPA
:NM
D ra
o
6
4
2
Control Stressed
70 mV
+40 mV
NAc D1RMSN
(B)(A)
ControlStressed
*
Figure 1. Stress-induced internalization of AMPA receptors (AMPARs) in D1 dopam
underlies some stress-induced depression-like behavioral changes. (A) Whole-cell
neurons (MSNs) at 70 mV and +40 mV for analysis of AMPAR- and NMDA receptoChronic stress decreased the AMPAR component, but not the NMDAR component. (B,C
AMPAR internalization prevented the stress-induced loss of sucrose preference (B),
permission, from [29]. Abbreviation: n.s., not significant.
282VTA discharge is increased in chronic stress models [3840] and that optogenetic activation of VTA neurons pro-motes depression-like behavior [41]. Given that: (i) drugsthat elevate dopamine, such as cocaine, are not pro-depres-sive; (ii) optogenetic stimulation of the PFC exerts anti-depressive effects [42,43]; and (iii) optogenetic activation ofD1R MSNs promotes resilience to social stress-induceddepression [32], it is unclear how the observed increasesin VTA firing are generated and why they should be pro-depressive. Further work is needed to reconcile thesedifferences, perhaps by identifying subpopulations ofinputs, synapses, or neurons within the NAc.
Prefrontal cortex[There are many subdivisions of the PFC (infralimbic,prelimbic, orbital, dorsolateral, and ventromedial) and,for simplicity, we omit any distinction of these regions.]The PFC is an important site at which cognitive evalua-tions, such as the controllability of a stressor [44] or thepleasantness of a stimulus [45], can influence affect andreward. Patients with depression display evidence of re-duced activity in the PFC and SSRI treatment restoresnormal activity levels [23,46]. In animal models, chronicstress results in atrophy of distal dendrites of pyramidalcells in the PFC and loss of dendritic spines [47,48]. Yuenet al. [49] reported that 7 days of chronic restraint stress
Tim
e im
mob
ile (s
)
100
200
Stress Stress
crose preference Forced swimn.s.
(C)
ControlControl pepde
Acve pepde
Untransfected
ControlControl pepde
Acve pepde
Untransfected
* * *
TRENDS in Neurosciences
receptor (D1R)-expressing medium spiny neurons in the nucleus accumbens (NAc)
rding of evoked excitatory synaptic currents from D1R-expressing medium spiny
DAR)-mediated components in slices from control and chronically stressed mice.) Transfection of D1R-expressing cells with a peptide that prevents stress-induced
but not stress-induced immobility in the forced swim test (C). Modified, with
produced a decrease in synaptic excitation in layer Vpyramidal cells in the PFC. Both AMPAR- and NMDAreceptor (NMDAR) components of the EPSP were equallyaffected, unlike in the NAc [29] and hippocampus (seebelow), and levels of AMPA A1 (GluA1) and NMDA N1(GluN1) glutamate receptor subunit proteins were down-regulated in parallel (Figure 2). The decreased expressionof GluA1 subunits was mediated by ubiquitination andproteosomal degradation. In addition to this postsynapticmechanism, there may also have been a decrease in pre-synaptic release probability (decreases in mEPSC frequen-cy, albeit no change in the paired-pulse ratio of evokedEPSPs). No change in postsynaptic density protein 95(PSD95) expression was detected, leaving uncertain thequestion of whether changes in synapse number and pre-synaptic function accompany the observed postsynapticchanges. Yuen et al. also observed no change in GluA1function or expression in the hippocampus with 7 days ofchronic restraint stress, a time at which GluA1 and GluN1are decreased in PFC, suggesting that the PFC is moresusceptible to stress. Consistent with the decrease inexcitation, expression of several activity-dependent genesis decreased in the PFC of humans with depression and inrodent models [42].
Activation of serotonin 2A receptors (5-HT2AR) triggersa rapid increase in the frequency of spontaneous glutama-tergic excitatory postsynaptic responses in pyramidal cellsin the PFC that is mediated by the excitation of a subset of
pyramidal cells in layer VVI, while evoked EPSCs aredepressed simultaneously [5052]. The effect of serotoninon mEPSC frequency is diminished after chronic stress[53], although the authors failed to report whether there isany change in the basal frequency or amplitude after stressor whether serotonin is less effective at exciting layer VVIcells. One potential explanation is that the effects of sero-tonin are weaker because the intracortical excitatory syn-apses formed by the deep layer cells are weakened bystress, although this has not been tested.
Lateral habenulaThe lateral habenula (LHb) has received increased atten-tion in the context of depression because of behavioralstudies indicating that it signals negative reward (i.e.,disappointment) [54,55]. The LHb forms a glutamatergicmonosynaptic excitatory projection directly to both dopa-minergic neurons in the VTA and serotonergic neurons inthe dorsal raphe (DR), as well to the mesopontine rostro-medial tegmental nucleus (RMTg), which then sends sub-stantial GABAergic projections to the VTA and DR[56,57]. Stimulation of the LHb inhibits discharge ofVTA neurons [58] and also DR cells. When animals per-form tasks with various rewarding, neutral, or punishingstimuli, neurons in the LHb are excited by both the absenceof an expected reward or the presence of a punishment, andthey are inhibited by the rewarding stimulus, especiallywhen the stimuli are unpredictable [59]. Thus, induction of
5
nd
DAR
s (o
s of
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 5AMPA
resp
onse
am
plitu
de
Smulaon strength
Control
Stress
(A)
AMPA
resp
onse
am
plitu
de
Cont Stress
Total Surface
GluA1
GluN1
Acn
(D)(C)
300
250
200
150
100
50
5 6 7 8 9
***
#
300
250
200
150
100
50ControlStress
ControlStress
9V
8V
7V
Figure 2. Stress-induced internalization of AMPA and NMDA receptors (AMPARs a
degradation. AMPAR-mediated [(A) 70 mV, representative currents at right] and NM
response to a range of stimulation intensities in slices from unstressed control rat
unpredictable stress (filled square). Stress depressed both components regardlesplasma membrane surface insertion of glutamate A1 (GluA1) and GluN1 receptor subuni
in AMPAR-mediated synaptic currents in response to chronic stress, but had no effe
permission, from [49].NM
DA re
spon
se a
mpl
itude
Smulaon strength
Control
Stress
(B)
Smulaon strength
Prot
eoso
me
inhi
bito
rCo
ntro
l
Control Stress
6# *
* *
# ** *
7 8 9
300
250
200
150
100
50
5 6 7 8 9
TRENDS in Neurosciences
NMDARs) in the prefrontal cortex is mediated by ubiquitination and proteosomal
-mediated [(B) +60 mV] excitatory currents recorded from layer V pyramidal cells in
pen circle) and rats subjected to chronic restraint stress (filled triangle) or chronic
stimulation intensity. (C) Chronic restraint stress decreased both expression andts. (D) Pharmacological inhibition of proteosomal degradation prevented decreases
ct in unstressed animals. Representative currents shown at right. Modified, with
283
LHb discharge by negative reward would inhibit bothdopamine and serotonin secretion.
In the learned helplessness model, LHb neurons dis-played an increased frequency of mEPSCs compared withnave controls with no change in their mean amplitude,reflecting an increased probability of glutamate releasefrom the nerve terminals of some population of presynapticneurons [60]. Consistent with this increased excitation, thespontaneous action potential discharge rate of LHb neu-rons was significantly higher after learned helplessness.Thus, these changes could suggest a physiological basis forexcessive disappointment or negative thinking in depres-sion. By contrast, patients with depression subjected toacute tryptophan depletion to induce depressive symptomsdisplayed significant decreases in activity-dependent posi-tron emission tomography (PET) signals in the LHb thatwere closely correlated with the occurrence and severity ofthe depressive symptoms [61].
AmygdalaThe amygdala is critical for learning to fear aversivestimuli and learning that previously aversive stimuli areno longer dangerous [62]. Unlike the hippocampus, theamygdala displays increases in volume in patients withdepression [24] and is hyperactive [63]. Unlike the hippo-campus and PFC, chronic stress induces dendritic hyper-trophy in the amygdala [64]. Relatively little is knownabout changes in synaptic function to, from, and withinthe amygdala in chronic stress models, but optogeneticactivation of amygdala inputs to the ventral hippocampuscan impair normal social behaviors [65].
HippocampusChronic stress induces changes in synaptic function thatoccur in parallel with the atrophy of distal apical dendriticbranches in pyramidal cells and decreases in the size andnumber of dendritic spines [66,67]. While stress has re-peatedly been shown to affect excitatory synapses in thehippocampus, published results are somewhat conflicting.At Schaffer collateral (SC) synapses onto CA1 cells, chronicstress is reported to have no effect on basal synapticstrength, although it does impair long-term potentiation(LTP), decreases GluA1 mRNA [6870], and enhancessynaptic currents mediated by NMDARs, while leavingAMPAR-mediated excitation unaltered in area CA3 [71].
One factor that could reconcile these discrepancies is thedendritic location of the synapses. Stress-induced changeshave recently been described in synaptic function at thesynapses in distal apical dendrites of CA1 pyramidal cellsformed by temporoammonic (TA) inputs from entorhinalcortex [72] (Figure 3). The authors observed that 36 weeksof chronic unpredictable stress (CUS), which is sufficient toproduce depression-like behavior in the sucrose preferencetest and other behavioral assays, resulted in a 50% de-crease in AMPAR-mediated excitation at TA-CA1 synap-ses. NMDAR-mediated excitation at TA-CA1 synapses wasunaffected, as was AMPAR-mediated excitation at neigh-boring SC-CA1 synapses. Accompanying the decreased
Feature Reviewexcitation was a corresponding layer-specific decrease inthe expression of GluA1 and PSD95 proteins. No changesin the levels of GluA2 or GluN1 proteins were detected.
284This synaptic pathology had deleterious behavioral con-sequences, as shown by the ability of CUS to disrupt long-term memory consolidation in a spatially cued Morriswater maze task, a function of TA-CA1 synapses [73]. Fur-ther strengthening the relation between this synapticpathology and the behavioral alterations produced byCUS, it was observed that changes in both AMPAR-medi-ated excitation and GluA1 protein levels were restored tonormal levels in stressed animals treated chronically, butnot acutely, with fluoxetine, just like the restoration ofnormal reward behavior.
Differences in the ability of allostatic load to affectexcitatory synaptic function may have a role in the deter-mination of susceptibility to becoming depressed. Schmidtet al. [70] examined the difference between mice that werevulnerable to behavioral changes after chronic social stressand those that were resilient. Vulnerable mice exhibited asignificant decrease in GluA1 and increase in GluA2 sub-units in area CA1, compared with resilient mice. Theauthors further examined genetic polymorphisms in genesencoding GluA1 subunits and found a SNP that correlatedwith vulnerability to stress. How this compares to humanGluA1 gene SNPs was not described, but this evidenceconverges with human data implicating AMPARs [74,75].
It has also been observed that chronic stress quantita-tively and qualitatively changes the way TA-CA1 excitato-ry synapses respond to serotonergic stimulation[76]. Whereas activation of postsynaptic 5-HT1BRs pro-duces a doubling of AMPAR-mediated excitation in controlanimals that recovers within ca. 30 min of washing out theagonist, activation of 5-HT1BRs in chronically stressedanimals produces a potentiation that is both significantlygreater in magnitude and more persistent (>12 h) afterremoval of the agonist (Figure 4). Normal transientresponses to 5-HT1BR activation are restored in CUSanimals administered fluoxetine chronically, but notacutely. The molecular changes in the signaling pathwaysthat underlie 5-HT1BR-induced potentiation that underliethis effect have not yet been determined, but they must beconfined to excitatory synapses in the postsynaptic cell,where this potentiation is induced and expressed, to ac-count for the stress-induced changes observed.
Not only do chronic stress models produce changes inglutamatergic receptor function and expression, but thesechanges may also be sufficient to cause changes in beha-viors related to depression. For example, Chourbaji et al.[77] found that knocking out the GluA1 gene globallyresulted in a depression-like phenotype in the learnedhelplessness test. Similarly, it has been reported that micein which the GluA1 gene is mutated so that it can no longerbe phosphorylated by calcium/calmodulin-dependent(CaM) kinase display a depression-like phenotype in thesucrose preference and novelty suppressed feeding tests,but not in the tail suspension test [76]. Together, thesedata make a strong argument that disruption of hippocam-pal glutamatergic transmission, specifically AMPAR func-tion, can accompany depression-like behavioral changes.
Trends in Neurosciences May 2015, Vol. 38, No. 5Human studiesAlthough less extensive than the literature on serotoninand its receptors, there is some evidence for changes in
Imipramine
100
200
Syna
pc
stre
ngth
(% c
ontr
ol)
150
50
250
300
Time (min)0 50 100
+ 5-HT1B R antagonist
Control
Before
+ 5-HT1B Ragonist
5-HT1BRagonist
Unstressed control
Stressed rats
100
200
300
Syna
pc
stre
ngth
(% c
ontr
ol)
Time (min)0 20 40 60 80 100 120 140
(B)(A) (C)
TRENDS in Neurosciences
Figure 4. Serotonin causes a serotonin 1B receptor (5-HT1BR)-dependent potentiation of temporoammonic (TA)CA1 synapses and this potentiation is altered by chronic
stress. (A) Field excitatory postsynaptic potentials (fEPSPs) are recorded in stratum lacunosum moleculare in response to stimulation of the TA pathway during application
of the tricyclic antidepressant imipramine in control saline (black) or saline containing the 5-HT1BR antagonist isamoltane. Elevation of endogenous serotonin produces a
doubling of synaptic strength. (B) Activation of 5-HT1BRs with the selective agonist anpirtoline promotes action potential discharge in CA1 cells. (C) Anpirtoline produces a
robust and reversible potentiation of TACA1 excitatory postsynaptic currents in slices from unstressed control animals, but produces an enhanced and persistent
potentiation in slices from rats subjected to chronic unpredictable stress. Modified, with permission, from [76].
*
Control CUS Stress +AD
GluA
1 ex
pres
sion
0
25
50
75
100
125
0
1
2
3
4
5
6
7
AMPA
:NM
DA ra
o
Control CUS Stress +AD
**
1 mV1 ms
Control Stre ss Stress + AD
0
0.1
0.2
0.3
0 0.1 0.2 0.3 0.4 0.5
AMPA
resp
onse
am
plitu
de
Smulaon strength
Stress
Control
(A)
(C)(B)
TRENDS in Neurosciences
Figure 3. Stress decreases AMPA receptor (AMPAR)-mediated excitation and glutamate A1 (GluA1) expression at temporoammonic (TA)CA1 cell synapses and chronic
fluoxetine reverses these effects. (A) Field excitatory postsynaptic potentials (fEPSPs) are recorded in stratum lacunosum moleculare in response to stimulation of the TA
pathway in Mg2+-free saline, allowing dissection of AMPAR- and NMDA receptor (NMDAR)-mediated components. AMPAR-mediated transmission was reduced over a
range of stimulation intensities after chronic unpredictable stress (CUS; red) compared with unstressed rats (blue). Quantification of AMPA:NMDA ratios (B) and GluA1
protein (C) reveals that the selective decrease in AMPAR-mediated transmission was due to decreased GluA1 expression. Administration of fluoxetine (antidepressant; AD)
for 3 weeks to stressed rats restored both AMPA:NMDA ratios and GluA1 expression. Modified, with permission, from [72].
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 5
285
glutamatergic function in patients with depression(Table 1), much of it mixed. Plasma levels of glutamateare decreased in some studies [78,79], but not others [80]. Aconsiderable difficulty in studies of this type is that gluta-mate is present in a large metabolic pool, rendering itimpossible to know what fraction of the measured gluta-mate is synaptic in origin. Postmortem studies revealeddecreases in multiple glutamate receptor subunits inpatients with depression, including the GluA1 subunit[81,82], as well as decreased AMPAR binding in the stria-tum of suicide completers [74,75]. Hopefully, reliable bio-markers can be found with which to more directly assayglutamatergic function in patients with depression.
Formal statement of the excitatory synapse hypothesis1. Major depression is caused by a weakening of specific
subsets of excitatory synapses in multiple brain regionsthat are critical in the determination of affect andreward. Chronic hyperactivity of the HPA axis inresponse to excessive stress, known to be an importantallostatic risk factor in the gene environmental axisdetermining susceptibility to depression, is one poten-tial mediator of these changes. High levels of GRs incells in these regions contribute to their vulnerability.
2. Many of the characteristic changes in behavior thatdefine the symptomatology of human depression, suchas anhedonia and depressed mood, result because
depression, such as sleep, sex drive, sociality, workingmemory, and attention, are similarly affected.
3. Restoration of excitatory synaptic strength is thecritical action of effective antidepressants, includingboth conventional agents, such as SSRIs and ECT, andnewer compounds, such as ketamine.
Several authors have highlighted stress-induced patho-physiological changes in reward circuits as the biologicalunderpinning of anhedonia and other symptoms of depres-sion [8385]. There is a strong correlation between anhe-donia and suicide [86,87]. Indeed, anhedonia has beenfound to be a more significant predictor of imminent suicidethan any other psychiatric variable [88]. Social anhedonia(e.g., loss of interest in social interaction) appears especial-ly predictive of suicidal ideation [87]. We can predict howchanges in excitatory synapses in regions such as thehippocampus and PFC might interact with reward circuitsin the corticomesolimbic system to change affective beha-viors and produce a depression-like behavioral phenotype(Figure 5).
One important target of outputs from both the hippo-campus and PFC is the NAc, and activation of either brainregion is positively correlated with NAc activity (e.g., [89]).One important target of the NAc is the VTA. Numerousstudies have shown that increased dopamine release iscritical for appetitive motivational processes, such as ini-
ing
alle
t in
DNF
corr
C i
ptominbfie
vera
ith
ing
late
ess
in ippo
s inract
g s
ine bol
moedo
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 5impaired excitatory synaptic transmission leads toreduced activity in the corticomesolimbic rewardcircuitry. Other circuits that are important in themediation of other behaviors that are altered in
Table 1. Summary of depression related changes in humans
Refs Protein or gene Major finding
Proteins and genes
[170] NMDAR Abnormal bind
[171] 5HTT gene Promoter short
[172] BDNF Val66Met varian
[173] BDNF (review) Lower serum B
[174] 5-HT1AR (review) C(-1019)G SNP
[175] mGluR5 Decreased in PF
[176] Multiple synaptic proteins, 5-HTRs Decreased syna5-hydroxytryptahippocampal su
In vivo PET findings in patients with depression
[175] mGluR5 Reduction in se
[177] 5-HT1AR (review) Mixed results, w
[197] 5-HT1BR Decreased bind
[179] 5-HT2R Reduction corre
[180] 5-HTT (review) Mixed results: l
[181] MAO type A Greater bindingstriatum, and h
Metabolism and activity (fMRI)
[182] N/A Hypoactive areastructures hype
[183] N/A Disrupted restin
[184] N/A Increased baseldecreased meta
[185] N/A Subgenual ACCfeedback in anh[46] N/A Successful SSRI tredorsal ACC
[186] N/A Reduced activation
286tiation of behavior, exertion of effort, sustained task en-gagement, and instrumental learning [90], precisely thebehaviors that are disturbed in depression. MSNs in theNAc are GABAergic and the D1R-expressing subset of
in brains of suicide completers
le variant promotes depression and suicidality after early life stress
teracts with stress to promote depression
in patients with depression, normalized with AD treatment
elates with MDD, suicidality, and decreased response to SSRIs
n unmedicated depressed postmortem samples
somal-associated protein, 25kDa (SNAP25), GLUR1, GLUR3; increasede (serotonin) receptor 2C (HTR2C), decreased HTR4 and HTR7 inlds of patients with depression
l regions, including hippocampus and PFC
clear trend for decrease in many brain areas
in ventral striatum and pallidum
s with ECT effect
binding or no effect
many areas, including PFC, anterior cingulate cortex (ACC), ventralcampus
clude frontal and temporal cortices; select subcortical and limbicive (meta-analysis)
tate interhemispheral coherence
metabolism of orbitofrontal cortex, amygdala, posterior cingulate;ism subgenual ACC and dorsal PFC
re active, dorsal ACC less active in response to reward-relatednic individuals
atment reverses metabolism deficits in dorsal PFC, medial PFC, and of temporal and occipital cortices during emotional task
sse
luta
mP
1
trans
d) to
latio
e NA
ase
e ro
essio
sulti
do
imul
, an
,70,7Normal(A) Depre
GABAKey: g
NAc
VTA VTA
LHb
RMTg
Hippo mPFC Hippo
3
(B)
Figure 5. The hypothesized impact of stress-induced changes in excitatory synaptic
The hippocampus and prefrontal cortex (PFC) send glutamatergic projections (re
expressing GABAergic medium spiny neurons (MSNs). These cells inhibit a popu
inhibitory synapses (blue) onto dopaminergic VTA cells (green) projecting back to th
cells, thereby disinhibiting VTA dopaminergic cells and promoting dopamine rele
glutamatergic cells in the lateral habenula (LHb) project to GABAergic neurons in th
VTA. Chronic stress induces anhedonia and other reward-related symptoms of depr
on D1R MSNs. There is also a potentiation of excitatory inputs onto LHb neurons, re
inhibition of VTA cells. These changes synergistically increase the inhibition of VTA
therapies [selective serotonin reuptake inhibitors (SSRIs), ketamine, deep brain st
serotonin-dependent strengthening of excitatory synapses in the hippocampus, PFC
stimuli (C). Evidence of changes in synaptic excitation: 1 [29]; 2 [4749,53,104]; 3 [69
[76]; tricyclic [189]; 7 ketamine [104]; 8, long-term potentiation (LTP)? [42].
Feature ReviewMSNs activates dopaminergic cells by inhibiting intrinsicVTA GABAergic cells [30]. Thus, the net effect of activationof D1R cells in response to inputs from the hippocampusand PFC is to promote dopamine release. Stress-inducedweakening of excitatory synapses onto D1R cells [29] wouldmake it harder for rewarding stimuli to trigger the normalactivation of dopaminergic VTA neurons and the release ofdopamine. Hence, D1R MSNs would be less effective ateliciting positive reinforcement or motivated behavior.Similarly, weakened excitatory synapses in the hippocam-pus and PFC would decrease the afferent drive that theNAc receives from these regions [91] and also diminish therelease of dopamine by reward-reinforcing stimuli (e.g.,[37,92]). Decreased drive to the amygdala from the PFCand hippocampus may also be important. These changesoccur in parallel to, and are synergistic with, the increasedinhibition of VTA cells mediated by changes in the LHbRMTg circuit, as discussed above.
It is important to note that we do not predict that everyexcitatory synapse is affected by stress, even in the sameregion. In the NAc, synapses on D1R cells are affected, butnot those on D2R cells [29,32]. In the hippocampus, TA-CA1 synapses are weakened by chronic stress, but SCsynapses are not [72]. Second, not every stress-sensitiveexcitatory synapse responds in the same manner. Forexample, synapses involved with positive valuation aremore likely to be weakened by stress, whereas synapsesinvolved in negative valuation become strengthened. Moresynapses need to be examined to clarify these issues.Finally, the range of behaviors that are altered in stressand depression provides clear evidence that there arelikely to be multiple synapses that are adversely affectedd
mate dopamine
AD treated
NAc
FC Hippo mPFC
NAc
VTA
2 6 7
8
LHb
RMTg
LHb
RMTg
4
5
TRENDS in Neurosciences
(C)
mission on the reward circuitry and the mechanisms of antidepressant (AD) action.
the nucleus accumbens (NAc), where they excite D1 dopamine receptor (D1R)-
n of interneurons within the ventral tegmental area (VTA) that form GABAergic
c, hippocampus, and PFC. When the hippocampus or PFC is active, they excite D1R
(A). There is a distinct circuit mediating responses to negative stimuli, in which
stromedial tegmentum (RMTg), which in turn inhibit dopaminergic neurons in the
n because of a weakening of excitatory synapses in the hippocampus and PFC and
ng in an increase in the excitation of RMTg neurons and subsequent increase in the
paminergic cells, thereby impairing dopamine release (B). Effective antidepressant
ation (DBS), or electroconvulsive therapy (ECT)] can all trigger an activity- and/or
d NAc, thereby restoring the normal release of dopamine in response to rewarding
2,187,188]); 4 [60]); and 5 [178]). Evidence of synaptic strengthening by ADs: 6 SSRI
Trends in Neurosciences May 2015, Vol. 38, No. 5by stress, with no one synapse likely to be uniquely re-sponsible for generating the depressed phenotype.
Restoration of excitatory synaptic strength within thehippocampus and PFC, as well as restoration of thestrength of excitatory synapses onto D1R cells, wouldrestore the ability of rewarding stimuli to promote dopa-mine secretion. Thus, our hypothesis predicts that anyagent or manipulation that strengthens key stress-sensi-tive excitatory synapses in the reward system may be aneffective AD. Indeed, optogenetic stimulation of the PFCexerts AD-like effects in the sucrose preference and socialinteraction tests in rodent models [42,43], equivalent to theeffects of chronic SSRIs. Furthermore, direct optogeneticactivation of D1R MSNs promotes resilience to socialstress-induced depression [32].
The excitatory synapse is also a place where serotoninand brain-derived neurotrophic factortropomyosin recep-tor kinase B (BDNFTrkB) receptor signaling converge.BDNF secretion is promoted by glutamatergic excitation.Considerable evidence ties BDNF and TrkB to synapse-strengthening processes, such as LTP. Application ofBDNF also results in morphological and functionalchanges that parallel LTP, including increased spine for-mation and higher amplitudes and frequency of miniatureEPSCs in CA1 neurons in the hippocampus [93]. Finally,there is considerable evidence that BDNFTrkB signalingpromotes the generation and survival of granule cells inthe dentate gyrus [94]. Proliferation of these cells and theirincorporation into the hippocampal circuitry should alsopromote activity within hippocampal circuits, thereby pro-moting hippocampal throughput and output. Thus,BDNFTrkB signaling is both promoted by excitation
287
and promotes excitation. Similarly, insufficient serotonin(Box 1) may lead to loss of on-going maintenance of synap-tic strength (see below). Conversely, insufficient synapticstrength may weaken the drive of neurons in the DR fromthe limbic cortices [95] and diminish serotonin release.
Therapeutic implications of the excitatory synapsehypothesisAlthough the development of SSRIs has improved thetreatment of major depression, there is still considerableneed for better therapeutic options. Can the excitatorysynapse hypothesis account for the actions of existingtherapies and point towards new strategies?
SSRIsOur understanding of how SSRIs relieve depressiondepends upon understanding how serotonin affects brainfunction. Two effects appear central to understanding thebeneficial actions of SSRIs: promoting neurogenesis andpromoting excitatory interactions at specific loci. There isstrong evidence that SSRIs promote the proliferation ofprogenitor cells in the subgranular zone of the dentategyrus, leading to the increased genesis and survival ofdentate granule cells [96]. Given the evidence that chronicstress impairs neurogenesis of these same cells [97], it hasbeen proposed that restoration of neurogenesis is a criticalfactor in the AD action of SSRIs [98], presumably byincreasing hippocampal output, although this is rarelydiscussed.
There is also considerable evidence that serotonin reg-ulates excitatory synaptic transmission in various brainregions through actions at multiple serotonin receptorsubtypes. In the hippocampus, elevation of serotonin levelswith SSRIs potentiates excitation at both mossy fiberCA3cell synapses, via an action at 5-HT4Rs [99], and TACA1cell synapses, via an action at 5-HT1BRs [76]. Furthermore,the ability of fluoxetine to restore normal sucrose prefer-ence and novelty suppressed feeding behaviors is compro-mised when 5-HT1BRs are blocked pharmacologically ordeleted genetically, suggesting that this potentiation isrequired for the therapeutic actions of SSRIs [76]. As dis-cussed above, serotonin also promotes excitatory synaptictransmission in the PFC.
Ketamine and NMDAR antagonistsOne of the greatest recent advances in our understandingof depression was the finding that antagonists of NMDAreceptors, particularly ketamine, have rapid, robust, andsustained (710 days) AD-like effects in humans[100,101]. Multiple studies have shown that acute admin-istration of ketamine, at doses that are said to be lowerthan those that produce psychotomimetic responses, pro-duces improvement in mood and reduces depressivesymptoms, including suicidal ideation, within 12 h;these improvements persist for up to 2 weeks. Rodentstreated acutely with ketamine exhibited an AD-like phe-notype in the forced swim [102], novelty suppressedfeeding, and learned helplessness tests [103]. Further-
Feature Reviewmore, ketamine restored sucrose preference and noveltysuppressed feeding behaviors rapidly in chronicallystressed animals [104].
288Although an area of on-going investigation, the gener-ally accepted model of the therapeutic effect of ketamine isconsistent with the hypothesized involvement of excitatorysynapses in depression. The actions of ketamine can bedivided into an induction phase of 12 h in duration,during which it is present in the brain at sufficient con-centrations to inhibit NMDARs, and an expression phase,in which symptoms are relieved for 12 weeks after wash-out of the drug from the brain.
There are two hypotheses about how ketamine actsduring the induction phase. Ketamine may exert a prefer-ential inhibition of NMDARs on GABAergic inhibitoryinterneurons [105107]. This could be due to differencesin subunit composition or because interneurons are moredepolarized than are pyramidal cells, relieving ion channelblock by Mg2+ and allowing NMDARs to contribute more totheir overall excitation. Thus, ketamine produces a milddisinhibition of the neuronal population, an increase inhigh-frequency oscillatory activity in rats [108,109] andhumans [110,111], and a neurochemically detectable surgeof glutamate release in the PFC and NAc [105,112,113]. Co-incident with the washout of ketamine at the end of theinduction phase, glutamate concentrations return to nor-mal [114]. Ketamine has recently been shown to exert anantidepressant-like action in the forced swim test in micelacking NMDARs in a subset of GABAergic interneurons[115], which could indicate that another population ofinterneurons is critical or that its actions do not dependon blocking interneuron NMDARs. Unfortunately, theauthors did not determine whether ketamine was ableto elicit increased activity in these animals.
The alternative hypothesis states that ketamine blockson-going activation of NMDARs mediated by spontaneoustransmitter release [116]. This on-going activity is postu-lated to produce constitutive suppression of signaling path-ways that promote excitatory synaptic transmission. Byrelieving this suppression, ketamine may promote expres-sion of several proteins that potentiate excitatory synap-ses, including AMPARs [104,116].
Regardless of which hypothesis is true, both share thestrengthening of excitatory synapses, persisting through-out the expression phase, as their common effector mecha-nism. Several activity-dependent consequences of a periodof increased network activity have been implicated, such asactivation of the mammalian target of rapamycin (mTOR)signaling pathway or deactivation of the eukaryotic trans-lation elongation factor 2 (EEF2) pathway, and rapidtranslation of proteins from pre-existing mRNAs, as wellas induction of the immediate early gene FBJ murineosteosarcoma viral oncogene homolog B (FosB) and itsdownstream genes. All three mechanisms ultimately leadto direct potentiation of synapses, the induction of synapse-related genes, and increased synthesis of synaptic proteinswithin hours [103,116], thereby strengthening patholog-ically weakened excitatory synapses [104]. Ketamineappears to induce LTP-like processes, as suggested bythe finding that ketamine triggers an increase in surfaceexpression of AMPARs [117]. Indeed, ketamine loses its
Trends in Neurosciences May 2015, Vol. 38, No. 5AD effect in the absence of functional AMPARs[117,118]. Ketamine also promotes BDNF synthesis andrelease, which may be required for its actions [116].
Unfortunately, the adverse effects of ketamine havehindered its clinical usefulness. Considerable effort isnow being made to develop other inhibitors of NMDARfunction that might preserve the rapid AD actions ofketamine without its adverse effects. Alternatively, thehypothesis predicts that any compound that either pro-motes LTP-like processes or promotes ketamine-likeincreases in oscillatory activity by any other means wouldalso strengthen excitatory transmission and exert an ADaction. Indeed, Kumar et al. [119] have shown that rhyth-mic optogenetic activation of the PFC exerts an AD-likeeffect in the forced swim test and increases oscillatoryactivity throughout the mesolimbic reward circuitry, in-cluding the VTA and NAc [119].
ScopolamineThe nonselective antagonist of muscarinic acetylcholinereceptors, scopolamine, exerts an antidepressant actionthat is slower than ketamine, but faster than SSRIs (ca.35 days) [120]. Similar to ketamine, scopolamineincreases signaling via the mTOR pathway in the PFCand increases the number and size of dendritic spines onlayer V pyramidal cells, as well as the ability of serotonin toincrease the frequency and amplitude of spontaneous glu-tamatergic synaptic currents [121]. These responses areprobably triggered by a short burst of neuronal activity,as suggested by the detection of elevated glutamatelevels with microdialysis, as seen with ketamine. Indeed,
blocking PFC AMPARs was sufficient to prevent the abilityof scopolamine to produce an AD-like response in the forcedswim test.
Positive allosteric modulators of AMPARsDrugs that block AMPAR desensitization and/or deactiva-tion are one promising way to exert AD action by directlypotentiating excitatory synapses. The AMPAR potentiatorLY392098, for example, exerted AD-like actions in the tailsuspension and forced swim tests, although it did notrestore sucrose preference after chronic stress[122,123]. In addition to their direct effects on excitation,AMPAR potentiators increase BDNF mRNA in the hippo-campus [124]. Finally, AMPAR potentiators may be usefulin accelerating the therapeutic response to SSRIs [125].
Deep brain stimulationDeep brain stimulation (DBS) has attracted considerableattention as a means to alleviate treatment-resistant de-pression, although the mechanisms underlying its thera-peutic effects and specific loci of action remain poorlyunderstood. In humans, repetitive stimulation of the sub-callosal cingulate gyrus has been shown to be therapeuticin some patients [126]. Similarly, in preclinical models,optogenetic and electrical stimulation of the medial PFChas an AD effect [42]. DBS of the rat PFC gray matter alsocaused an AD-like response in the forced swim, noveltysuppressed feeding, and sucrose preference tests [127].
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 5Box 3. Outstanding questions
How does chronic stress impair synaptic structure and function?Changes associated with stress have been attributed to a variety ofmediators, including corticosteroids, peptide hormones [e.g., corti-cotrophin-releasing hormone (CRH)], opioid peptides, monoamines(e.g., serotonin or dopamine), inflammatory cytokines, endocanna-binoids, and neurosteroids, to name a few [190]. Glucocorticoids areby no means singly important but are likely the best known. In oneexample, chronic administration of corticosterone to nave rats andmice was sufficient to produce depressive-like behavior andrecapitulate several molecular markers of chronic stress models[191], presumably via glucocorticoid receptors (GRs). Neuronsthroughout the cortex, hippocampus, and ventral striatum expressGRs at high levels, rendering them vulnerable to chronic stress. Thedownstream mediators of persistent GR activation are unknown butultimately alter expression of synaptic proteins through alteredsynthesis, turnover, and degradation [49]. What makes somesynapses selectively vulnerable? Distal dendrites are particularlysensitive to the atrophic effects of chronic stress and contribute, atleast in part, to the localization of the altered synaptic compositionand function [72,187]. Studies continue to unravel new roles for allthe known mediators of stress, sometimes with differing short- andlong-term actions, and to elucidate the molecular mediators betweenstress and impaired synaptic function.
How do SSRIs act in the NAc and LHb? Activation of 5-HT1BRstriggered a persistent inhibition of excitatory synaptic transmissionin the NAc in unstressed animals [192,193]. The hypothesis predictsthat this action of serotonin would promote, not relieve, thesymptoms of depression. It remains to be determined what theeffects of chronic SSRI or acute ketamine administration are onthese synapses. Similarly, do SSRIs renormalize plastic changes insynapses in the LHb and RMTg? This will be particularly informative
because the direction of the depression-associated synapticplasticity in this circuit (strengthening) appears opposite to thatseen in the hippocampus, PFC, and NAc (weakening). Why do SSRIs induce immediate elevation of serotonin levels [145]and rapid potentiation of excitatory synapses [76], but a delayedtherapeutic response? The hypothesis predicts that an immediatestrengthening of stress-weakened synapses should relieve symp-toms rapidly. Indeed, cocaine produces a rapid strengthening ofexcitatory synapses in the NAc [194] and an immediate euphoricresponse. It may be that multiple bouts of serotonin-inducedpotentiation are necessary to render them persistent, perhaps inconjunction with slower changes in neurotrophin signaling, for theeffects to be sustained. Only repeated administration of cocaine, forexample, leads to a lasting disinhibition [30] or increases in mEPSCfrequency [195] in VTA dopamine neurons. Alternatively, SSRIsmay elicit acute actions in other brain regions, such as theamygdala, that may oppose their actions in reward circuits. Notenough is known about the actions of SSRIs in mesolimbic rewardcircuits. Finally, it will be important to determine whether ADs thataffect norepinephrine and DA levels also produce changes in thesecircuits.
How and where does DBS work? The effects described in somepatients offer encouragement that DBS may offer sorely neededrelief for the patients who are most severely depressed and/ortreatment resistant [126], but the low rate of surgical successsuggests that we do not yet know enough about what structures orpathways to target and perhaps what stimulus parameters to use.Preclinical studies are likely to suggest the best places torenormalize reward circuitry.
Can compounds be developed that retain the beneficial rapid ADaction of ketamine without the adverse effects? Current efforts todevelop NMDAR antagonists with unique pharmacological proper-ties and subunit selectivity appear promising [196], although theadverse effect profile is not yet clear. If our understanding of how
ketamine works is correct, then any drug that promotes synchro-nized oscillatory activity in the PFC and hippocampus should alsohave an AD action, similar to that of ketamine.
289
The hypothesis described here predicts that repetitivestimulation, particularly of PFC and hippocampal inputsto the NAc, may induce activity-dependent synaptic poten-tiation or induce activity-dependent genes and transcrip-tion factors, similar to ketamine. Whether and where in thebrain these processes occur remain to be determined.Interestingly, direct stimulation of the ventral striatalreward circuitry may be equally efficacious in humanswith depression [128,129]. Finally, preclinical studies havesuggested that DBS acts indirectly by promoting the re-lease of serotonin [127].
Electroconvulsive therapyThere is a compelling overlap between direct methods ofexciting the brains of patients with depression to improvemood and the high-frequency stimuli that induce LTP atexcitatory synapses within and between many of the brainareas discussed above. Electroconvulsive therapy (ECT), inwhich a generalized seizure is induced by delivering 20120-Hz stimulation, is one of the oldest treatments fordepression still in use, although typically reserved forpatients resistant to pharmacological treatment, or wheresuch treatment is contraindicated, such as many olderpatients.
ECT-like stimulation in rats is known to induce an LTP-like potentiation of excitatory synapses in the dentategyrus that persists for up to 40 days [130]. Similar toconventional LTP, this ECT-induced potentiation is inhib-ited by an NMDAR antagonist [131] and is accompanied byan increase in phosphorylation of ser831 of GluA1 andS1301 of GluN2B. Changes in overall receptor expressionwere observed in some studies [132], but not others[133]. Interestingly, ECT also promoted an NMDAR-de-pendent increase in BDNF expression in animal models[134,135].
Concluding remarksIn conclusion, although there are many important ques-tions that remain (Box 3), there is considerable evidencethat dysfunction of excitatory synapses contributes to thepathology of depression and that many effective antide-pressants act to restore normal synaptic strength. Werecognize that the evidence in support of the hypothesisis far from complete and that the circuits are more complexthan we have indicated here. Our intent in formulatingthis hypothesis is to stimulate further work and to providea conceptual framework with which to interpret theresults. We hope that these efforts lead to a better under-standing of the causes of depression and thereby to bettertherapeutic options that increase the number of patientswho can be treated effectively and that their symptoms canbe relieved more quickly.
AcknowledgmentsWe are grateful to Bradley Alger for teaching us the importance of a wellformulated hypothesis. We thank our colleagues Mary Kay Lobo, BrianMathur, Todd Gould, Robert Schwarcz, and T. Chase Francis for theircomments on the manuscript. S.M.T. and X.C. were supported by grantR01 MH086828, A.J.K. and A.M.V. were supported by T32 GM008181,
Feature ReviewM.D.K. was supported by T32 NS063391, and T.A.L. was supported byT32 NS007375.
290References1 Kessler, R.C. et al. (2005) Lifetime prevalence and age-of-onsetdistributions of DSM-IV disorders in the National ComorbiditySurvey Replication. Arch. Gen. Psychiatry 62, 593602
2 Belmaker, R.H. and Agam, G. (2008) Major depressive disorder. N.Engl. J. Med. 358, 5568
3 Centers for Disease Control and Prevention (2009) Suicide Facts at aGlance, Centers for Disease Control and Prevention
4 Gartlehner, G. et al. (2011) Second-Generation Antidepressants in thePharmacologic Treatment of Adult Depression: An Update of the2007 Comparative Effectiveness Review, Agency for HealthcareResearch and Quality
5 Gaynes, B. and Warden, D. (2009) What did STAR* D teach us?Results from a large-scale, practical, clinical trial for patients withdepression. Psychiatr. Serv. 60, 14391445
6 Duman, R.S. and Aghajanian, G.K. (2012) Synaptic dysfunction indepression: potential therapeutic targets. Science 338, 6872
7 Nestler, E. et al. (2002) Neurobiology of depression. Neuron 34, 13258 Flint, J. and Kendler, K.S. (2014) The genetics of major depression.Neuron 81, 484503
9 Kendler, K. (1998) Anna-Monika-Prize paper. Major depression and theenvironment: a psychiatric genetic perspective. Pharmacopsychiatry31, 59
10 Billings, A.G. et al. (1983) Socialenvironmental factors in unipolardepression: comparisons of depressed patients and nondepressedcontrols. J. Abnorm. Psychol. 92, 119133
11 Korte, S.M. et al. (2005) The Darwinian concept of stress: benefits ofallostasis and costs of allostatic load and the trade-offs in health anddisease. Neurosci. Biobehav. Rev. 29, 338
12 Sapolsky, R.M. (1994) Why Zebras Dont Get Ulcers: A Guide to Stress,Stress Related Diseases, and Coping, W.H. Freeman
13 McEwen, B.S. (2000) Allostasis and allostatic load: implications forneuropsychopharmacology. Neuropsychopharmacology 22, 108124
14 Pariante, C.M. (2009) Risk factors for development of depression andpsychosis. Glucocorticoid receptors and pituitary implications fortreatment with antidepressant and glucocorticoids. Ann. N. Y.Acad. Sci. 1179, 144152
15 Holsboer, F. (2000) The corticosteroid receptor hypothesis ofdepression. Neuropsychopharmacology 23, 477501
16 Anacker, C. et al. (2011) The glucocorticoid receptor: pivot of depressionand of antidepressant treatment? Psychoneuroendocrinology 36,415425
17 Spijker, A. and van Rossum, E. (2012) Glucocorticoid sensitivity inmood disorders. Neuroendocrinology 95, 179186
18 Brown, E.S. (2009) Effects of glucocorticoids on mood, memory, andthe hippocampus. Treatment and preventive therapy. Ann. N. Y.Acad. Sci. 1179, 4155
19 Frol, A.B. et al. (2013) A comparison of clinician-ratedneuropsychological and self-rated cognitive assessments in patientswith asthma and rheumatologic disorders. Allergy asthma Proc. 34,170175
20 McEwen, B.S. (2012) Brain on stress: how the social environment getsunder the skin. Proc. Natl. Acad. Sci. U.S.A. 109 (Suppl.), 1718017185
21 Yun, J. et al. (2010) Chronic restraint stress impairs neurogenesis andhippocampus-dependent fear memory in mice: possible involvementof a brain-specific transcription factor Npas4. J. Neurochem. 114,18401851
22 Magarinos, A.M. and McEwen, B.S. (1995) Stress-induced atrophy ofapical dendrites of hippocampal CA3c neurons: comparison ofstressors. Neuroscience 69, 8388
23 Fales, C.L. et al. (2009) Antidepressant treatment normalizeshypoactivity in dorsolateral prefrontal cortex during emotionalinterference processing in major depression. J. Affect. Disord. 112,206211
24 Frodl, T. et al. (2003) Larger amygdala volumes in first depressiveepisode as compared to recurrent major depression and healthycontrol subjects. Biol. Psychiatry 53, 338344
25 Lorenzetti, V. et al. (2009) Structural brain abnormalities in majordepressive disorder: a selective review of recent MRI studies. J. Affect.Disord. 117, 117
26 Pittenger, C. and Duman, R.S. (2008) Stress, depression, and
Trends in Neurosciences May 2015, Vol. 38, No. 5neuroplasticity: a convergence of mechanisms. Neuropsycho-pharmacology 33, 88109
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 527 Sesack, S.R. and Grace, A.a. (2010) Cortico-basal ganglia rewardnetwork: microcircuitry. Neuropsychopharmacology 35, 2747
28 Smith, K.S. et al. (2011) Disentangling pleasure from incentivesalience and learning signals in brain reward circuitry. Proc. Natl.Acad. Sci. U.S.A. 108, E255E264
29 Lim, B.K. et al. (2012) Anhedonia requires MC4R-mediated synapticadaptations in nucleus accumbens. Nature 487, 183189
30 Bocklisch, C. et al. (2013) Cocaine disinhibits dopamine neurons bypotentiation of GABA transmission in the ventral tegmental area.Science 341, 15211525
31 Christoffel, D.J. et al. (2011) IkB kinase regulates social defeat stress-induced synaptic and behavioral plasticity. J. Neurosci. 31, 314321
32 Francis, T.C. et al. (2015) Nucleus accumbens medium spiny neuronsubtypes mediate depressionrelated outcomes to social defeat stress.Biol. Psychiatry 77, 212222
33 Golden, S.A. et al. (2013) Epigenetic regulation of RAC1 inducessynaptic remodeling in stress disorders and depression. Nat. Med.19, 337344
34 Tye, K.M. et al. (2013) Dopamine neurons modulate neural encodingand expression of depression-related behaviour. Nature 493,537541
35 Valenti, O. et al. (2012) Different stressors produce excitation orinhibition of mesolimbic dopamine neuron activity: responsealteration by stress pre-exposure. Eur. J. Neurosci. 35, 13121321
36 Belujon, P. et al. (2014) Restoring mood balance in depression:ketamine reverses deficit in dopamine-dependent synapticplasticity. Biol. Psychiatry 76, 927936
37 Cabib, S. and Puglisi-Allegra, S. (2012) The mesoaccumbensdopamine in coping with stress. Neurosci. Biobehav. Rev. 36, 7989
38 Krishnan, V. et al. (2007) Molecular adaptations underlyingsusceptibility and resistance to social defeat in brain rewardregions. Cell 131, 391404
39 Cao, J-L. et al. (2010) Mesolimbic dopamine neurons in the brainreward circuit mediate susceptibility to social defeat andantidepressant action. J. Neurosci. 30, 1645316458
40 Barik, J. et al. (2013) Chronic stress triggers social aversion viaglucocorticoid receptor in dopaminoceptive neurons. Science 339,332335
41 Chaudhury, D. et al. (2013) Rapid regulation of depression-relatedbehaviours by control of midbrain dopamine neurons. Nature 493,532536
42 Covington, H.E. et al. (2010) Antidepressant effect of optogeneticstimulation of the medial prefrontal cortex. J. Neurosci. 30, 1608216090
43 Vialou, V. et al. (2014) Prefrontal cortical circuit for depression- andanxiety-related behaviors mediated by cholecystokinin: role of DFosB.J. Neurosci. 34, 38783887
44 Amat, J. et al. (2005) Medial prefrontal cortex determines howstressor controllability affects behavior and dorsal raphe nucleus.Nat. Neurosci. 8, 365371
45 Berridge, K.C. and Kringelbach, M.L. (2013) Neuroscience of affect:brain mechanisms of pleasure and displeasure. Curr. Opin. Neurobiol.23, 294303
46 Kennedy, S.H. et al. (2001) Changes in regional brain glucosemetabolism measured with positron emission tomography afterparoxetine treatment of major depression. Am. J. Psychiatry 158,899905
47 Liston, C. et al. (2006) Stress-induced alterations in prefrontal corticaldendritic morphology predict selective impairments in perceptualattentional set-shifting. J. Neurosci. 26, 78707874
48 Radley, J.J. et al. (2008) Repeated stress alters dendritic spinemorphology in the rat medial prefrontal cortex. J. Comp. Neurol.507, 11411150
49 Yuen, E.Y. et al. (2012) Repeated stress causes cognitive impairmentby suppressing glutamate receptor expression and function inprefrontal cortex. Neuron 73, 962977
50 Aghajanian, G.K. and Marek, G.J. (1997) Serotonin induces excitatorypostsynaptic potentials in apical dendrites of neocortical pyramidalcells. Neuropharmacology 36, 589599
51 Lambe, E.K. et al. (2000) Serotonin induces EPSCs preferentially inlayer V pyramidal neurons of the frontal cortex in the rat. Cereb.
Cortex 10, 97498052 Beque, J-C. et al. (2007) Mechanism of the 5-hydroxytryptamine 2Areceptor-mediated facilitation of synaptic activity in prefrontal cortex.Proc. Natl. Acad. Sci. U.S.A. 104, 98709875
53 Liu, R-J. and Aghajanian, G.K. (2008) Stress blunts serotonin- andhypocretin-evoked EPSCs in prefrontal cortex: role of corticosterone-mediated apical dendritic atrophy. Proc. Natl. Acad. Sci. U.S.A. 105,359364
54 Matsumoto, M. and Hikosaka, O. (2007) Lateral habenula as asource of negative reward signals in dopamine neurons. Nature447, 11111115
55 Proulx, C.D. et al. (2014) Reward processing by the lateral habenula innormal and depressive behaviors. Nat. Neurosci. 17, 11461152
56 Goncalves, L. et al. (2012) Differential projections from the lateralhabenula to the rostromedial tegmental nucleus and ventraltegmental area in the rat. J. Comp. Neurol. 520, 12781300
57 Sego, C. et al. (2014) Lateral habenula and the rostromedialtegmental nucleus innervate neurochemically distinct subdivisionsof the dorsal raphe nucleus in the rat. J. Comp. Neurol. 522,14541484
58 Christoph, G.R. et al. (1986) Stimulation of the lateral habenulainhibits dopamine-containing neurons in the substantia nigra andventral tegmental area of the rat. J. Neurosci. 6, 613619
59 Matsumoto, M. and Hikosaka, O. (2009) Representation of negativemotivational value in the primate lateral habenula. Nat. Neurosci. 12,7784
60 Li, B. et al. (2011) Synaptic potentiation onto habenula neurons in thelearned helplessness model of depression. Nature 470, 535539
61 Morris, J.S. et al. (1999) Covariation of activity in habenula and dorsalraphe nuclei following tryptophan depletion. Neuroimage 10, 163172
62 Duvarci, S. and Pare, D. (2014) Amygdala microcircuits controllinglearned fear. Neuron 82, 966980
63 Drevets, W.C. (2003) Neuroimaging abnormalities in the amygdala inmood disorders. Ann. N. Y. Acad. Sci. 985, 420444
64 Vyas, A. et al. (2002) Chronic stress induces contrasting patterns ofdendritic remodeling in hippocampal and amygdaloid neurons. J.Neurosci. 22, 68106818
65 Felix-Ortiz, A.C. and Tye, K.M. (2014) Amygdala inputs to the ventralhippocampus bidirectionally modulate social behavior. J. Neurosci.34, 586595
66 Magarinos, A.M. and McEwen, B.S. (1995) Stress-induced atrophy ofapical dendrites of hippocampal CA3c neurons: involvement ofglucocorticoid secretion and excitatory amino acid receptors.Neuroscience 69, 8998
67 Pawlak, R. et al. (2005) Tissue plasminogen activator andplasminogen mediate stress-induced decline of neuronal andcognitive functions in the mouse hippocampus. Proc. Natl. Acad.Sci. U.S.A. 102, 1820118206
68 Bartanusz, V. et al. (1995) Stress-induced changes in messenger RNAlevels of N-methyl-D-aspartate and AMPA receptor subunits inselected regions of the rat hippocampus and hypothalamus.Neuroscience 66, 247252
69 Alfarez, D.N. et al. (2003) Chronic unpredictable stress impairs long-term potentiation in rat hippocampal CA1 area and dentate gyrus invitro. Eur. J. Neurosci. 17, 19281934
70 Schmidt, M.V. et al. (2010) Individual stress vulnerability is predictedby short-term memory and AMPA receptor subunit ratio in thehippocampus. J. Neurosci. 30, 1694916958
71 Kole, M.H.P. et al. (2002) The antidepressant tianeptine persistentlymodulates glutamate receptor currents of the hippocampal CA3commissural associational synapse in chronically stressed rats.Eur. J. Neurosci. 16, 807816
72 Kallarackal, A.J. et al. (2013) Chronic stress induces a selectivedecrease in AMPA receptor-mediated synaptic excitation athippocampal temporoammonic-CA1 synapses. J. Neurosci. 33,1566915674
73 Remondes, M. and Schuman, E.M. (2004) Role for a cortical input tohippocampal area CA1 in the consolidation of a long-term memory.Nature 431, 699703
74 Freed, W.J. et al. (1993) Properties of [3H]AMPA binding inpostmortem human brain from psychotic subjects and controls:increases in caudate nucleus associated with suicide. Exp. Neurol.
121, 4856
291
Feature Review Trends in Neurosciences May 2015, Vol. 38, No. 575 MeadorWoodruff, J.H. et al. (2001) Striatal ionotropic glutamatereceptor expression in schizophrenia, bipolar disorder, and majordepressive disorder. Brain Res. Bull. 55, 631640
76 Cai, X. et al. (2013) Local potentiation of excitatory synapses byserotonin and its alteration in rodent models of depression. Nat.Neurosci. 16, 464472
77 Chourbaji, S. et al. (2008) AMPA receptor subunit 1 (GluR-A)knockout mice model the glutamate hypothesis of depression.FASEB J. 22, 31293134
78 Auer, D.P. et al. (2000) Reduced glutamate in the anterior cingulatecortex in depression: an in vivo proton magnetic resonancespectroscopy study. Biol. Psychiatry 47, 305313
79 Pfleiderer, B. et al. (2003) Effective electroconvulsive therapy reversesglutamate/glutamine deficit in the left anterior cingulum of unipolardepressed patients. Psychiatr. Res. 122, 185192
80 Mitani, H. et al. (2006) Correlation between plasma levels ofglutamate, alanine and serine with severity of depression. Prog.Neuropsychopharmacol. Biol. Psychiatry 30, 11551158
81 Choudary, P.V. et al. (2005) Altered cortical glutamatergic andGABAergic signal transmission with glial involvement indepression. Proc. Natl. Acad. Sci. U.S.A. 102, 1565315658
82 Beneyto, M. et al. (2007) Abnormal glutamate receptor expression inthe medial temporal lobe in schizophrenia and mood disorders.Neuropsychopharmacology 32, 18881902
83 Nestler, E.J. and Carlezon, W.A. (2006) The mesolimbic dopaminereward circuit in depression. Biol. Psychiatry 59, 11511159
84 Akiskal, H.S. and McKinney, W.T. (1973) Depressive disorders:toward a unified hypothesis. Science 182, 2029
85 Leuner, B. and Shors, T.J. (2013) Stress, anxiety, and dendriticspines: what are the connections? Neuroscience 251, 108119
86 Spijker, J. et al. (2010) Predictors of suicidality in depressive spectrumdisorders in the general population: results of the Netherlands MentalHealth Survey and Incidence Study. Soc. Psychiatry Psychiatr.Epidemiol. 45, 513521
87 Winer, E.S. et al. (2014) Anhedonia predicts suicidal ideation in alarge psychiatric inpatient sample. Psychiatry Res. 218, 124128
88 Fawcett, J. et al. (1990) Time-related predictors of suicide in majoraffective disorder. Am. J. Psychiatry 147, 11891194
89 Floresco, S.B. et al. (2001) Glutamatergic afferents from thehippocampus to the nucleus accumbens regulate activity of ventraltegmental area dopamine neurons. J. Neurosci. 21, 49154922
90 Salamone, J.D. and Correa, M. (2012) The mysterious motivationalfunctions of mesolimbic dopamine. Neuron 76, 470485
91 ODonnell, P. and Grace, A.A. (1995) Synaptic interactions amongexcitatory afferents to nucleus accumbens neurons: hippocampalgating of prefrontal cortical input. J. Neurosci. 15, 36223639
92 Legault, M. et al. (2000) Chemical stimulation of the ventralhippocampus elevates nucleus accumbens dopamine by activatingdopaminergic neurons of the ventral tegmental area. J. Neurosci.20, 16351642
93 Tyler, W.J. and Pozzo-Miller, L. (2003) Miniature synaptictransmission and BDNF modulate dendritic spine growth and formin rat CA1 neurones. J. Physiol. 553, 497509
94 Sairanen, M. et al. (2005) Brain-derived neurotrophic factor andantidepressant drugs have different but coordinated effects onneuronal turnover, proliferation, and survival in the adult dentategyrus. J. Neurosci. 25, 10891094
95 Weissbourd, B. et al. (2014) Presynaptic partners of dorsal rapheserotonergic and GABAergic neurons. Neuron 83, 645662
96 Malberg, J.E. et al. (2000) Chronic antidepressant treatmentincreases neurogenesis in adult rat hippocampus. J. Neurosci. 20,91049110
97 Gould, E. et al. (1998) Proliferation of granule cell precursors in thedentate gyrus of adult monkeys is diminished by stress. Proc. Natl.Acad. Sci. U.S.A. 95, 31683171
98 Sahay, A. and Hen, R. (2007) Adult hippocampal neurogenesis indepression. Nat. Neurosci. 10, 11101115
99 Kobayashi, K. et al. (2008) Chronic fluoxetine bidirectionallymodulates potentiating effects of serotonin on the hippocampalmossy fiber synaptic transmission. J. Neurosci. 28, 62726280
100 Berman, R.M. et al. (2000) Antidepressant effects of ketamine in
depressed patients. Biol. Psychiatry 47, 351354
292101 Zarate, C.A. et al. (2006) A randomized trial of an N-methyl-D-aspartate antagonist in treatment-resistant major depression.Arch. Gen. Psychiatry 63, 856864
102 Garcia, L.S.B. et al. (2008) Acute administration of ketamine inducesantidepressant-like effects in the forced swimming test and increasesBDNF levels in the rat hippocampus. Prog. Neuropsychopharmacol.Biol. Psychiatry 32, 140144
103 Li, N. et al. (2010) mTOR-dependent synapse formation underliesthe rapid antidepressant effects of NMDA antagonists. Science 329,959964
104 Li, N. et al. (2011) Glutamate N-methyl-D-aspartate receptorantagonists rapidly reverse behavioral and synaptic deficits causedby chronic stress exposure. Biol. Psychiatry 69, 754761
105 Moghaddam, B. et al. (1997) Activation of glutamatergicneurotransmission by ketamine: a novel step in the pathway fromNMDA receptor blockade to dopaminergic and cognitive disruptionsassociated with the prefrontal cortex. J. Neurosci. 17, 29212927
106 Homayoun, H. and Moghaddam, B. (2007) NMDA receptorhypofunction produces opposite effects on prefrontal cortexinterneurons and pyramidal neurons. J. Neurosci. 27, 1149611500
107 Dwyer, J.M. and Duman, R.S. (2013) Activation of mammalian targetof rapamycin and synaptogenesis: role in the actions of rapid-actingantidepressants. Biol. Psychiatry 73, 11891198
108 Carlen, M. et al. (2012) A critical role for NMDA receptors inparvalbumin interneurons for gamma rhythm induction andbehavior. Mol. Psychiatry 17, 537548
109 Kittelberger, K. et al. (2012) Comparison of the effects of acute andchronic administration of ketamine on hippocampal oscillations:relevance for the NMDA receptor hypofunction model ofschizophrenia. Brain Struct. Funct. 217, 395409
110 Maksimow, A. et al. (2006) Increase in high frequency EEG activityexplains the poor performance of EEG spectral entropy monitorduring S-ketamine anesthesia. Clin. Neurophysiol. 117, 16601668
111 Cornwell, B.R. et al. (2012) Synaptic potentiation is critical for rapidantidepressant response to ketamine in treatment-resistant majordepression. Biol. Psychiatry 72, 555561
112 Lorrain, D.S. et al. (2003) Effects of ketamine and N-methyl-D-aspartate on glutamate and dopamine release in the rat prefrontalcortex: modulation by a group II selective metabotropic glutamatereceptor agonist LY379268. Neuroscience 117, 697706
113 Razoux, F. et al. (2007) Ketamine, at a dose that disrupts motorbehavior and latent inhibition, enhances prefrontal cortex synapticefficacy and glutamate release in the nucleus accumbens.Neuropsychopharmacology 32, 719727
114 Aan Het Rot, M. et al. (2012) Ketamine for depression: where do we gofrom here? Biol. Psychiatry 72, 537547
115 Pozzi, L. et al. (2014) Mice lacking NMDA receptors in parvalbuminneurons display normal depression-related behavior and response toantidepressant action of NMD
Recommended
