Cholinergic interneurons in the nucleus accumbensregulate depression-like behaviorJennifer L. Warner-Schmidta,1, Eric F. Schmidtb, John J. Marshalla, Amanda J. Rubina, Margarita Arango-Lievanoc,Michael G. Kaplittc, Ines Ibañez-Tallonb,d, Nathaniel Heintzb, and Paul Greengarda,1

Laboratories of aMolecular and Cellular Neuroscience and bMolecular Biology and Howard Hughes Medical Institute, The Rockefeller University, New York,NY 10065; cLaboratory of Molecular Neurosurgery, Weill Cornell Medical College, New York, NY 10065; and dMolecular Neurobiology Group, Max-Delbrück-Centrum, 13125 Berlin, Germany

Contributed by Paul Greengard, June 5, 2012 (sent for review May 17, 2012)

A large number of studies have demonstrated that the nucleusaccumbens (NAC) is a critical site in the neuronal circuits controllingreward responses,motivation, andmood,but theneuronal cell type(s)underlying these processes are not yet known. Identification of theneuronal cell types that regulate depression-like states will guide usin understanding the biological basis of mood and its regulation bydiseases like major depressive disorder. Taking advantage of recentfindingsdemonstrating that theserotonin receptor chaperone, p11, isan important molecular regulator of depression-like states, here weidentify cholinergic interneurons (CINs) as a primary site of action forp11 in the NAC. Depression-like behavior is observed in mice afterdecrease of p11 levels inNACCINs. This phenotype is recapitulated bysilencing neuronal transmission in these cells, demonstrating thataccumbal cholinergic neuronal activity regulates depression-likebehaviors and suggesting that accumbal CIN activity is crucial forthe regulation of mood and motivation.

acetylcholine | neurotransmission | s100a10 | antidepressant

The nucleus accumbens (NAC) has been identified as a criticalsite in the neuronal circuits controlling reward responses,

motivation, and mood (1–3). For example, in major depressivedisorder (MDD): (i) NAC activity is reduced in response topositive stimuli (4) and (ii) MDD symptoms can be reversed bydeep brain stimulation in the NAC (5, 6). Although these regionalstudies implicate the NAC, the cell type(s) responsible for thepathophysiology remain to be elucidated. In the case of MDD, itis crucial that we determine which neuronal cell types regulatedepressive-like behaviors to eventually develop highly targetedantidepressants that could be both faster acting and have feweroff-site effects. MDD is a debilitating psychiatric conditioncharacterized by anhedonia (defined as a loss of interest inpleasurable things), loss of motivation, negative affect, behavioraldespair, and changes in cognition and basic drives such as eatingand sleeping. Despite its complexity, central features of the dis-ease, such as anhedonia and behavioral despair, can be modeledin rodents to probe the underlying neuronal circuitry.The serotonin receptor (5HTR) interacting protein p11 is

a small intracellular protein that regulates the localization ofcertain 5HTRs, including 5HTR1B and 5HTR4 (7, 8), at thecellular surface. Constitutive p11 knockout (KO) mice show bothanhedonia (e.g., reduced sucrose preference) and increased be-havioral despair (e.g., increased immobility) (7–9). Recently, weidentified the NAC as a key brain region in which loss of p11induces depression-like behaviors (9). The NAC is composed ofseveral different cell types (10). Approximately 95% of theneurons in this region are medium spiny neurons (MSNs), theprojection neurons of the NAC. The cholinergic interneuronshave a characteristically large soma and make up less than 1% ofthe neurons in the striatum (10). Cholinergic interneurons pri-marily target MSNs via muscarinic and nicotinic acetylcholinereceptors, although GABAergic interneurons also receive somecholinergic input.In the current study, we identify cholinergic interneurons as

a primary site of action for p11 in the NAC and show that ma-nipulation of p11 in this single cell type regulates anhedonia and

behavioral despair in rodent models of depression. The de-pression-like phenotype that we observe in mice lacking p11 inaccumbal cholinergic interneurons is recapitulated by silencingneuronal transmission in these cells. This work identifies accumbalcholinergic interneurons as critical modulators of depression-likestates and suggests that the activity of these neurons is crucial forthe regulation of mood and motivation.

ResultsAnalysis of Accumbal Cell Types Expressing p11. We reported thatreduced p11 expression in the NAC leads to a depressive-likephenotype (9). Therefore, we sought to determine the cell typesin the NAC that express p11. Immunohistochemical analysis ofhuman NAC tissue indicated that p11 was highly expressed incholinergic interneurons, which are identified by their expressionof choline acetyltransferase (ChAT) (Fig. 1A). Similarly, in themouse, ChAT neurons showed the highest levels of p11 in theNAC (Fig. 1B) compared with other cell types. We next usedthe translating ribosome affinity purification (TRAP) technique(11, 12) to isolate ribosome bound mRNA from multiple celltypes in the NAC. We used bacTRAP mice expressing the EGFP-L10a transgene in cholinergic interneurons, dopamine receptor1 (D1) expressing striatonigral MSNs, and D2-expressing stria-topallidal MSNs. Subsequent quantification of p11 message levelsby RT-quantitative PCR (qPCR) indicated that p11 expression inaccumbal cholinergic interneurons was enriched 30-fold com-pared with noncholinergic cell types in the NAC [±9.09 SEM,F(1,4)=30.02, P < 0.01]. Both D1- and D2-receptor containingMSNs express p11 but at lower levels compared with the cho-linergic interneurons (Fig. 1 C and D) [D1-MSNs vs. rest of theNAC: 1.3-fold ± 0.04 SEM, t(1,6) =3.611, P < 0.05; D2-MSNs vs.rest of the NAC: 1.25-fold ± 0.02 SEM, t(1,6) =2.979, P < 0.05].

Effect of Cell Type-Specific Deletion of p11 on Depression-LikeBehavior. Given the high levels of p11 expressed by cholinergicinterneurons, we next examined whether these cells regulatedepressive-like behavior, including anhedonia and behavioraldespair. For this purpose, we used an intersectional strategy byusing p11 conditional floxed mice (13) and four different BACtransgenic mice expressing CRE recombinase under differentBAC promoters to target the different cell populations (14).First, ChAT-CRE–expressing mice were used to target cholin-ergic neurons, including but not limited to the NAC cholinergicinterneurons. Second, dopamine D2-receptor (D2-) CRE micetargeted striatopallidal MSNs and ChAT neurons in the striatumbut also expressed CRE in D2-containing neurons in the cortex,

Author contributions: J.L.W.-S., E.F.S., I.I.-T., N.H., and P.G. designed research; J.L.W.-S.,E.F.S., J.J.M., and A.J.R. performed research; M.A.-L., M.G.K., and I.I.-T. contributed newreagents/analytic tools; J.L.W.-S., E.F.S., and J.J.M. analyzed data; and J.L.W.-S., E.F.S.,I.I.-T., N.H., and P.G. wrote the paper.

The authors declare no conflict of interest.1To whom correspondence may be addressed. E-mail: jschmidt@rockefeller.edu orgreengard@rockefeller.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1209293109/-/DCSupplemental.

11360–11365 | PNAS | July 10, 2012 | vol. 109 | no. 28 www.pnas.org/cgi/doi/10.1073/pnas.1209293109

Dow

nloa

ded

by g

uest

on

Janu

ary

26, 2

021

hypothalamus, and midbrain. However, the only point of in-tersection between the ChAT-CRE and D2-CRE mouse lines isstriatal cholinergic interneurons. We took advantage of this in-tersection and compared behavioral results from these two lines.In contrast to the D2-CRE mice, adenosine 2a (A2a)-CRE isa striatopallidal MSN-specific line, targeting only those cells.Finally, dopamine D1-receptor (D1-) CRE expressing mice tar-geted striatonigral MSNs. Loss of p11 in the relevant cell typewas confirmed by immunohistochemistry.These cell type-specific p11 KO mice were tested in de-

pression-related behavior tests. Anhedonia is modeled in rodentsby measuring preference for palatable sucrose solution. In thecurrent study, we assessed anhedonia and also performance intwo widely used models of behavioral despair, the tail suspensiontest (TST) and the forced swim test (FST). Both the TST and theFST are models of depression that assess the response of ananimal to an inescapable stressor (15). Behavioral analysesrevealed that ChAT- and D2-p11KO mice had a reduced pref-erence for sucrose (Fig. 2A), and increased immobility in theTST (Fig. 2B), consistent with the anhedonic and depressive-likephenotype observed in the constitutive p11KO mice (7). ChAT-p11KO mice also showed increased immobility in the FST (Fig.

S1) compared with controls. In contrast, A2a-p11KO mice be-haved similarly to controls in sucrose preference and TST (Fig. 2A and B), indicating that the effects observed in the D2-p11KOmice can be attributed to the loss of p11 in the cholinergicinterneurons, which express D2 receptors, and not because ofp11 loss in the D2-expressing MSNs. D1-p11KO mice showed nodifference in sucrose preference (Fig. 2A) and reduced immo-bility in the TST (Fig. 2B) compared with controls. All fourtransgenic mouse lines tested responded normally to the selec-tive serotonin reuptake inhibitor (SSRI) antidepressant cit-alopram in the TST (Fig. 2B) and showed no differences inlocomotor activity (Fig. S2).

Silencing Neurotransmission in Cholinergic Interneurons InducesDepression-Like Behaviors. To definitively identify accumbal ChATneurons as regulators of depression-like behaviors and to de-termine how the activity of these neurons is involved in the be-havioral outputs, we silenced neurotransmission in these cells byusing virus-mediated delivery of two membrane-tethered toxins (t-toxins) against calcium voltage-gated channels. This approach hasbeen used in other cell types and has been shown to effectivelysilence neurotransmission in cells expressing these t-toxins (16).Mice expressing CRE recombinase under the ChAT bac promoterwere infected with the CRE-dependent AAV-t-toxins MPE andAPC or with control AAV-PE virus. Expression of the t-toxins orcontrol virus was determined by immunohistochemistry (Fig. 3A),and the number of ChAT-positive cells expressing the viruses wasdetermined (Fig. 3 B and C). In mice infected with the controlvirus (AAV-PE), 77% of ChAT neurons expressed AAV-PE, asdetected by EGFP/ChAT colocalization (Fig. 3B). In mice infec-ted with the AAV-MPE/APC viruses, 78% of ChAT neuronsexpressed both AAV-MPE and APC, as detected by colocalizationof EGFP, Cherry, and ChAT, and indicating the near-completeinfection of NAC ChAT neurons by the viruses (Fig. 3C). Allneurons expressing AAV-MPE/APC were ChAT positive, dem-onstrating that there was no ectopic expression of the t-toxinviruses in non-ChAT neurons. Mice infected with the t-toxinsshowed significantly reduced sucrose preference (Fig. 3D) andincreased immobility in both the TST (Fig. 3E) and the FST (Fig.3F) compared with controls. There was no effect of the t-toxins ontotal distance traveled in an open field (Fig. 3G), illustrating thatthe observed effects on immobility could not be attributed tochanges in locomotor activity. Taken together, these data suggestthat silencing neurotransmission in accumbal cholinergic inter-neurons mimics the effect of p11KO and induces a depression-likebehavioral phenotype, suggesting that downstream neuronal ac-tivity of these neurons is necessary to maintain motivation andprevent behavioral despair.

Cell Type-Specific Overexpression of p11 Rescues KO Phenotype.Given our previous studies implicating the NAC as the site wherep11 loss exerts its effects on depression-related behaviors (9),and our current data showing that ChAT-p11KO and D2-p11KOmice are anhedonic and have increased behavioral despair, wetook a second approach that permitted temporal, spatial, andcell type-specific p11 expression. For this purpose, we generateda virus that selectively overexpressed p11 only in neuronsexpressing CRE recombinase (Fig. 4 A and B). Thus, miceexpressing CRE recombinase in ChAT neurons were bred withconstitutive p11KO mice, and the NAC was infected with theCRE-dependent AAV. The NAC can be subdivided into coreand shell regions, which share distinct anatomical and functionalcircuitry (17). Because cholinergic interneurons are an extremelysparse population of neurons (<1%) in the mouse NAC, we havenot attempted to distinguish between core and shell regions inthe current study. Behavioral analysis of these mice showed thatp11 overexpression specifically in NAC ChAT neurons restorednormal behavior in constitutive p11KO mice, reflected both inthe sucrose preference test and the TST (Fig. 4 C and D). Therewas no effect of p11 overexpression on locomotor activity(Fig. 4E). In wild-type (WT) mice, p11 overexpression in NAC

mergeHUMAN ChATHUMAN p11

MOUSE ChAT mergeMOUSE p11

MOUSE ChAT mergeMOUSE p11

A

B

mergeHUMAN ChATHUMAN p11

merge

merge

met-enk (D2)

drd1a (D1)

C

D

MOUSE p11

MOUSE p11

Fig. 1. Accumbal cholinergic interneurons express relatively high levels ofp11. Colocalization of p11 (red) with choline acetyltransferase (green) in thehuman nucleus accumbens (A) or the nucleus accumbens of wild-type mice(B). The ChAT positive neuron from the 40× image (white box) is magnifiedbelow 4×. (C and D) Colocalization of p11 (red) with medium spiny neuroncell-type markers (green) in the nucleus accumbens of wild-type mice: met-enkephalin, a maker of D2-receptor–containing striatopallidal medium spinyneurons (C), and drd1a, a marker of D1-receptor–containing striatonigralmedium spiny neurons (D). Filled arrows indicate a cell that coexpresses p11and the cell-type marker (ChAT, met-enkephalin, or drd1a); open arrowsindicate cells expressing p11 but not the marker. (Scale bars: 20 μm.)

Warner-Schmidt et al. PNAS | July 10, 2012 | vol. 109 | no. 28 | 11361

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Janu

ary

26, 2

021

ChAT neurons had no effect on behavior (Fig. 4 C–E), which isconsistent with previous results (9). Control virus injections(AAV-YFP) also did not influence behavioral responses in WTor p11KO mice (Fig. 4 C–E). Overexpression of p11 in ChATneurons of the dorsal striatum (caudate/putamen; CPU) had noeffect on the behavioral responses of p11KO mice tested in thesucrose preference test, TST, and open field (Fig. 4 F–H). Theseresults demonstrate that the influence of p11 in ChAT neuronson depression-like behaviors is specific to cells located in theNAC and not generalized to ChAT neurons in the CPU.

DiscussionStriatal cholinergic interneurons have been linked to addictivebehaviors (18, 19), reward learning (19, 20), cognitive function(21), and feeding behaviors (22), all of which can be dysregulatedin patients suffering from depression. The NAC, specifically, hasbeen strongly linked to depressive symptoms (1, 4–6) and isa brain region where p11 levels modulate depressive-likebehaviors (9). Using an intersectional loss-of-function approach,we demonstrate that deletion of p11 specifically in NAC

cholinergic interneurons led to depression-like behaviors. Thiseffect was reversed by subsequent overexpression of p11 in NACChAT neurons of constitutive p11KO mice, using a CRE-de-pendent AAV. Further, we demonstrate that ChAT interneuronsplay a key role in the NAC circuit regulating mood, because si-lencing neurotransmitter release from these cells using mem-brane-tethered toxins resulted in a depression-like phenotype.Taken together, these results show that NAC cholinergic inter-neurons are crucial for the generation of depression-relatedbehavior, identifying this cell type as an important substratein the circuitry that regulates mood, motivation, and rewardresponses.

NAC and Its Involvement in Mood Circuitry. The NAC is a compo-nent of a dispersed circuit that modulates mood and motivation.Human imaging studies and rodent models have demonstratedthat its activity is linked to depressive-like states (1, 5, 23, 24). TheNAC receives dopaminergic inputs from the ventral tegmentalarea (VTA), serotoninergic inputs from the raphe, and gluta-matergic inputs from regions such as the amygdala, hippocampus,

WTChAT-p11KO

A

Suc

rose

Pre

fere

nce

(% W

T)

Tail

susp

ensi

on te

st im

mob

ility

(% W

T/ve

hicl

e)

*

#***

D2-p11KOA2a-p11KOD1-p11KO

0

20

40

60

80

100

120

140

160WTChAT-p11KOD2-p11KOA2a-p11KOD1-p11KO

##

#

#

*

* *

vehicle citalopram

B

60

70

80

90

100

110

120Fig. 2. Cholinergic neuronsmediate the effects of p11 ondepression-like behaviors. (A)ChAT-, D2-, A2a-, and D1- p11KO mice were tested for su-crose preference, a model forthe hallmark depressive symp-tom of anhedonia [F(4,107) =2.578, P < 0.05]. (B) Baselineimmobility and the response tocitalopram, an SSRI antidepres-sant, were measured in the tailsuspension test in ChAT-, D2-,A2a-, or D1-p11 KO mice [ge-notype: F(4,111) = 8.604, P < 0.01; citalopram: F(1,111) = 64.5, P < 0.01; interaction: F(4,111) = 0.3805, n.s.]. All data are presented as means ± SEM. Statisticallysignificant effects of genotype (*P < 0.05) or citalopram (#P < 0.05) are noted. n.s., not significant.

75

80

85

90

95

100

105

110

Suc

rose

pre

fere

nce

(% a

av-G

FP)

*

B

D E

C

0

20

40

60

80

100

120

140

160

180**

Tail

Sus

pens

ion

Test

Imm

obili

ty (%

aav

-GFP

)

Forc

ed S

wim

Tes

tIm

mob

ility

(% a

av-G

FP)

0

500

1000

1500

2000

2500

3000

Tota

l dis

tanc

e (c

m)

A

0

20

40

60

80

100

120

140

160

*

AAV-PE AAV-MPE/APC AAV-PE AAV-MPE/APC AAV-PE AAV-MPE/APC AAV-PE AAV-MPE/APC

0

10

20

30

40

50

60

70

80

90

100

0

10

20

30

40

50

60

70

80

90

100

AAV-MPE/APCAAV-PE

% o

f ChA

T po

sitiv

e ce

lls

% o

f ChA

T po

sitiv

e ce

lls

EGFP

Cherry

ChAT

MERGE

F G

EGFP/Cherry/ChAT

Cherry/ChAT

EGFP/ChAT

ChAT only

EGFP/ChAT

ChAT only

Fig. 3. Silencing accumbal cholinergic interneurons causes depression-like behaviors. (A) Immunohistochemical detection of EGFP (green), Cherry (red), andChAT (blue) in the NAC of a ChAT-CRE positive mouse infected with the AAV-MPE/APC t-toxins mixture. Quantification of the number of ChAT neuronsexpressing AAV-PE control virus (B) or AAV-MPE/APC viruses (C) indicated near complete infection of ChAT neurons in the NAC by the viruses and no ectopicvirus expression in non-ChAT neurons (Results). Mice infected with the AAV-t-toxins (AAV-MPE/APC) or with the AAV-control virus (AAV-PE) were tested inthe sucrose preference test [t(16) = 2.259, P < 0.05] (D), the TST [t(18) = 4.095, P < 0.01] (E), the FST [t(18) = 2.565, P < 0.05] (F), or the open field test (G). All dataare presented as means ± SEM. *P < 0.05, **P < 0.01.

11362 | www.pnas.org/cgi/doi/10.1073/pnas.1209293109 Warner-Schmidt et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

26, 2

021

thalamus, and infralimbic prefrontal cortex (PFC), all of whichhave been linked to depression and/or antidepressant action(Fig. 5A). Although MSNs are the primary projection neurons ofthe NAC, local cholinergic interneurons function to regulate theiractivity (10) via AChRs expressed on the MSN cell surface. Whilecholinergic interneurons account for only a small percentage ofcells in the NAC (∼1%), they exert a robust influence overaccumbal function. Our results emphasize this point in that de-letion of p11 specifically from these cells was sufficient to inducea depression-like phenotype, likely a result of decreased activityof the ChAT cells, because silencing the cells with membranetethered-toxins also resulted in depressive-like behavioralresponses. Such an effect may have altered behavior as a result ofaltered MSN activity, because it has been shown that photo-inhibition of NAC cholinergic interneurons enhances MSNspiking in vivo (19).It is notable that some depression-related behaviors (e.g., su-

crose preference) may be explained by disruptions in rewardlearning, a feature of MDD in humans (25). Silencing neuro-transmission in accumbal ChAT neurons reduces sucrose pref-erence (Fig. 3A) and cocaine-induced reward learning (19).We identify NAC cholinergic interneurons as a cell type that is

responsible for inducing both anhedonia and behavioral despairin rodent models of depression. It has been suggested that ace-tylcholine (ACh) plays an important role in depression and itstreatment (26, 27), and cholinergic interneurons are the solesource of acetylcholine in the NAC. However, a cholinergic hy-pothesis of depression suggests that increasing cholinergic trans-mission is prodepressive and that anticholinergics may function asantidepressants, supported in part by the observation that tricyclicantidepressants have central anticholinergic properties (26). This

hypothesis is further supported by the more recent finding thatscopolamine, a centrally acting antimuscarinic agent, is a potentantidepressant in humans (28). When considering the cellularcircuitry that may underlie these effects, it is important to notethat (i) there are several sources of ACh in the brain, e.g., basalforebrain and pontomesencephalic cholinergic neurons, whichexert widespread and complex cholinergic modulation of manybrain areas including the PFC, hippocampus, amygdala, VTA,and dorsal raphe (Fig. 5A) and (ii) the cellular mechanisms un-derlying the actions of anticholinergics (e.g., scopolamine) are notyet understood.Within the accumbens, cholinergic interneurons are the only

source of ACh, acting locally to regulate the activity of otherinterneurons and NAC efferents. Silencing neurotransmission inaccumbal ChAT neurons would prevent release of acetylcholinethat modulates downstream cellular responses to glutamate,dopamine, and other transmitters within the NAC, ultimatelyaffecting the output of the NAC through the activity of D1- orD2-containing MSNs (Fig. 5B). However, there is new evidencethat striatal cholinergic interneurons also modulate fast gluta-matergic neurotransmission (29), which would also be silencedby the methods used in the current study. Therefore, althoughthe cholinergic hypothesis outlined above suggests that anti-cholinergics are antidepressants, we find that silencing cholin-ergic interneurons locally in the NAC is prodepressive in rodentmodels of the disease. One simple explanation is that the effectsof scopolamine and other anticholinergics are due to ACh ac-tivity in brain areas outside the NAC. Within the NAC, and giventhe complexity of the local microcircuitry and the diverse inputsand outputs, our identification of a single cell type that regulatesanhedonia and behavioral despair is an important step forward

A

C

ac

ac

RFP ChAT p11 merge

ChA

T-C

re p

ositi

vep1

1 K

O m

ouse

ChA

T-C

re n

egat

ive

p11

KO

mou

se

B

RFP ChAT p11 merge

loxP loxP

RFP/STOP p11X

loxP loxP

RFP/STOP p11X

+ Cre

- Cre

Tail

Sus

pens

ion

Test

Suc

rose

Pre

fere

nce

(%W

T/aa

v-Y

FP)

#

**

Tota

l Dis

tanc

e (c

m)

0

200

400

600

800

1000

1200

1400

1600

1800

WT p11 KO

E

70

75

80

85

90

95

100

105

110

115

WT p11 KO

D

Imm

obili

ty (%

WT/

aav-

YFP

)

#*

95

100

105

110

115

120

125

WT p11KO

0

20

40

60

80

100

120

140

160

180

0

200

400

600

800

1000

1200

1400

1600

1800

2000F G H

0

20

40

60

80

100

120

140

Suc

rose

Pre

fere

nce

(%K

O/a

av-Y

FP)

aav-YFP (NAC)aav-p11 (NAC)

aav-YFP (NAC)aav-p11 (NAC)

aav-YFP (NAC)aav-p11 (NAC)

aav-YFP (CPU)aav-p11 (CPU)

aav-YFP (CPU)aav-p11 (CPU)

aav-YFP (CPU)aav-p11 (CPU)

p11KO p11KO p11KO

Tail

Sus

pens

ion

Test

Imm

obili

ty (%

KO

/aav

-YFP

)

Tota

l Dis

tanc

e (c

m)

Fig. 4. Overexpression of p11 in NAc cholinergicinterneurons of constitutive p11KO mice restoresnormal behavior. (A) Schematic of the virus used inthese experiments (see text). Constitutive p11 KOmice were bred with ChAT-CRE mice to generate p11KOs that expressed CRE in ChAT neurons. (B) Immu-nohistochemical detection of RFP (red), ChAT (blue),and p11 (green) in the NAC of a p11 KO mouseexpressing (Upper) or not expressing (Lower) CRE inChAT neurons. Arrows indicate ChAT-positive cho-linergic interneurons overexpressing (Upper) or notover-expressing (Lower) p11. (C–E) Control virus (aav-YFP) or p11 overexpressing virus (aav-p11) wasinjected to the NAc of CRE-positive WT or p11 KOmice before behavioral testing in the TST [F(1,32) =6.384, P < 0.05] (C), the sucrose preference test(F(1,32) = 5.681, P < 0.05) (D), or the open field test[main effect AAV: F(1,37) = 3.505, n.s.; main effectgenotype: F(1,37) = 0.2654, n.s.] (E). All data are pre-sented as means ± SEM. Significant effects of geno-type (#P < 0.05) or p11 overexpression (*P < 0.05,**P < 0.01) are noted. (F–H) CRE-positive constitutivep11KO mice were injected with aav-p11 or aav-YFPinto the dorsal striatum (CPU). No effect of p11overexpression was observed in sucrose preference(main effect virus: F(1,26) = 0.3801, n.s.; main effectgenotype: F(1,26) = 1.136, n.s.; interaction virus × ge-notype: F(1,26) = 0.1272, n.s.) (F), tail suspension testimmobility (main effect virus: F(1,25) = 1.994, n.s.;main effect genotype F(1,25) = 1.521, n.s.; Interactionvirus × genotype: F(1,25) = 3.874, n.s.) (G), or loco-motor activity (main effect virus: F(1,26) ≤0.001, n.s.;main effect genotype: F(1,26) =1.244, n.s.; interactionvirus × genotype: F(1,26) = 1.022, n.s.) (H). All data arepresented as means ± SEM.

Warner-Schmidt et al. PNAS | July 10, 2012 | vol. 109 | no. 28 | 11363

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Janu

ary

26, 2

021

in mapping the functional cellular circuitry dysregulated in MDDand related diseases.Because silencing neurotransmission in ChAT neurons mimics

the behavioral phenotype observed when p11 is deleted fromthese cells, we hypothesize that the removal of p11 disruptsChAT neuron activity and/or neurotransmission. This effect maybe achieved by changes in the cellular response to serotoninbecause the expression of serotonin receptors at the cell surfaceis regulated by p11 levels (7). It is also likely that other pathwaysmay contribute to the altered physiological responses, becausep11 is known to bind other transmembrane ion channels andsignaling proteins (30). Here, we clearly identify accumbal ChATneurons as a cellular substrate for regulation of depression-likebehavior by p11. Future molecular and pharmacological studies,including TRAP molecular profiling will be necessary to validatethis hypothesis.

Anatomical Dissociation Between Roles of p11 in Mediating Depressive-Like States and Antidepressant Responses. Current antidepressanttherapies have relatively low treatment response rates and areaccompanied by a host of unpleasant side effects. These treat-ments are based on drugs developed more than 50 y ago. It istherefore crucial to incorporate the significant progress that hasbeen made toward our understanding the neural circuitry medi-ating MDD into developing novel treatments that specificallytarget the underlying cell types. Previous work has shown thatconstitutive p11KO mice, which lack p11 in the whole brain andbody, have both a depressive-like phenotype and are less sensitiveto antidepressant treatment (7). p11 is normally expressed ina variety of cell types throughout the central nervous system (31).The current study demonstrates that p11 in NAC cholinergic

interneurons is sufficient for regulating anhedonia and behavioraldespair, identifying these cells as modulators of mood.It is notable that ChAT-p11KO mice retain their ability to

respond to SSRI antidepressant treatment. This result suggestsan anatomical dissociation between p11’s involvement in regu-lating depression-like states and antidepressant responses. Infact, this notion is consistent with recent evidence showing thatcorticostriatal p11-expressing neurons, which have a strong mo-lecular response to an SSRI also mediate behavioral antide-pressant responses (32). Loss of p11 in these cells does notinduce a depression-like phenotype (32). Our results are alsoconsistent with the observation that CAMK2α-p11KO mice,which lack p11 in most forebrain neurons (including the corti-costriatal p11 cells and striatal MSNs, but not striatal ChATneurons), show no response to SSRI antidepressants (13). Thesemice also do not show a depression-like phenotype, presumablybecause p11 levels in the accumbal cholinergic interneuronsare preserved.The complexity of the neural circuitry underlying mood points

toward distinct sites for pathophysiology and treatment. Forexample, our findings implicate NAC cholinergic interneurons inmediating depression-related behaviors, but the p11-expressingcorticostriatal cells underlying antidepressant responses sendprojections predominantly to dorsal striatum with few efferentsto the NAC (32). Although there appear to be no direct con-nections between these two cell types, they are both significantcomponents of parallel basal ganglia loops, which have in-dependent but integrated roles in cognitive and motivationalfeatures of behavior (33–35). These observations present de-finitive evidence that the cell types generating behavioral phe-notypes and those required for effective therapy are distinct.Such an anatomical dissociation between disease and treatmentis common for complex circuit-based diseases. In the case ofParkinson’s disease, the dopaminergic cells in substantia nigradegenerate, but current effective therapies target downstreamdopamine-receptive cells (36). The present study combined withrecent results from the cortex (32) sets the precedent that an in-depth analysis of multiple cell types will be required to un-derstand both the genesis of the depressive behavioral phenotypeand the mechanisms required for effective treatment.

MethodsAnimals. Male C57BL/6 mice (8–10 wk old; Charles River Laboratories) wereused for experiments in Fig. 1. For all other experiments, male transgenicmice and WT littermates (8–24 wk old) were used. Mice were housed 2–5 percage with ad libitum access to food and water. p11 constitutive KO micewere generated and maintained at the Rockefeller University (7). Cell type-specific p11 KO mice were generated at The Rockefeller University by breedingfloxed p11 mice (13) with mice expressing the CRE recombinase under ChAT(GM60), Drd2 (ER44), Adora2a (KG139), or Drd1a (EY262) promoters (CRE miceprovided by GENSAT). For AAV experiments, ChAT-CRE mice were bred withWT or p11 constitutive KO mice. For bacTRAP experiments, ChAT-, D1-, or D2-bacTRAP mice were used (12). Animal use and procedures were in accordancewith the National Institutes of Health guidelines and approved by the In-stitutional Animal Care and Use committees.

Immunohistochemistry. Immunohistochemistry was performed using stan-dard procedures (see SI Methods).

Behavioral Analysis. Sucrose preference was performed as described (7). Theopen field locomotor test, TST, and FST were performed as described (8).Citalopram hydrobromide (Sigma) was dissolved in sterile saline and wasdelivered acutely (20 mg/kg, i.p.) 30 min before the tail suspension test.

AAV Vector Construction and Virus Production. p11 viruses. The CRE inducibleoverexpressing vectors were obtained by subcloning either the mouse p11coding sequence or a flag-tagged YFP downstream of an RFP-STOP cassetteflanked by two loxP sequences (ATAACTTCGTATAGCATACATTATACGAA-GTTAT) into an AAV vector containing a CMV/Chicken β-actin hybrid pro-moter followed by composite chicken β-actin/rabbit β-globin intron (giftfrom Matthew J. During, Ohio State University, Columbus, Ohio). In the ab-sence of the CRE recombinase, RFP is expressed. In the presence of the CRE

A

PFCHIP

THAL

HYPO

VTADR/LC

STR

NAC

AMY

VP

B

NAC output neurons

Basal ForebrainCholinergic neurons

PontomesencephalicCholinergic neurons

ACh ACh

ACh

ACh

AChACh

D2-MSNChAT

SS PV CAL

D1-MSN

Fig. 5. Circuitry that may mediate depressive-like behaviors and the actionsof antidepressants. (A) A simplified summary of the inputs (solid blackarrows) and outputs (dashed black arrows) of the NAC (for more detailedanatomy, see Groenwegen et al., 1999). Red arrows indicate cholinergicinputs to these areas, including basal forebrain and pontomesencephaliccholinergic neurons. In the NAC, cholinergic interneurons are the only sourceof acetylcholine, acting locally within the NAC. (B) In the NAC, there are atleast six different cell types. Drd1a containing medium spiny neurons (D1-MSN) and Drd2 containing medium spiny neurons (D2-MSN) are the primaryoutput neurons of the NAC. Cholinergic interneurons (ChAT) and three typesof GABAergic interneurons expressing somatostatin/nNOS (SS), parvalbumin(PV), or calretinin (CAL), also reside in the NAC, acting locally to control theactivity of the MSNs. AMY, amygdala; DR/LC, dorsal raphe/locus coeruleus;HIP, hippocampus; HYPO, hypothalamus; NAC, nucleus accumbens; PFC,prefrontal cortex; STR, striatum; THAL, thalamus; VP, ventral pallidum; VTA,ventral tegmental area.

11364 | www.pnas.org/cgi/doi/10.1073/pnas.1209293109 Warner-Schmidt et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

26, 2

021

recombinase, the loxP sequences are recombined, RFP-STOP cassette is ex-cised, and p11 or flag YFP is expressed. Virus stocks were prepared bypackaging the vector plasmid into AAV serotype 2 particles using a helper-free plasmid transfection system. The viral vectors were purified by usingheparin affinity chromatography and dialyzed against PBS. AAV titers weredetermined by qPCR with primers to a fragment of the AAV backbone andadjusted to 1011 genomic particles per milliliter.Vector construction and AAV production of the tethered peptide toxin antagonists ofCav2.1 and Cav2.2 calcium channels (MPE/APC t-toxins). T-toxin expression cas-settes (named MPE and APC) contained the sequences encoding the MVIIAconotoxin (from Conus magus) fused to the PDGF-receptor transmembranedomain followed by EGFP (MPE), or the AgaIVA agatoxin (from Agenelopsisaperta), followed by the TM domain and mCherry (APC). The no-toxincontrol contains the transmembrane domain of the PDGF receptor and EGFP(PE) (16). Double floxed AAV constructs were generated by insertion of theinverted t-toxin expression cassettes MPE or APC or the no-toxin control PEbetween double lox 2722 and lox P incompatible sites (DFI). The t-toxindouble floxed cassettes were further subcloned into the pENN-AAV-CB7vector. AAV Vector production of the AAV2/8 serotype was performed byUniversity of Pennsylvania vector core. The titers (genome copies per milli-liter) of the AAVs were as follows: 6.62e12 for AAV2/8.CB7.CI.DFI-MPE, 7e12

for AAV2/8.CB7.CI.DFI-APC, and 8.44e12 for AAV2/8.CB7.CI.DFI-PE. The MPEand APC virus stocks were combined (1:1) immediately before use, and 1 μLof the combined MPE/APC (t-toxin) stock or no-toxin control was injectedbilaterally into the NAC by using standard stereotaxic surgical procedures.Stereotaxic virus injections were performed under ketamine and xylazineanesthesia as described (9).

bacTRAP and RT-qPCR. Experiments were performed as described (12). ChAT-,D1-, or D2-bacTRAP mice, expressing EGFP-tagged L10a ribosomal protein

under the choline acetyltransferase promoter, were backcrossed to C57BL/6mice. Six to nine male mice from each line were pooled for each sample, andexperiments were performed in triplicate. Striatum were dissected and ho-mogenized, cleared by centrifugation, and polysomes were immunopreci-pitated by using antibodies for EGFP. mRNA samples were collected by usingthe Absolutely RNA Nanoprep kit (Stratagene) according to the manu-facturer’s instructions. RNA was quantified, and quality was checked byBioanalyzer. Amplification and cDNA synthesis was performed by using theWT-Ovation RNA Amplification system (Nugen) according to the manu-facturer’s instructions. Samples were run in triplicate and normalized toβ-actin by using TaqMan assays (Applied Biosystems). Assays used were asfollows: S100A10 (Mm00501457_m1) and β-actin (Mm00607939_s1).

Statistical Analysis. Comparisons were made by two-tailed unpaired t test orANOVA using Prism 5 software (GraphPad). In experiments composed ofmore than two groups, data were first analyzed by two-way ANOVA fol-lowed by post hoc Bonferroni test. Statistical significance was set at P ≤ 0.05.

ACKNOWLEDGMENTS. We thank Sebastian Auer, James A. Belarde, andSarah Meller for their technical assistance. This work has been supported byUS Army Medical Research Acquisition Activity Grants W81XWH-09-1-0401(to J.L.W.-S.), W81XWH-09-1-0402 (to P.G.), W81XWH-09-1-0108 (to P.G.),and W81XWH-09-0381 (to M.G.K.); National Institutes of Health (NIH)/National Institute of Mental Health Grant MH090963 (to P.G., N.H., andE.F.S.); Howard Hughes Medical Institute (HHMI) (to N.H. and E.F.S.);American Recovery and Reinvestment Act NIH/National Institute on DrugAbuse Grant 1RC2DA028968 (to P.G., N.H., and E.F.S.); National Institute ofAging Grant AG09464 (to P.G.); The JPB Foundation (to P.G. and M.G.K.); TheFisher Center for Alzheimer’s Research Foundation (to P.G.); The Simons Foun-dation (to N.H. and P.G.); The Helmholtz Association 31-002 (to I.I.-T.); andThe Sonderforschungsbereich SFB 665 (to I.I.-T.). N.H. is an HHMI Investigator.

1. Nestler EJ, Carlezon WA, Jr. (2006) The mesolimbic dopamine reward circuit indepression. Biol Psychiatry 59:1151–1159.

2. Price JL, DrevetsWC (2010) Neurocircuitry of mood disorders. Neuropsychopharmacology35:192–216.

3. Wise RA (2009) Roles for nigrostriatal—not just mesocorticolimbic—dopamine in re-ward and addiction. Trends Neurosci 32:517–524.

4. Heller AS, et al. (2009) Reduced capacity to sustain positive emotion in major de-pression reflects diminished maintenance of fronto-striatal brain activation. Proc NatlAcad Sci USA 106:22445–22450.

5. Bewernick BH, et al. (2010) Nucleus accumbens deep brain stimulation decreasesratings of depression and anxiety in treatment-resistant depression. Biol Psychiatry67:110–116.

6. Schlaepfer TE, et al. (2008) Deep brain stimulation to reward circuitry alleviates an-hedonia in refractory major depression. Neuropsychopharmacology 33:368–377.

7. Svenningsson P, et al. (2006) Alterations in 5-HT1B receptor function by p11 indepression-like states. Science 311:77–80.

8. Warner-Schmidt JL, et al. (2009) Role of p11 in cellular and behavioral effects of 5-HT4receptor stimulation. J Neurosci 29:1937–1946.

9. Alexander B, et al. (2010) Reversal of depressed behaviors in mice by p11 genetherapy in the nucleus accumbens. Sci Transl Med 2:54ra76.

10. Tepper JM, Bolam JP (2004) Functional diversity and specificity of neostriatal inter-neurons. Curr Opin Neurobiol 14:685–692.

11. Doyle JP, et al. (2008) Application of a translational profiling approach for thecomparative analysis of CNS cell types. Cell 135:749–762.

12. Heiman M, et al. (2008) A translational profiling approach for the molecular char-acterization of CNS cell types. Cell 135:738–748.

13. Warner-Schmidt JL, Vanover KE, Chen EY, Marshall JJ, Greengard P (2011) Antide-pressant effects of selective serotonin reuptake inhibitors (SSRIs) are attenuated byantiinflammatory drugs in mice and humans. Proc Natl Acad Sci USA 108:9262–9267.

14. Gong S, et al. (2007) Targeting Cre recombinase to specific neuron populations withbacterial artificial chromosome constructs. J Neurosci 27:9817–9823.

15. Cryan JF, Holmes A (2005) The ascent of mouse: Advances in modelling humandepression and anxiety. Nat Rev Drug Discov 4:775–790.

16. Auer S, et al. (2010) Silencing neurotransmission with membrane-tethered toxins. NatMethods 7:229–236.

17. Meredith GE, Baldo BA, Andrezjewski ME, Kelley AE (2008) The structural basis formapping behavior onto the ventral striatum and its subdivisions. Brain Struct Funct213:17–27.

18. Williams MJ, Adinoff B (2008) The role of acetylcholine in cocaine addiction. Neuro-psychopharmacology 33:1779–1797.

19. Witten IB, et al. (2010) Cholinergic interneurons control local circuit activity andcocaine conditioning. Science 330:1677–1681.

20. Joshua M, Adler A, Mitelman R, Vaadia E, Bergman H (2008) Midbrain dopaminergicneurons and striatal cholinergic interneurons encode the difference between reward

and aversive events at different epochs of probabilistic classical conditioning trials.J Neurosci 28:11673–11684.

21. Ragozzino ME, Mohler EG, Prior M, Palencia CA, Rozman S (2009) Acetylcholine ac-tivity in selective striatal regions supports behavioral flexibility. Neurobiol Learn Mem91:13–22.

22. Hajnal A, Székely M, Gálosi R, Lénárd L (2000) Accumbens cholinergic interneuronsplay a role in the regulation of body weight and metabolism. Physiol Behav70:95–103.

23. Pizzagalli DA, et al. (2009) Reduced caudate and nucleus accumbens response to re-wards in unmedicated individuals with major depressive disorder. Am J Psychiatry166:702–710.

24. Wacker J, Dillon DG, Pizzagalli DA (2009) The role of the nucleus accumbens androstral anterior cingulate cortex in anhedonia: Integration of resting EEG, fMRI, andvolumetric techniques. Neuroimage 46:327–337.

25. Eshel N, Roiser JP (2010) Reward and punishment processing in depression. Biol Psy-chiatry 68:118–124.

26. Janowsky DS, el-Yousef MK, Davis JM, Sekerke HJ (1972) A cholinergic-adrenergichypothesis of mania and depression. Lancet 2:632–635.

27. Mineur YS, Picciotto MR (2010) Nicotine receptors and depression: Revisiting andrevising the cholinergic hypothesis. Trends Pharmacol Sci 31:580–586.

28. Drevets WC, Furey ML (2010) Replication of scopolamine’s antidepressant efficacy inmajor depressive disorder: A randomized, placebo-controlled clinical trial. Biol Psy-chiatry 67:432–438.

29. Higley MJ, Soler-Llavina GJ, Sabatini BL (2009) Cholinergic modulation of multi-vesicular release regulates striatal synaptic potency and integration. Nat Neurosci12:1121–1128.

30. Svenningsson P, Greengard P (2007) p11 (S100A10)—an inducible adaptor proteinthat modulates neuronal functions. Curr Opin Pharmacol 7:27–32.

31. Egeland M, Warner-Schmidt J, Greengard P, Svenningsson P (2011) Co-expressionof serotonin 5-HT(1B) and 5-HT(4) receptors in p11 containing cells in cerebralcortex, hippocampus, caudate-putamen and cerebellum. Neuropharmacology 61:442–450.

32. Schmidt EF, et al. (2012) Identification of the cortical neurons that mediate antide-pressant responses. Cell 149:1152–1163.

33. Bornstein AM, Daw ND (2011) Multiplicity of control in the basal ganglia: Compu-tational roles of striatal subregions. Curr Opin Neurobiol 21:374–380.

34. Graybiel AM (2008) Habits, rituals, and the evaluative brain. Annu Rev Neurosci31:359–387.

35. Voorn P, Vanderschuren LJ, Groenewegen HJ, Robbins TW, Pennartz CM (2004)Putting a spin on the dorsal-ventral divide of the striatum. Trends Neurosci27:468–474.

36. Limousin P, Martinez-Torres I (2008) Deep brain stimulation for Parkinson’s disease.Neurotherapeutics 5:309–319.

37. Groenewegen HJ, Wright CI, Beijer AV, Voorn P (1999) Convergence and segregationof ventral striatal inputs and outputs. Ann NY Acad Sci 877:49–63.

Warner-Schmidt et al. PNAS | July 10, 2012 | vol. 109 | no. 28 | 11365

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Janu

ary

26, 2

021