Reexposure to nicotine during withdrawal increasesthe pacemaking activity of cholinergichabenular neuronsAndreas Görlicha,b, Beatriz Antolin-Fontesa,b, Jessica L. Ablesa, Silke Frahmb, Marta A. Slimakb, Joseph D. Doughertyc,and Inés Ibañez-Tallona,b,1

aLaboratory of Molecular Biology, The Rockefeller University, New York, NY 10065; bMolecular Neurobiology Group, Max Delbrück Center for MolecularMedicine, 13125 Berlin, Germany; and cDepartment of Genetics and Department of Psychiatry, Washington University School of Medicine, St. Louis, MO 63110

Edited by Jean-Pierre Changeux, Centre National de la Recherche Scientifique, Institut Pasteur, Paris, France, and approved September 3, 2013 (received forreview July 11, 2013)

The discovery of genetic variants in the cholinergic receptor nico-tinic CHRNA5-CHRNA3-CHRNB4 gene cluster associated withheavy smoking and higher relapse risk has led to the identificationof the midbrain habenula–interpeduncular axis as a critical relaycircuit in the control of nicotine dependence. Although clear rolesfor α3, β4, and α5 receptors in nicotine aversion and withdrawalhave been established, the cellular and molecular mechanisms thatparticipate in signaling nicotine use and contribute to relapse havenot been identified. Here, using translating ribosome affinity pu-rification (TRAP) profiling, electrophysiology, and behavior, wedemonstrate that cholinergic neurons, but not peptidergic neurons,of the medial habenula (MHb) display spontaneous tonic firing of2–10 Hz generated by hyperpolarization-activated cyclic nucleotide-gated (HCN) pacemaker channels and that infusion of the HCN pace-maker antagonist ZD7288 in the habenula precipitates somatic andaffective signs of withdrawal. Further, we show that a strong, α3β4-dependent increase in firing frequency is observed in these pace-maker neurons upon acute exposure to nicotine. No change in thebasal or nicotine-induced firing was observed in cholinergic MHbneurons from mice chronically treated with nicotine. We observe,however, that, during withdrawal, reexposure to nicotine doublesthe frequency of pacemaking activity in these neurons. These find-ings demonstrate that the pacemaking mechanism of cholinergicMHb neurons controls withdrawal, suggesting that the height-ened nicotine sensitivity of these neurons during withdrawalmay contribute to smoking relapse.

nAChRs | TRAP profiling

The tobacco epidemic kills nearly six million smokers a year,primarily from lung cancer. The success rate for smoking

cessation without pharmacological treatment is only 3–5% andless than 30% with nicotine-replacement therapies (1). Nicotine,the addictive component of tobacco, mediates its action by ac-tivating nicotinic acetylcholine receptors (nAChRs), which re-spond to the endogenous neurotransmitter acetylcholine (ACh).Like other drugs of abuse, chronic nicotine exposure promotesadaptations in neuronal circuits that sustain the use of cigarettes(2, 3). Upon nicotine cessation in humans, a withdrawal syn-drome—characterized by negative somatic and affective symp-toms such as irritability, anxiety, depressed mood, and loss ofconcentration—develops (2, 3) and contributes to the high prob-ability of smoking relapse. In rodents that have been chronicallytreated with nicotine, withdrawal can be induced either by phar-macological precipitation or by cessation of nicotine administration.Somatic and affective signs of withdrawal differentially manifest inthese two cases (3, 4). Additionally, studies in knockout mice haverevealed a distinct contribution of various nAChR subtypes to so-matic and affective withdrawal signs (5, 6), illustrating the molecularcomplexity of nicotine dependence, withdrawal, and relapse.Converging evidence from genome-wide association studies

and animal models has strongly implicated the nicotinic receptor

subunits α3, α5, and β4 with heavy tobacco use and decreasedsuccess in smoking-cessation therapy in humans (7, 8), and withnicotine aversion and withdrawal in laboratory animals (6, 9, 10).These nicotinic receptor subunits (with the exception of α5) arenot present in the mesolimbic dopamine tract (typically impli-cated in addiction disorders) (2, 11) but are concentrated in themedial habenula (MHb) and interpeduncular nucleus (IPN),which together comprise a major cholinergic tract in the mam-malian brain that conveys convergent information from the limbicforebrain to the midbrain via the fasciculus retroflexus (2, 12, 13).A number of studies have begun to dissect the medial and

lateral habenular nuclei in an effort to determine whether sub-population differences in cytoarchitecture and connectivity couldbe correlated to molecular markers and electrophysiologicalproperties (14–17). For instance, the dorsal and ventral parts ofthe MHb can be clearly divided based on their respective en-richment in the neuropeptide Tachykinin 1, also known as sub-stance P (SP), or the ACh synthesizing enzyme choline ace-tyltransferase (ChAT), and the segregation of their efferents todifferent parts of the IPN (18). A recent study showed that MHbChAT-positive neurons corelease glutamate and ACh upon dif-ferent patterns of blue-light stimulation in transgenic ChAT-Channelrhodopsin mice (19), but that sufficient ACh to activatenAChRs in IPN neurons is detected only once a certain thresholdof tonic firing is reached. Given that it has been reported thatmedial habenular neurons generate tonic trains of action poten-tials (15), similar to neurons in the mesolimbic dopamine tract, wewanted to address whether this property was restricted to cholin-ergic neurons, and whether this tonic firing could be modulated invivo by nicotine. Here, we analyzed the pacemaking activity andtranslating ribosome affinity purification (TRAP) molecular profileof MHb neurons. By comparing ChAT and SP neurons, we foundthat cholinergic MHb neurons are equipped with pacemaker

Significance

According to the World Health Organization, tobacco con-sumption causes the death of close to 6 million people eachyear, yet successful attempts to quit smoking are very rare. Thepresent study identifies a group of neurons in the brain thatrespond differently to nicotine after a period of abstinence,suggesting that altered activity of these neurons may con-tribute to difficulties with smoking cessation.

Author contributions: A.G., B.A.-F., and I.I.-T. designed research; A.G., B.A.-F., J.L.A., S.F.,M.A.�S., J.D.D., and I.I.-T. performed research; A.G., B.A.-F., J.D.D., and I.I.-T. analyzed data;and A.G. and I.I.-T. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE43164).1To whom correspondence should be addressed. E-mail: iibanez@rockefeller.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1313103110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1313103110 PNAS | October 15, 2013 | vol. 110 | no. 42 | 17077–17082

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Mar

ch 4

, 202

0

hyperpolarization-activated cyclic nucleotide-gated (HCN) chan-nels that confer them with intrinsic pacemaking activity, and thatinfusion of the HCN antagonist ZD7288 into the MHb of nicotinenaïve mice precipitates somatic and affective signs of withdrawal.This pacemaking activity is increased by nicotine via activation ofα3β4-containing (α3β4*) nAChRs and is not changed in micetreated with chronic nicotine, but is further potentiated by expo-sure to nicotine in mice undergoing nicotine withdrawal.

ResultsCholinergic, but Not Peptidergic, Neurons in theMHb Show SpontaneousAction-Potential Firing. We confirmed the clear segregation of cho-linergic and peptidergic neurons in the two mouse models used inthis study: ChAT-DW167 mice and Tabac mice. ChAT-DW167mice are transgenic for a bacterial artificial chromosome (BAC) ofthe ChAT gene and regulatory regions driving expression of anenhanced green fluorescent protein (eGFP)-tagged L10a ribo-somal subunit (eGFP-L10a) (20). Tabac (Transgenic α3β4α5cluster) mice carry a BAC of the Chrnb4-Chrna3-Chrna5 genecluster driving expression of eGFP in α3β4* nAChR-positiveneurons (10). As shown in Fig. 1 A and B, cholinergic neuronsare concentrated in the ventral two-thirds of the MHb whereasSP-containing cells are exclusively localized in the dorsal partof the nucleus. ChAT-DW167 and Tabac mice show colocali-zation of eGFP and ChAT immunoreactivity (Fig. 1B and Fig.S1), which indicates that both BAC vectors accurately drivetransgene expression to cholinergic neurons of the MHb.We next used ChAT-DW167 mice for patch-clamping ex-

periments in brain slices. We performed whole-cell recordings incholinergic and peptidergic neurons of the MHb (Fig. 1 C–F).Eighty-five percent of the ChAT-positive neurons in the MHbshowed tonic action-potential firing in the current clampmode (58out of 68 neurons) (Fig. 1C). Tonic firing could be recorded for upto 1 h. The action-potential frequency ranged from2 to 10Hz, withan average frequency of 4.2 ± 0.25 Hz. None of the patched GFP-negative, peptidergic neurons in the dorsal 1/3 of theMHb showedthis pacemaking activity (n = 5) (Fig. 1D). Upon current injections(200 pA for 300ms), eGFP–positive cholinergic neurons respondedwith three ormore action potentials (APs) (Fig. 1E) whereas eGFP-negative peptidergic neurons in the dorsal 1/3 of the MHb showedonly one or two APs (Fig. 1F). To address whether the spontaneousactivity is synchronizedwithin theMHb, we performed cell-attachedrecordings from two adjacent eGFP-positive cholinergic neurons(Fig. 1G). In all five of these double cell-attached recordings, neu-rons showed a similar firing frequency, but tonic firing was not syn-chronized (Fig. 1H). These results show that only ChAT-positiveneurons in theMHbhave pacemaking activity and that spontaneousAPs are unsynchronized between different cells.

TRAP Analysis of Cholinergic MHb Neurons Reveals Enrichment ofChannels and Receptors Involved in Pacemaking. Given the cleardistinction between cholinergic and peptidergic neurons in theirpacemaking activity, we sought to identify the channels andreceptors that could confer this property uniquely to the cho-linergic subpopulation. We applied the TRAP methodology tohabenular homogenates from the bacTRAP ChAT-DW167 micetargeting cholinergic neurons (20) (Fig. 2 A and B). This meth-odology employs affinity purification of eGFP-tagged ribosomesand the mRNAs that are being translated in the targeted cellpopulation (20). Three replicates of RNA captured from cho-linergic neurons (TRAP) and total RNA from the habenulardissection (Total) were subjected to amplification followed byhybridization with Affymetrix microarrays.As shown in Fig. 2A and Dataset S1, the TRAP samples show

robust enrichment of known markers of MHb neurons, includingChrna3, Chrnb4, Tac2, and Gpr151, in agreement with otherstudies (14, 17), whereas nonneuronal transcripts (e.g., glial genes)aredepleted(red inFig. 2A).We identified transcripts for over 1,100genes that were at least twofold enriched in MHb cholinergic neu-rons compared with total RNA (Dataset S2). Comparative analysiswith all previously collected TRAP cell types (20, 21) identified

390 transcripts that were relatively enriched and/or specific toMHb cholinergic neurons as measured by the specificity indexstatistic (pSI < 0.01) (21), indicating that these cells have a highlyunique transcriptional signature within the brain. Hierarchicalclustering with all previous profiled cell types confirms the relativeuniqueness of this cell type (Fig. 2B). These neurons fall only looselynear a neuron cluster with various neuromodulatory populationsof cells, such as hypothalamic hypocretin neurons and cholinergicneuronsof thebasal forebrain, and they are quitedistinct fromothercholinergic populations such as spinal cord and hindbrain motorneurons. Conducting a Gene Ontologies analysis on the 390 tran-scripts that distinguish these cells reveals a remarkable prepon-derance of molecules with transcription-factor activity, ion-bindingactivity (mostly channels), and receptors, includingWnts (Fig. S2and Dataset S3). This characteristic gene expression profile sug-gests that these cells have a unique physiological profile as well.To better understand the physiology of these neurons, we

further analyzed the ion channels and receptors enriched inthese cells (Dataset S1). First, we observed extremely high ex-pression levels of the big potassium (BK) channel (encoded bythe potassium large conductance calcium-activated channel gene;Kcnma1) and the Cacnb3 (voltage-gated calcium channel auxiliary

Fig. 1. Cholinergic, but not peptidergic, neurons in the MHb show spon-taneous action-potential firing. (A) Immunohistochemical detection of eGFP-tagged L10a ribosomal subunit in a coronal section from the ChAT-DW167TRAP line. The boxed area indicates the MHb, shown in detail in B. (Scalebar: 1 mm.) (B) Immunofluorescent detection of eGFP, ChAT, and SP in MHbof the ChAT-DW167 mouse. The Insets demonstrate each individual channel.ChAT and eGFP colocalize in the ventral part (outlined right) of MHbwhereas Substance P is expressed in the dorsal part (outlined left). (Scale bar:100 μm.) ChAT, acetylcholine transferase; d, dorsal; MHb, medial habenula;3V, third ventricle; v, ventral. (C) Patch-clamp recordings show spontaneouspacemaking activity in ChAT-positive neurons. (D) ChAT-negative neurons inthe dorsal 1/3 of the MHb show no spontaneous activity. (E) ChAT-positiveMHb neurons respond to 200 pA current injection with a train of APs. (F)ChAT-negative neurons have a very diminished response to 200 pA currentinjection. (G) Representative image of a double cell-attached recording fromadjacent ChAT-positive neurons. (H) ChAT-positive neurons showed un-synchronized firing, but similar action-potential frequency.

17078 | www.pnas.org/cgi/doi/10.1073/pnas.1313103110 Görlich et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 4

, 202

0

b3 subunit). In addition, the L-type voltage-gated calcium channela1d (Cacna1d, also known as Cav1.3), although not present in themicroarray, is highly expressed in these neurons (see Allen BrainAtlas: http://mouse.brain-map.org/experiment/show/68798037) andhas been shown to drive pacemaker activity in dopamine (DA)neurons (22). We were also interested in the pacemaking HCNchannels, which are differentially expressed in the habenula (16,23). Our microarray data demonstrated a clear enrichment of theHcn3 channel in these neurons whereas Hcn1 andHcn2 were at orbelow the level of background. BecauseHcn4was not present in ourmicroarrays but seems to be specific for the cholinergic part of theMHb (16, 23), we assessed its expression by reverse transcriptionquantitative polymerase chain reaction (RT-qPCR) on TRAPsamples [F(5,6) = 69.83, P < 0.0001, ANOVA] (Fig. 2C). Alto-gether, these data indicate that cholinergic MHb neurons differ-entially express high levels of Hcn3, Hcn4, BK, L-type Cav1.3, andα3β4* nAChRs that may endow these neurons with specific elec-trical properties that underlie their pacemaking capability.

Block of HCN Channel-Mediated Pacemaker Activity in MHb NeuronsResults in a Withdrawal-Like Phenotype. To further characterize thepacemaking activity of MHb neurons, we used pharmacologicalblockers. In five of five ChAT-positive neurons, the tonic AP-firingwas unchanged by the application of a drug mixture containingAMPA, NMDA, GABAA, GABAB, and nAChR blockers [50 μM2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sul-fonamide (NBQX), 50 μM D-(-)-2-amino-5-phosphonopentanoicacid (d-APV), 100 μM Picrotoxin, 1 μM CGP55845, and 3 μMmecamylamine, respectively] (Fig. 3A). Tonic AP-firing is there-fore driven by intrinsic factors and independent of glutamatergic,GABAergic, or cholinergic inputs.HCN channels mediate spontaneous pacemaking activity in

the heart and in other brain areas (23) and are enriched in MHbneurons (Fig. 2C). To test whether these channels regulate

pacemaking in cholinergic neurons, we applied the nonselectiveHCN blocker ZD7288 (20 μM) to MHb slices. As shown in Fig.3B, spontaneous firing was strongly reduced by 62 ± 10.5%. Wealso observed a smaller but significant contribution of the L-Typecalcium (20 μM nifedipine increased tonic firing by 33 ± 3.9%)and BK channels (1 μM of paxilline, decreased tonic firing by25 ± 2.0%.) to the pacemaking frequency (Fig. S3). These resultssupport our findings from the TRAP analysis indicating thatHCN, BK, and L-type calcium ionic currents are present in MHbcholinergic neurons and contribute to the generation and mod-ulation of their intrinsic pacemaking activity.To assess the behavioral relevance of the pacemaker activity of

cholinergic neurons, ZD7288 was bilaterally microinjected intothe MHb of mice through implanted cannulae (Fig. 3D). Imme-diately after the infusion, mice were observed for 15 min for anyunique behavior. Microinjection of ZD7288 increased the numberof typical somatic signs of withdrawal, including scratching, jump-ing, body and paw tremors, genital licking, and retropulsion (P =0.031) (Fig. 3C). Mice injected with ZD7288 also showed reducedcenter area horizontal activity (P = 0.036) (Fig. 3C). This reducedcenter activity is indicative for increased anxiety, an affective signof nicotine withdrawal. These data demonstrate that inhibition ofMHb pacemaker activity in vivo precipitates a withdrawal-likephenotype even in mice that have not experienced nicotine.

Tonic Firing Frequency Increases uponNicotineApplication andDependson α3β4* nAChRs.Given the high concentration of nAChRs in thecholinergic subpopulation (Fig. S1, Fig. 2A, and Dataset S1) andtheir critical role in nicotine consumption and withdrawal (6, 9,10), we next wanted to determine the contribution of nAChRs tothe spontaneous pacemaking activity in the MHb. The 81 nMnicotine free base increased tonic AP firing by a factor of 1.6 (Fig.4A), an effect that was abolished by the application of the non-specific nAChR blocker mecamylamine (3 μM) (Fig. 4B). The nic-otine-dependent increase in the number of spontaneous APs in theMHb is therefore mediated by nAChRs and is not caused by effectsof nicotine on HCN (24), L-type calcium (25), or BK channels (26).To determine which nAChR combination is responsible for the

effect of nicotine on MHb pacemaking firing, we used a numberof antagonists and peptide toxins. Application of 50 nM Bun-garotoxin (α7*-blocker), 100 nM Conotoxin MII (α3β2*, β3*-blocker), 10 μM 3-(4)-dimethylaminobenzylidine anabaseine(DMAB) (α4β2*, α7*-blocker), or 1 μM dihydro-β-erythroidinehydrobromide (DHβE) (α4β2*, α4β4*-blocker) did not alter theeffects of nicotine on tonic AP firing (Fig. 4D and Fig. S4) whereaspreincubation with the selective blockers for α3β4* nAChRsconotoxin AuIB (Fig. 4 C and D) and SR16584 (Fig. 4D and Fig.S4) abolished the increase in tonic firing produced by nicotine.The specificity for α3β4 nAChRs was further supported by theuse of Tabac mice in which Chrnb4 overexpression increasesα3β4* nAChR levels (10). In these mice, nicotine led to a 3.27 ±0.50-fold increase in spontaneous AP-activity in MHb neurons(Fig. S4). These results demonstrate that the nicotine-meditatedincrease in pacemaking activity is independent of α3β2*, α4β2*,α4β4*, α7*, or β3* nAChR subunits but is the result of α3β4*nAChRs activation.

Nicotine Withdrawal, but Not Chronic Nicotine, Potentiates Nicotine-Induced Pacemaking Activity. We next sought to determine thefunctional consequences of chronic nicotine treatment andwithdrawal on the spontaneousAPactivity in theMHb.To analyzetolerance to chronic nicotine, we used a previously describedprotocol (4, 27).Mice were given nicotine in the drinking water for28 d and recorded on the last day or withdrawn from nicotine andrecorded on the next day (see experimental paradigm in Fig. S5).The water contained 2% saccharin to mask the bitter taste ofnicotine, which was increased from 65 mg/L to 163 mg/L nicotinefree base during the first week, and maintained at 163 mg/L nic-otine free base during the following weeks. Based on the volumeintake (Fig. S5), mice consumed 26.2mg·kg−1·day−1 of nicotine freebase. InC57BL/6Jmice, this dose correlates to plasma cotinine (the

Tota

l Hab

enul

a R

NA

0

0

000

00

00

0

00

0

0

0

MHb TRAP RNA

GlulMbp

PlpMogMagMobpAldh1L1

Galc

Tac2Pou4f1

Chrna3Chrnb4

Sstr2Slc18a3

1 2 3 4 5

12

34

5

ooo

oo

o

o

oo

o

ooo

o

Hypocretin N. /hypSerotonergic N. /hdb

Cholinergic N. /bfCholinergic N. /str

Cholinergic N. /hbPnoc+ N. /ctxL5 Corticospinal N. /ctx

L6 Corticothalamic N. /ctxN. /ctx

L5 Corticostriatal N. /ctxDrd1 MSN /strDrd2 MSN /strCholinergic N. /scCholinergic N. /hdb

Stellate & Basket N. /cbPurkinje N. /cb

Cort N. /ctxInner Golgi N. /cb

Granule N. /cbOligodendrocytes /ctxOligodendrocytes /cb

Oligodendrocytes Progenitor Cells /ctxAstroglia /cbBG /cb

BG & Unipolar Brush N. /cbAstrocytes /ctx

A BR

elat

ive

expr

essi

on TotalIP

Hcn3 ChAT

C

0

5

10

15*

*

Hcn4

*

Fig. 2. TRAP profiling of MHb cholinergic neurons reveals a unique tran-scriptional signature. (A) Scatterplot of microarray data comparing meantranscript levels in cholinergic cell TRAP (x axis) with total habenular dissectionRNA (y axis), for all 17,000 available transcripts. Transcripts known to be foundin cholinergic cells (blue) are robustly enriched in the TRAP sample whereasnonneuronal transcripts are depleted (red, glial genes). Biotinylated controlsare unchanged between samples (green circles). Black lines at 0.5-, 1-, and2-fold. Red line is background threshold as established by fold change of glialgenes. Axes in log scale. (B) Hierarchical clustering of the transcriptional profileof cholinergic MHb neurons compared with all previously measured cell typesreveals that these cells are highly unique. They are only loosely similar to otherneuromodulatory populations and are clearly distinct from other cholinergiccells, such as motor neurons of spinal cord and hindbrain. bf, basal forebrain;BG, Bergman glia; cb, cerebellum; ctx, cortex; hb, habenula; hdb, hindbrain;hyp, hypothalamus; MSN,medium spiny neuron; N, Neurons; sc, spinal cord; str,striatum. (C) Quantification (mean ± SEM) of expression of the Hcn3 and Hcn4isoforms and ChAT genes by qRT-PCR in ChAT-TRAP Immunoprecipitate (IP)versus total RNA. Hcn3, Hcn4, and ChAT are significantly enriched in the IPfraction compared with total. *P < 0.05, Bonferroni post hoc test.

Görlich et al. PNAS | October 15, 2013 | vol. 110 | no. 42 | 17079

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Mar

ch 4

, 202

0

major metabolite of nicotine) levels of 43 ng/mL (27). This con-centration is comparable with nicotine plasma levels measured inthe afternoon in smokers (10-50 ng/mLof cotinine, or 60–310 nMofnicotine) (28). Controlmice were given 2% saccharin.Wepreparedacutebrain slices fromthesemiceandanalyzedMHbChAT-positiveneurons. Recordings in these spontaneously active neurons showedthat the basal pacemaking frequency was indistinguishable be-tween nicotine-treated and saccharin-control groups (Fig. 5A)and that bath application of nicotine resulted in a similar increaseof firing frequency in both groups (fold increase for saccharin-control, 1.63± 0.11; nicotine-treated, 1.71± 0.08) (Fig. 5B andC).This fold increase was the same as that observed in untreated mice

(Fig. 4A), indicating that chronic treatment with nicotine does notchange either the basal or the nicotine-enhanced pacemaking.In a second group, mice were treated as described before, but

given water only to drink for the 14–18 h before preparation of theacute brain slices. The basal AP frequency before nicotine appli-cation was unchanged between the groups, as in the previousexperiments (control withdrawal, 4.07 ± 0.39 Hz; nicotine with-drawal, 3.49 ± 0.28 Hz; P = 0.234) (Fig. 5D). However, when nic-otine was bath applied, the fold increase of spontaneous APfrequency was significantly higher in mice undergoing nicotinewithdrawal compared with the saccharin-withdrawal control group(control, 1.58 ± 0.12; nicotine withdrawal, 2.07 ± 0.17; P = 0.0014)(Fig. 5EandF).This increasewasblockedby thenAChRantagonistmecamylamine (Fig. 5G), demonstrating that this potentiationdepends on the activity of nAChRs. Behaviorally, mice dem-onstrate spontaneous somatic signs of withdrawal during thesame time frame aswhen the nicotine recordings were performed(that is 14–18 h after removing nicotine from the drinking bottlefollowing 4 wk chronic nicotine exposure) (Fig. 5H). We concludethat chronic exposure to nicotine alters the activity of nAChRs incholinergic neurons of the MHb such that, during withdrawal,these neurons display significantly increased pacemaking activityupon reexposure to nicotine.

DiscussionThe cellular and molecular mechanisms underlying how thebrain signals nicotine withdrawal and recognizes its previousexposure to nicotine remain poorly understood. The studiespresented here establish that cholinergic habenular neuronspossess intrinsic pacemaking activity that is mediated by HCNpacemaker channels and that inhibition of pacemaking activity,by infusion of HCN antagonists in the MHb, precipitates with-drawal-like behavior. In addition we find that fast fluctuations innicotine levels, but not chronic nicotine, accelerate the pace-making frequency of these neurons via α3β4* nAChR. Strikingly,mice undergoing withdrawal from nicotine double the pace-making activity of these neurons upon reexposure to nicotine.Given the genetic association of α3β4* receptors with smokingrelapse, our data suggest that altered pacemaking activity inmedial habenular neurons upon nicotine withdrawal may con-tribute to difficulties with smoking cessation.It is intriguing that the pacemaking activity in the MHb is re-

stricted to cholinergic neurons when one considers themechanismsof ACh release from habenular neurons and the signaling path-ways that maymodulate this release. For example, it was recently

A B

C D

0 10 20 30 400.0

0.5

1.0

1.5

2.0

Time [min]

Nor

mal

ized

#AP

s/20

s ZD 7288

n=5

0 5 10 150.0

0.5

1.0

1.5

2.0

Time [min]

Nor

mal

ized

#AP

s/20

s

n=5

Drug mixture

3V

MHb

LHb

Hipp

*

Saline ZD72880

5

10

15

Num

ber o

f som

atic

sig

ns

Saline ZD72880

50

100

150

200

Act

ivity

in C

e nte

rAre

a

*

Fig. 3. Block of HCN pacemaker channels mediating theautonomous tonic firing of MHb neurons precipitates with-drawal-like somatic and affective signs. (A) Application of 50μM NBQX, 50 μM d-APV, 100 μM Picrotoxin, 1 μM CGP55845,and 3 μM Mecamylamine (drug mixture) to block gluta-matergic, GABAergic, and cholinergic inputs had no effect onspontaneous action-potential firing in ChAT-positive MHbneurons. (B) The nonselective HCN blocker ZD7288 (20 μM)reduced tonic AP firing by 60%. (C) After microinjection ofZD7288 into the MHb, mice (n = 8) displayed increasednumber of withdrawal-like somatic signs, not present in micemicroinfused with saline (n = 8). They were also less active inthe center area, indicative of increased anxiety, an affectivewithdrawal sign. *P < 0.05, unpaired t test. (D) Schematicdiagram showing injection sites. Image displays red-fluores-cent dye infused into MHb. Tissue counterstained with DAPI.Hipp, hippocampus; LHb, lateral habenula; MHb, medialhabenula; 3V, third ventricle. (Scale bar: 100 μm.)

Fig. 4. Tonic firing increases upon nicotine application and depends on α3β4*nAChRs. (A) Normalized number of APs per 20 s showing a 1.6-fold increase intonic AP firing after nicotine application. Insets show representative cell-at-tached recordings from MHb neurons. Coapplication of the nAChR blockermecamylamine (3 μM) (B) or the α3β4* nAChR blocker Conotoxin AuIB (C)abolished the effects of nicotine on AP activity. (D) Effects of nicotine uponcoapplication with the indicated nAChR blockers. Percent frequency changewas calculated as the change between baseline AP activity (first 5 min) and thelast minute of nicotine application. The concentrations used were as follows:Nicotine, 81nM;Mecamylamine, 3 μM;ConotoxinMII, 250 nM; ConotoxinAuIB,10 μM; SR 16584, 25 μM; DHβE, 1 μM; DMAB, 10 μM; Bungarotoxin, 50 nM.

17080 | www.pnas.org/cgi/doi/10.1073/pnas.1313103110 Görlich et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 4

, 202

0

demonstrated that cholinergic MHb neurons corelease glutamateand ACh, but, only upon tonic photostimulation (≥20 Hz), theamount of released ACh is sufficient to activate nAChRs inpostsynaptic IPNneurons (19). Interestingly, both in this study andin recordings from rat MHb neurons in vivo (29), this frequencythreshold for ACh release is not reached under basal conditions.However, given the large α3β4*-dependent increase in pacemak-ing activity that we have observed in response to acute nicotine, itseems probable that elevation of ACh levels in the MHb asa consequence of physiological input activity or because of tobaccosmoking will be important for the regulation of ACh release in theIPN. The fact that HCN, L-type Cav1.3, and BK channels co-operate to generate spontaneous pacemaking activity in MHbneurons suggests that additional pathways impinging on thesechannels can also alter tonic firing and thus ACh release in theIPN. For example, the identification of “cAMP-dependent proteinkinase regulator activity” as a major functional category in cho-linergic MHb neurons is interesting in light of the central role ofcAMP in dopaminergic signaling (30) and the regulation of neu-ronal activity in reward circuitry (31). We anticipate that furtherstudies of this pathway and of other molecular components iden-tified in the TRAP studies reported here can lead to an improvedunderstanding of tonic firing in the MHb and its regulation in re-sponse to the many inputs to this important midbrain circuit.Given the complex behavioral effects of nicotine, and recent

studies demonstrating a critical role for the MHb in nicotineintake, aversion, and withdrawal (6, 9, 10), the demonstrationthat pacemaking activity in the MHb responds to nicotine andthat this modulation is altered during withdrawal is important. Itis evident from these studies that the activity of HCN channels ispositively modulated by nicotine and that this effect is mediatedby nAChRs. However, whether this modulation is occurring di-rectly by altering HCNs (i.e., phosphorylation or cAMP changes)or indirectly through other signal transduction events initiated bynAChRs will require further investigation. Our finding that thebasal pacemaking of these α3β4*-enriched neurons is not changedis consistent with studies indicating that chronic nicotine treatmentpreferentially up-regulates α4β2* receptors (32) but not α3β4*receptors (33, 34). During a single withdrawal episode, 3H nicotine

binding sites increase in cerebral cortex and midbrain (35). Incontrast, repeated cycles of nicotine treatment and withdrawalresult in progressive up-regulation of 3H epibatidine sites in striatumand hippocampus but not in the cortex (36). Given these studies,and the cell-specific response we have demonstrated here, it wouldbe interesting to compare cytisine-resistant radioligand bindingstudies under a variety of withdrawal conditions to distinguish theresponses of α4β2* and α3β4* receptors. It seems probable that cell-specific regulation of particular nAChR combinations may underliethe different behavioral and physiological effects of chronic nicotineexposure, withdrawal, and relapse. This specificity for nAChRsubtypes is particularly important in light of recent human geneticassociation studies showing that smokers with the high geneticrisk variants in the CHRNA5-CHRNA3-CHRNB4 locus not onlyhave a predisposition to heavy smoking but will also respond differ-ently to pharmacological treatments for smoking cessation (7, 8).Finally, it is important to consider the pacemaking activity of

cholinergic neurons in the MHb and its modulation by nicotine inthe context of the widespread neural circuitry mediating nicotinedependence andwithdrawal (37).Nicotine, likemanyother drugs ofabuse, is known to have an impact in the mesocorticolimbic DAreward center (2, 38). Interestingly, DA neurons show both spon-taneous tonic firing (with a frequency of 1–5 Hz, similar to MHbneurons) (39) and phasicfiring. The existence of these twomodes offiring is thought to contribute to the malleability of the DA system(38), and it follows that this could also be the case in the MHb.Phasic burst firing induced by afferent stimuli causes transient DArelease and signals acute positive reinforcement by nicotine (38).Tonic firing maintains constant DA levels and is known to decreaseduring nicotine withdrawal (40). The actions of nicotine in this sys-temare complexbecause it coactivatesDAneurons andGABAergicinterneurons, which both express nAChRs (2). For instance, nico-tine greatly increases the frequency of the tonic discharge of DAneurons by activation of β2* nAChRs (41) but also contributes toburst firing by activation of β2* in GABAergic interneurons (42).Furthermore, subcellular compartmentalization of the nAChRsplays an important role in their physiological activities (2). Givenour demonstration that tonic activity in cholinergic neurons in theMHb is modulated by nicotine, and the complex distribution of

Fig. 5. Nicotine withdrawal, but not chronic nico-tine treatment, increases the response to nicotine.(A) Nicotine-treated mice show similar basal APfrequencies compared with control animals. (B) Theeffects of 81 nM nicotine on slices of control andtreated mice were indistinguishable. (C) The foldincrease in tonic firing after nicotine applicationwas not different between the control and thetreated group. (D) The basal AP frequencies in slicesfrom treated mice withdrawn from nicotine 14–18 hbefore slice preparation (withdrawal group) andcontrol animals were indistinguishable. (E) Nicotineeffects on AP firing were enhanced in the with-drawal group compared with the control group. (F)The fold increase in tonic firing after nicotine ap-plication was significantly higher in the withdrawalgroup, compared with the control group (P =0.0014). (G) Coapplication of 3 μM Mecamylamineand 81 nM nicotine in slices from mice withdrawnfrom nicotine completely blocks the nicotine-inducedincrease in spontaneous AP firing. (H) Quantificationof spontaneous somatic signs of withdrawal 14–18 hafter nicotine (n = 9) and saccharin (n = 10) treat-ment cessation. *P < 0.05, unpaired t test.

Görlich et al. PNAS | October 15, 2013 | vol. 110 | no. 42 | 17081

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Mar

ch 4

, 202

0

nAChRs in the MHb–IPN axis, we anticipate that mechanismsregulating the output of this circuitry will also be complex. Con-sequently, a detailed understanding of the Hb–IPN mechanismsthat control nicotine withdrawal and relapse will require furthercell-specific molecular and electrophysiological studies of addi-tional cell types in this interesting and ancient midbrain circuit.

Materials and MethodsDetails for all materials andmethods can be found in SIMaterials andMethods.

Mice. Transgenic ChAT-DW167 and Tabac mice have been previously de-scribed (SI Materials and Methods). All studies were done in accordance withNational Institutes of Health and Institutional Animal Care and Use guide-lines. All procedures and protocols were approved by the Animal CareCommittee of The Rockefeller University and the Max Delbrück Center.

Slice Preparation and Electrophysiological Recordings. Patch pipettes hadresistances of 4–8 MΩ when filled with a solution containing (in mM): 105K-gluconate, 30 KCl, 10 Hepes, 10 phosphocreatine, 4 ATP-Mg2+, 0.3 GTP(pH adjusted to 7.2 with KOH). Electrophysiological responses were recordedat 33–36 °C.

Translating Ribosome Affinity Purification and Quantitative RT-PCR. Translatingribosome affinity purification (TRAP) was conducted in triplicate as previouslydescribed using the TRAP line ChAT-DW167 (SI Materials and Methods). Immu-noprecipitated (IP) and total mRNA fractions purified for TRAP were subjectedto RT-PCR.

Histology and Confocal Imaging. Staining of ChAT-DW167 and Tabac micewith anti-GFP, anti-ChAT, and anti-SP antibodies was done as described in SIMaterials and Methods.

Chronic Nicotine Treatment and Nicotine Withdrawal. All nicotine doses arereported as free base. Micewere given nicotine in the drinkingwater for 28 d,and withdrawal studies were conducted using the spontaneous nicotinewithdrawal model as previously described (4, 27).

Surgical Procedures for Microinjections. Mice were implanted with cannulaguides into the MHb (coordinates, anteroposterior, −1.7 mm from bregma;mediolateral, ±0.3 mm from midline; dorsoventral, −2.9 mm from skulllevel). ZD7288 or saline was microinjected at a volume of 0.5 μL bilaterallyfor over 90 s.

Statistical Analysis. Sets of data are presented by their mean values and SEMs.The unpaired two-tailed Student t test was used when comparing two sets ofdata with normal distribution.

ACKNOWLEDGMENTS.We thank CuidongWang, Syed Samin Shehab, MonikaSchwarz-Harsi, and Julio Santos-Torres for technical assistance. This workwas supported by Deutsche Forschungsgemeinschaft (DFG) Grant GO 2334/1-1 (to A.G.), by National Institute of Neurological Disorders and Stroke(NINDS) Grant 4R00NS067239-03 (to J.D.D.), by Helmholtz AssociationGrant 31-002 and Sonderforschungsbereich (SFB) Grant SFB665 (to I.I.T.),by the Irma L. and Abram S. Croll Charitable Trust (J.L.A.), and by the NationalInstitutes of Health/National Institute of Mental Health Conte Center GrantMH074866.

1. Stead LF, Perera R, Bullen C, Mant D, Lancaster T (2008) Nicotine replacement therapyfor smoking cessation. Cochrane Database Syst Rev (1):CD000146.

2. Changeux JP (2010) Nicotine addiction and nicotinic receptors: Lessons from geneti-cally modified mice. Nat Rev Neurosci 11(6):389–401.

3. Paolini M, De Biasi M (2011) Mechanistic insights into nicotine withdrawal. BiochemPharmacol 82(8):996–1007.

4. Grabus SD, et al. (2005) Nicotine physical dependence and tolerance in the mousefollowing chronic oral administration. Psychopharmacology (Berl) 178(2-3):183–192.

5. Jackson KJ, Martin BR, Changeux JP, Damaj MI (2008) Differential role of nicotinicacetylcholine receptor subunits in physical and affective nicotine withdrawal signs.J Pharmacol Exp Ther 325(1):302–312.

6. Salas R, Sturm R, Boulter J, De Biasi M (2009) Nicotinic receptors in the habenulo-interpeduncular system are necessary for nicotine withdrawal in mice. J Neurosci29(10):3014–3018.

7. Thorgeirsson TE, et al. (2008) A variant associated with nicotine dependence, lungcancer and peripheral arterial disease. Nature 452(7187):638–642.

8. Bierut LJ, et al. (2007) Novel genes identified in a high-density genome wide associ-ation study for nicotine dependence. Hum Mol Genet 16(1):24–35.

9. Fowler CD, Lu Q, Johnson PM, Marks MJ, Kenny PJ (2011) Habenular α5 nicotinicreceptor subunit signalling controls nicotine intake. Nature 471(7340):597–601.

10. Frahm S, et al. (2011) Aversion to nicotine is regulated by the balanced activity of β4and α5 nicotinic receptor subunits in the medial habenula. Neuron 70(3):522–535.

11. Exley R, McIntosh JM, Marks MJ, Maskos U, Cragg SJ (2012) Striatal α5 nicotinicreceptor subunit regulates dopamine transmission in dorsal striatum. J Neurosci32(7):2352–2356.

12. Grady SR, et al. (2009) Rodent habenulo-interpeduncular pathway expresses a large va-riety of uncommon nAChR subtypes, but only the alpha3beta4* and alpha3beta3beta4*subtypes mediate acetylcholine release. J Neurosci 29(7):2272–2282.

13. Sutherland RJ (1982) The dorsal diencephalic conduction system: A review of theanatomy and functions of the habenular complex. Neurosci Biobehav Rev 6(1):1–13.

14. Aizawa H, Kobayashi M, Tanaka S, Fukai T, Okamoto H (2012) Molecular character-ization of the subnuclei in rat habenula. J Comp Neurol 520(18):4051–4066.

15. Kim U, Chang SY (2005) Dendritic morphology, local circuitry, and intrinsic electro-physiology of neurons in the rat medial and lateral habenular nuclei of the epi-thalamus. J Comp Neurol 483(2):236–250.

16. Poller WC, et al. (2011) Lateral habenular neurons projecting to reward-processingmonoaminergic nuclei express hyperpolarization-activated cyclic nucleotid-gatedcation channels. Neuroscience 193:205–216.

17. Quina LA, Wang S, Ng L, Turner EE (2009) Brn3a and Nurr1 mediate a gene regulatorypathway for habenula development. J Neurosci 29(45):14309–14322.

18. Contestabile A, et al. (1987) Topography of cholinergic and substance P pathways inthe habenulo-interpeduncular system of the rat: An immunocytochemical and mi-crochemical approach. Neuroscience 21(1):253–270.

19. Ren J, et al. (2011) Habenula “cholinergic” neurons co-release glutamate and acetylcholineand activate postsynaptic neurons via distinct transmission modes. Neuron 69(3):445–452.

20. Doyle JP, et al. (2008) Application of a translational profiling approach for thecomparative analysis of CNS cell types. Cell 135(4):749–762.

21. Dougherty JD, et al. (2013) The disruption of Celf6, a gene identified by translationalprofiling of serotonergic neurons, results in autism-related behaviors. J Neurosci33(7):2732–2753.

22. Putzier I, Kullmann PH, Horn JP, Levitan ES (2009) Cav1.3 channel voltage de-pendence, not Ca2+ selectivity, drives pacemaker activity and amplifies bursts innigral dopamine neurons. J Neurosci 29(49):15414–15419.

23. Santoro B, et al. (2000) Molecular and functional heterogeneity of hyperpolarization-activated pacemaker channels in the mouse CNS. J Neurosci 20(14):5264–5275.

24. Griguoli M, et al. (2010) Nicotine blocks the hyperpolarization-activated current Ihand severely impairs the oscillatory behavior of oriens-lacunosum moleculare inter-neurons. J Neurosci 30(32):10773–10783.

25. Stevens TR, Krueger SR, Fitzsimonds RM, Picciotto MR (2003) Neuroprotection bynicotine in mouse primary cortical cultures involves activation of calcineurin andL-type calcium channel inactivation. J Neurosci 23(31):10093–10099.

26. Kuhlmann CR, Trümper JR, Tillmanns H, Alexander Schaefer C, Erdogan A (2005) Nico-tine inhibits large conductance Ca(2+)-activated K(+) channels and the NO/-cGMP sig-naling pathway in cultured human endothelial cells. Scand Cardiovasc J 39(6):348–352.

27. Sparks JA, Pauly JR (1999) Effects of continuous oral nicotine administration on brainnicotinic receptors and responsiveness to nicotine in C57Bl/6 mice. Psychopharma-cology (Berl) 141(2):145–153.

28. Matta SG, et al. (2007) Guidelines on nicotine dose selection for in vivo research.Psychopharmacology (Berl) 190(3):269–319.

29. Zhao H, Rusak B (2005) Circadian firing-rate rhythms and light responses of rathabenular nucleus neurons in vivo and in vitro. Neuroscience 132(2):519–528.

30. Girault JA, Greengard P (2004) The neurobiology of dopamine signaling. Arch Neurol61(5):641–644.

31. Nestler EJ (2004) Molecular mechanisms of drug addiction. Neuropharmacology47(Suppl 1):24–32.

32. Gaimarri A, et al. (2007) Regulation of neuronal nicotinic receptor traffic and ex-pression. Brain Res Brain Res Rev 55(1):134–143.

33. Marks MJ, et al. (2004) Subsets of acetylcholine-stimulated 86Rb+ efflux and [125I]-epibatidine binding sites in C57BL/6 mouse brain are differentially affected by chronicnicotine treatment. Neuropharmacology 46(8):1141–1157.

34. Nguyen HN, Rasmussen BA, Perry DC (2004) Binding and functional activity of nico-tinic cholinergic receptors in selected rat brain regions are increased following long-term but not short-term nicotine treatment. J Neurochem 90(1):40–49.

35. Pietilä K, Lähde T, Attila M, Ahtee L, Nordberg A (1998) Regulation of nicotinic re-ceptors in the brain of mice withdrawn from chronic oral nicotine treatment. NaunynSchmiedebergs Arch Pharmacol 357(2):176–182.

36. Hilario MR, Turner JR, Blendy JA (2012) Reward sensitization: Effects of repeated nico-tine exposure and withdrawal in mice. Neuropsychopharmacology 37(12):2661–2670.

37. Volkow ND, Wang GJ, Fowler JS, Tomasi D (2012) Addiction circuitry in the humanbrain. Annu Rev Pharmacol Toxicol 52:321–336.

38. Sesia T, Grace AA (2012) Shifting pharmacology of nicotine use and withdrawal:Breaking the cycle of drug abuse. Proc Natl Acad Sci USA 109(8):2697–2698.

39. Grace AA (1991) Phasic versus tonic dopamine release and the modulation of dopaminesystem responsivity: A hypothesis for the etiology of schizophrenia.Neuroscience 41(1):1–24.

40. Grieder TE, et al. (2012) Phasic D1 and tonic D2 dopamine receptor signaling doubledissociate the motivational effects of acute nicotine and chronic nicotine withdrawal.Proc Natl Acad Sci USA 109(8):3101–3106.

41. Picciotto MR, et al. (1998) Acetylcholine receptors containing the beta2 subunit areinvolved in the reinforcing properties of nicotine. Nature 391(6663):173–177.

42. Tolu S, et al. (2013) Co-activation of VTA DA and GABA neurons mediates nicotinereinforcement. Mol Psychiatry 18(3):382–393.

17082 | www.pnas.org/cgi/doi/10.1073/pnas.1313103110 Görlich et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 4

, 202

0