Calcium signaling cascade links dopamine D1–D2receptor heteromer to striatal BDNF productionand neuronal growthAhmed Hasbia, Theresa Fanb, Mohammad Alijaniaramb, Tuan Nguyenb, Melissa L. Perreaulta, Brian F. O’Dowda,b,and Susan R. Georgea,b,c,1

bCentre for Addiction and Mental Health, Toronto, ON, Canada M5T 1R8; and Departments of aPharmacology and cMedicine, University of Toronto,Toronto, ON, Canada M5S 1A8

Edited by Floyd E. Bloom, The Scripps Research Institute, La Jolla, CA, and approved September 1, 2009 (received for review April 8, 2009)

Although the perturbation of either the dopaminergic system orbrain-derived neurotrophic factor (BDNF) levels has been linked toimportant neurological and neuropsychiatric disorders, there is noknown signaling pathway linking these two major players. We foundthat the exclusive stimulation of the dopamine D1–D2 receptorheteromer, which we identified in striatal neurons and adult rat brainby using confocal FRET, led to the activation of a signaling cascadethat links dopamine signaling to BDNF production and neuronalgrowth through a cascade of four steps: (i) mobilization of intracel-lular calcium through Gq, phospholipase C, and inositol trisphos-phate, (ii) rapid activation of cytosolic and nuclear calcium/calmodu-lin-dependent kinase II�, (iii) increased BDNF expression, and (iv)accelerated morphological maturation and differentiation of striatalneurons, marked by increased microtubule-associated protein 2 pro-duction. These effects, although robust in striatal neurons from D5�/�

mice, were absent in neurons from D1�/� mice. We also demon-strated that this signaling cascade was activated in adult rat brain,although with regional specificity, being largely limited to thenucleus accumbens. This dopaminergic pathway regulating neu-ronal growth and maturation through BDNF may have consider-able significance in disorders such as drug addiction, schizophrenia,and depression.

brain-derived neurotrophic factor activation � calcium signaling pathway �calcium/calmodulin-dependent kinase II � neuronal maturation �GPCR oligomerization

Dopamine promotes neuronal differentiation, maintenance,and survival (1–4) by modulating the transcription of different

genes. Little however, is known regarding the molecular events thatgovern these dopamine-mediated effects. Evidence has emergedindicating a positive relationship between functions mediated bydopamine and brain-derived neurotrophic factor (BDNF) and itsreceptor TrkB (2–7). However, a direct mechanism bridging do-pamine signaling to BDNF has not yet been described. Classically,dopamine exerts its actions through D1-like (D1, D5) and D2-like(D2, D3, D4) receptors, which regulate activation or inhibition ofcAMP accumulation, through Gs/olf or Gi/o proteins, respectively(8). Other signaling cascades have also been reported (9, 10),including phosphatidylinositol turnover in brain through D1-likereceptor activation (11, 12), but no such activation was observedwhen the cloned D1 receptor was expressed (13–15). These obser-vations led us to the discovery of the dopamine D1–D2 receptorheterooligomer, which is able to mobilize intracellular calcium(15–18). However, the signaling cascade and the physiologicalfunctions of the dopamine D1–D2 receptor heteromer in brain areunknown.

Because calcium is involved in the activation of BDNF signaling(19), we hypothesized that this pathway may be central to dopamineactivation of BDNF and subsequent neuronal maturation anddifferentiation.

In this context, we describe a signaling pathway that linksdopamine action through the D1–D2 receptor heterooligomer to

the expression of BDNF in postnatal striatal neurons and in adultrat brain by a mechanism involving activation of Gq, phospholipaseC (PLC), the mobilization of intracellular calcium, activation ofcytoplasmic and nuclear calcium/calmodulin (CaM)-dependent ki-nase II� (CaMKII�) and subsequently an increase in BDNFexpression. Furthermore, we also highlight the physiological con-sequences resulting from the activation of this signaling pathway onneuronal maturation and growth during development and its exis-tence in nucleus accumbens of adult rat brain. Finally, usingconfocal FRET, we demonstrate the presence of the D1–D2receptor heterooligomer as a physical entity in striatal neurons andrat brain.

ResultsDopamine D1 and D2 Receptors Form Heterooligomers in StriatalNeurons. Immunocytochemistry revealed the majority of postnatalstriatal neurons in culture for 7–21 days expressed D1 and D2receptors mainly at the cell surface and on neurites with a highdegree of colocalization in �90% of the neurons (Fig. 1A and Fig.S1a). Confocal FRET analysis showed that D1 and D2 receptorswere in close proximity with a relative distance of 5–7 nm (50–70Å) localized in microdomains, where FRET efficiency (E) rangedfrom 0.1 to 0.5, higher in the soma and proximal dendrites and lowerin distal processes (Fig. 1B and Fig. S1b). Coimmunoprecipitationof D1–D2 receptor complexes from the striatum is shown (Fig. S1c).Data showed D2 receptor as a broad band of 55- to 70-kDa proteinand an oligomeric form �170 kDa, whereas D1 receptor existed asbands at �55, �70, and �75–80 kDa, probably caused by differentdegrees of glycosylation. The data clearly indicate that D1 and D2receptors can be coimmunoprecipitated from the striatum.

Coexpression, together with the coimmunoprecipitation, and theproximity of the receptors within same neurons indicated a physicalinteraction and heteromer complex formation between the nativelyexpressed dopamine D1 and D2 receptors.

Intracellular Calcium Mobilization Through the Dopamine D1–D2 Re-ceptor Heteromer. The calcium signaling pathway was evaluated byusing cameleon (20). In the absence of extracellular calcium,agonist treatments showed rapid increases in cameleon FRET,reflecting an immediate rise in intracellular calcium (Fig. 1B Insets).Both dopamine and SKF 83959, a D1-D2 heteromer agonist (17,18), dose-dependently mobilized intracellular calcium (Fig. 1B),with a higher maximal peak effect for dopamine (Emax 0.5 versus

Author contributions: A.H. and S.R.G. designed research; A.H., T.F., M.A., T.N., and M.L.P.performed research; A.H., M.L.P., B.F.O., and S.R.G. analyzed data; and A.H. and S.R.G.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1To whom correspondence should be addressed. E-mail: s.george@utoronto.ca .

This article contains supporting information online at www.pnas.org/cgi/content/full/0903676106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0903676106 PNAS � December 15, 2009 � vol. 106 � no. 50 � 21377–21382

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

021

0.35 for SKF 83959), but equivalent EC50 values (170 nM fordopamine and 152 nM for SKF 83959). SCH 23390, a D1-specificantagonist, and raclopride, a D2-specific antagonist, blocked theeffects of both dopamine and SKF 83959 (Fig. 1C), suggesting thatdopamine- or SKF 83859-induced calcium mobilization was medi-ated by the D1–D2 receptor heterooligomer.

Quinpirole, a D2 agonist, at concentrations �10 �M, showed noeffect (Fig. S2a) but at higher concentrations (�10 �M) triggeredcalcium mobilization (Fig. S2a). These quinpirole concentrationsare far higher than its affinity constant (Kd 1.5–2.2 nM) for D2receptor in rat striatal membranes (21) and would not be specific forD2 receptors. However, 1–100 nM quinpirole used concomitantlywith equivalent concentrations of SKF 83959 showed an effect oneto two times greater than the effect of SKF 83959 alone (Fig. S2b),indicating a synergistic effect of D1 and D2 receptor activation. SKF83822, which exclusively activates the cAMP pathway (17, 18),showed no significant effect on intracellular calcium release (Fig.S2c), suggesting the noninvolvement of this pathway. To ensurespecificity of D1–D2 receptor heteromer involvement, the calciummobilization, performed with striatal neurons from dopamine D5receptor null (D5�/�) mice showed robust SKF 83959-triggeredintracellular calcium rise (Fig. S2d), with an amplitude equivalentto that observed in neurons from WT animals (FRET 0.27 � 0.04

versus 0.31 � 0.03, respectively), which was abolished by SCH 22390and raclopride (Fig. S2d), suggesting the involvement of D1 and D2receptors. No calcium mobilization, however, was observed inneurons from D1�/� mice. Thus, intracellular calcium mobilizationby dopamine or SKF 83959 in striatal neurons involved the D1–D2receptor heterooligomer and a synergistic cooperation between thereceptors within the heteromer, without involvement of the dopa-mine D5 receptor.

Characteristics of the D1–D2 Receptor Heteromer-Mediated CalciumRelease. Calcium mobilization studies performed in the absence ofextracellular calcium excluded calcium entry via calcium channels.The exclusive involvement of intracellular calcium was confirmedby preincubation with 10 �M thapsigargine (TPG), which attenu-ated calcium release caused by SKF 83959 by 80% and that causedby dopamine by �75% (Fig. S2e). The source of intracellularcalcium was assessed with 100 �M 2-aminoethoxydiphenyl borate(2-APB), which inhibits inositol trisphosphate (IP3) receptor-dependent calcium stores and abolished calcium mobilization elic-ited by SKF 83959 (Fig. 1D).

Lack of involvement of the Gs–AC pathway in calcium mobili-zation by the D1–D2 receptor heteromer was shown by SQ22536(SQ; 10 �M), an inhibitor of AC, which had no effect on dopamine-dependent mobilization of intracellular calcium (Fig. 1E). Pertussistoxin (PTX), an inhibitor of Gi/o proteins, had no effect on thecalcium release induced by SKF 83959 or dopamine (Fig. 1F),suggesting noninvolvement of Gi/o proteins in this process. Gqinvolvement was assessed by preincubation with YM 254890 (YM;100 nM), a Gq-specific inhibitor (22). No significant effect on basalcalcium levels was noted, but SKF 83959- and dopamine-triggeredcalcium mobilization was inhibited by 90% (Fig. 2A), implicating arole for Gq/11. Immunocytochemistry of Gq/11 (Fig. 2B) revealedit localized largely in the cytosol under basal conditions. When

0.0

0.1

0.2

0.3

Basal SKF 83959 SKF 83959 +SCH 23390

SKF 83959 + Racl

0.0

0.1

0.2

0.3

0.4

0.5

DopamineBasal Dopamine+SCH 23390

Dopamine+ Racl

A

B

C

FR

ET

Ratio

FR

ET

Ratio

D1 D2 Merged FRET

100 nM SKF 83959

0 10 20 30 40 50 60-0.1

0.0

0.1

0.2

0.3

0.4

Time (sec)

Net

FR

ET

Ratio

* *

100 nM Dopamine

0 25 50 75-0.10.00.10.20.30.40.5

Time (sec)

Net

FR

ET R

atio

-12 -11 -10 -9 -8 -7 -6 -5 -40.0

0.1

0.2

0.3

0.4

0.5

0.6

Dopamine [log (M)]

0.7

FR

ET

Ratio

-13 -12 -11 -10 -9 -8 -7 -6 -5 -4

0.0

0.1

0.2

0.3

0.4

SKF 83959 [log(M)]

FR

ET

Ratio

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

Net

FR

ET

Ratio

Net

FR

ET

Ratio

-0.1

0.0

0.1

0.2

0.3

0.4D E* *

*

DopamineBasal Dopamine+ SQ

BasalBasal+ SQ+ 2APB

Basal SKF 83959SKF 83959+ 2APB

0.00

0.25

0.50

0.75

1.00

Net

FR

ET

Ratio

F*

*

Basal Basal+ PTX + PTX

Dopamine Dopamine

Fig. 1. Colocalization, FRET analysis, and calcium mobilization of dopamineD1–D2 receptor heteromer in striatal neurons. (A) Immunocytochemistryshowing endogenously expressed dopamine D1 and D2 receptor colocaliza-tion (merged) and interaction (FRET). (B) D1–D2 receptor heteromer activa-tion leads to rapid (Insets) and dose-dependent mobilization of intracellularCa2� by dopamine or SKF 83959. (C) D1 antagonist SCH (10 �M) or D2antagonist raclopride (10 �M) abolished intracellular Ca2� mobilization trig-gered by D1-D2 receptor heteromer activation by 100 nM dopamine or 100 nMSKF 83959. (D) 2-APB (100 �M), inhibitor of IP3 receptors, abolished theintracellular Ca2� increase elicited by 100 nM SKF 83959. (E) SQ 22536 (SQ, 10�M), an adenylyl cyclase inhibitor, had no effect on intracellular Ca2� mobi-lization triggered by 100 nM dopamine. (F) Pertussis toxin (PTX) (50 nM),inhibitor of Gi/o proteins had no effect on dopamine-induced Ca2� mobiliza-tion. *, significant difference from basal (P � 0.01). Data represent mean�SEM of values from at least three different experiments.

Control 2 min. 5 min.

0.00

0.05

0.10

0.15

Net

FR

ET

Rat

io

*

*

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

Net

FR

ET

Ratio

A*

Basal Basal+ YM + YM

SKF 83959 SKF 83959

SKF 83959+ U73

SKF 83959Dopamine+ U73

B

C

Nuclear pCaMKIIα

0 min 2 min 5 min0

250

500

750

1000

1250

1500

Dopamine (100 nM)

pCaM

KIIα

incr

ease

(% o

f ba

sal)

Total pCaMKII α

0 min 2 min 5 min0

100200300400500600700

Dopamine (100 nM)

pCaM

KIIα

incr

ease

(% o

f ba

sal)

pCaMKII α

E F

D 2 min. 5 min.0 min.

Dopamine

SKF 83959

Fig. 2. Involvement of Gq, intracellular calcium, and CaMKII� in the D1-D2heteromer signaling pathway in striatal neurons. (A) Inhibition of 100 nM SKF83959-triggered intracellular Ca2� mobilization by Gq-specific inhibitor YM254890 (YM, 100 nM). (B) Time-dependent trafficking of endogenous Gq, aftertreatment with vehicle (control, 0 min) or SKF 83959 (100 nM) for 2 and 5 min. (C)Intracellular Ca2� mobilized by SKF 83959 with or without U73122, inhibitor ofPLC. (D) Increase in phosphoCaMKII� (pCaMKII�) following treatment with ve-hicle or 100 nM SKF 83959 for 2 min or 5 min in cytosol and nucleus. (E)Quantification of total pCaMKII� fluorescence (cytosolic � nuclear) over thebasal. (F) Quantification of pCaMKII� fluorescence in the nuclear compartmentover the basal level. Data represent mean �SEM of values from at least threedifferent experiments. *, significant difference from control (P � 0.01).

21378 � www.pnas.org�cgi�doi�10.1073�pnas.0903676106 Hasbi et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

021

neurons were treated with 100 nM SKF 83959 or dopamine for 2–5min, Gq was more concentrated at the cell surface and less presentin the cytosol. Quantification of Gq fluorescence (Fig. S3) revealedagonists increased Gq at the cell surface by 28 � 2.7% after 2 minand �65 � 4% after 5 min (Fig. S3) of treatment, whereas cytosoliclabeling decreased by 21% and 50%, respectively (Fig. S3). To-gether, these results indicated the involvement of Gq, and theexclusion of Gs and Gi/o proteins, in calcium mobilization throughthe D1–D2 receptor heteromer.

The involvement of Gq and sensitivity of the calcium stores to IP3suggested a role for PLC. Treatment with U73122, a PLC inhibitor,attenuated calcium elevations seen with SKF 83959 and dopamine(Fig. 2C), indicating that PLC was involved in the calcium mobili-zation triggered by activation of the D1–D2 receptor heteromer.Thus identified is a signaling pathway in striatal neurons, wherebyactivation of the dopamine D1–D2 receptor heteromer led to thesubsequent rapid activation and translocation of Gq, activation ofPLC and mobilization of intracellular calcium from stores sensitiveto IP3.

Induction of CaMKII� Activation and Nuclear Translocation. Becauseintracellular calcium changes affect CaM, we investigated whetherCaMKII� was modulated through the D1–D2 receptor heteromer-dependent calcium signal. In the absence of extracellular calcium,striatal neurons were treated with vehicle, SKF 83959, or dopaminefor 2 or 5 min (Fig. 2D). In the basal state, a weak signal forphosphoCaMKII� (pCaMKII�) was observed mainly in the cy-tosol. Upon treatment, total pCaMKII� fluorescence increased�2- and 4-fold after 2 and 5 min, respectively (Fig. 2E). Not onlywas there an increase in cellular pCaMKII� levels, but there was a

robust increase of pCaMKII� in the nucleus (Fig. 2D). At basalstate, no pCaMKII� was observed in the nucleus, suggesting that iteither translocated from the cytosol to the nucleus or preexistingnuclear CaMKII� was activated. Quantification of nuclearpCaMKII� levels showed that the effect was rapid, occurring within2 min of treatment, and further dramatically increased after 5 minof treatment, when nuclear pCaMKII� levels rose over basal by�12-fold (Fig. 2F). The SKF 83959-induced activation ofCAMKII� was blocked by raclopride (Fig. S4b), indicating theinvolvement of both dopamine D1 and D2 receptors. CaMKII� wassimilarly activated in striatal neurons from D5�/� mice (Fig. S4a)treated (5 min) with dopamine or SKF 83959. Here again, theincrease in pCaMKII� in cytosol was accompanied by an increasein the nucleus (Fig. S4). No CaMKII� activation by dopamine orSKF 83959 was observed in neurons from D1�/� mice. Takentogether, these results indicated that calcium signaling, elicited byspecific activation of the dopamine D1–D2 receptor heteromer,activated CaMKII� in the cytosol and in the nuclei of striatalneurons with no involvement of the D5 receptor.

Activation of BDNF in Striatal Neurons. Because BDNF gene expres-sion is modulated by nuclear isoforms of CaMKII (23), we assessedwhether BDNF was activated by the D1–D2 heteromer signalingpathway. In the absence of extracellular calcium in the basal state,BDNF immunolabeling showed a weak expression level, mainlylocalized to cytosol, with little or no BDNF expression in neuritesand axons (Fig. 3A, control). Treatment with dopamine or SKF83959 showed a time-dependent increase in BDNF levels, evidentafter 1 h and significantly higher after 2 h at three and seven timeshigher than basal levels (Fig. 3 A and B). It was noted that a

Fig. 3. Striatal enhanced BDNF expression and neuronal maturation throughD1-D2 receptor heteromer activation. (A) Increased BDNF expression after treat-ment with vehicle (control), SKF 83959 (100 nM) for 2 hours, or dopamine (100nM) for 2 or 72 hours, in the absence of extracellular calcium. Note perinuclearBDNF (white arrows), and axonal/dendritic BDNF (yellow arrows). Nuclei werelabeled with DAPI. Magnified images are shown in the Bottom frames. (B)Quantification of BDNF immunofluorescence after treatment with dopamine for5 min, 1 hr, and 2 hrs. (C) Western blot analysis showing an agonist-induced BDNFexpression and the inhibition of this effect by the antagonists, raclopride (10 �M)and SCH (10 �M). (D) Confocal images of striatal neurons treated, in the absenceof extracellular calcium, with vehicle (Left, control), SKF 83959 (Upper Right, 100nM)ordopamine(LowerRight, 100nM), frompostnatalday4throughday10(PN4–10). Note the difference in arborization and connections between control andagonist-treated neurons. (E) Confocal microscopy of immunolabeled MAP2 fol-lowing treatment (PN 4–10), in the absence of extracellular calcium, with vehicle(Left, control), SKF 83959 (Upper Right, 10 nM), or dopamine (Lower Right, 500nM). Nuclei were labeled with DAPI. Note the dense and extensive MAP2 labelingin agonist-treated neurons.

Fig. 4. D1-D2 receptor heteromer localization and enhancement of BDNFexpression in rat nucleus accumbens. (A) Confocal FRET analysis of D1 and D2receptor interactioninratnucleusaccumbensshell region.Anti-D2-Alexa350andanti-D1-Alexa 488 were used as donor and acceptor dipoles. Analysis showsprocessed FRET (pFRET) and distances separating the two receptors . FRET signalwas detected only in neurons coexpressing D1 and D2 receptors. (B) Microdo-mains [regions of interest (ROIs)] within neurons coexpressing D1 and D2 recep-tors (Bi, Upper) or expressing D2 exclusively (Bii, Lower) were analyzed. AverageFRET efficiency (E) and distance between donor and acceptor are shown. Adistance �10 nm indicates no FRET. (C) BDNF expression in rat nucleus accumbenscore and shell regions from animals treated with vehicle (saline), SKF 83959, orSKF 83822. (D) Quantification of BDNF-positive neurons (Left) and densitometrywithin each neuron (Right) from animals treated with vehicle (saline), SKF 83959,or SKF 83822 were assessed. Results represent mean �SEM of values from 4–5rats/group.

Hasbi et al. PNAS � December 15, 2009 � vol. 106 � no. 50 � 21379

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

021

significant perinuclear, likely ER localization of BDNF was appar-ent after agonist treatment during all observation periods, suggest-ing novel synthesis of BDNF (Fig. 3A, white arrows). Agonist-induced BDNF expression was also observed in neurites and axons(Fig. 3A, yellow arrows). A Western blot indicating agonist-inducedproduction of BDNF is shown in Fig. 3C. This effect was inhibitedby the D1 and D2 antagonists (Fig. 3C and Fig. S4b), indicating theinvolvement of the D1–D2 receptor heteromer. BDNF expressionwas also enhanced in striatal neurons from D5�/� mice aftertreatment with SKF 83959 (Fig. S4c), whereas no alteration inBDNF expression was observed in striatal neurons from D1�/�

mice (Fig. S4d), suggesting the exclusive involvement of D1 but notD5 receptors.

Consequences of Activating the D1–D2 Signaling Pathway: NeuronalGrowth and Differentiation. Because the D1–D2 signaling cascadereached the nucleus through CaMKII� and activated local synthesisof BDNF, we surmised a link may exist between the activation ofthe calcium signaling pathway and neuronal developmental and/orsurvival signaling by BDNF. Striatal neurons were treated with10–100 nM SKF 83959 or dopamine from postnatal days 4–10 andcompared with vehicle treatment. Treatments were performedovernight (12–14 h) in the absence of calcium, followed by removalof treatment media and replacement by fresh culture media. Asignificant proportion of neurons treated by the dopaminergicagonists differentiated earlier (Fig. 3D and Fig. S5), with a pheno-type equivalent to that of neurons in culture for at least 15 days (Fig.3D), when they usually reach maturation (3, 4). After agonisttreatment, the neurons were no longer isolated as usually seen upto day 10, but instead had enhanced growth of neurites, formingconnections with others at a distance, indicative of a more differ-entiated and mature stage (Fig. 3D and Fig. S5). We confirmedthese observations by microtubule-associated protein 2 (MAP2)labeling (Fig. 3E and Fig. S5a), which is the major protein thatregulates the structure and stability of microtubules, neuronalmorphogenesis, cytoskeleton dynamics, and organelle trafficking inaxons and dendrites (1). In newborn rat brain, MAP2a appearsbetween postnatal days 10 and 20, during the dendritic growthphase when neurons have reached their mature morphology (1). Inuntreated cells, MAP2 labeling was restricted to the cell body andcertain small processes, probably axons (Fig. 3E and Fig. S5a). Incontrast, striatal neurons treated by SKF 83959 or dopamineshowed dense MAP2 expression, within long processes extendingbetween neurons making contact. This finding indicated that moreMAP2, probably MAP2a, was produced because of dopamineagonist treatments and led to accelerated neuronal maturation andgrowth. The SKF 83959 effect was inhibited in the presence ofraclopride, indicating the involvement of both D1 and D2 receptors(Fig. S5b). Moreover, this effect, while present in neurons derivedfrom D5�/� mice (Fig. S5c), was absent in neurons derived fromD1�/� mice (Fig. S5d), underscoring the necessary involvement ofthe D1 receptor. These data, as a whole, suggested that activationof the D1–D2 receptor heteromer signaling pathway was a keycascade transducing dopaminergic signals involved in the develop-ment, maturation, and differentiation of striatal neurons, throughactivation of BDNF signaling.

Evidence of Activation of the D1–D2 Receptor Heteromer SignalingPathway in Adult Rodent Brain. Colocalization of D1 and D2receptors in some neurons within the striatum (24–27), althoughrepresenting an important finding, does not indicate whether suchreceptors form heteromers. Immunohistochemistry in rat striatumshowed that certain neurons coexpressed D1 and D2 receptors,whereas others expressed only one of the receptors. In caudateputamen, few neurons coexpressed D1 and D2 receptors. Withinthe nucleus accumbens, there were more neurons coexpressing D1and D2 receptors, mainly at the cell surface (Fig. 4 and Fig. S6).Confocal FRET analysis using fluorophore-labeled antibodies

measured the interaction between the colocalized receptors, asillustrated for nucleus accumbens shell (Fig. 4) and core (Fig. S6).FRET analysis showed that in most neurons where D1 and D2receptors were colocalized the receptors were physically closeenough to generate a FRET signal, whereas, noncolocalized re-ceptors were unable to generate any FRET (Fig. 4B and Fig. S6).The distance between the receptors showed D1 and D2 were inclose proximity with an average distance of 4–7 nm (40–70 Å), andhigh FRET efficiency was detected in nucleus accumbens. Thesedata clearly indicated in adult rat nucleus accumbens the presenceof neurons coexpressing D1 and D2 receptors, which existed closeenough to permit energy transfer and therefore may be consideredto be physically interacting, forming receptor heteromers.

We then assessed whether BDNF expression in the adult ratstriatum could be stimulated through activation of the D1–D2heteromer. Rats were treated once daily for a total of threeinjections with saline (control), 0.4 mg/kg of SKF 83959, or 0.4mg/kg of SKF 83822. Changes in BDNF expression were evident innucleus accumbens (Fig. 4C) and not in caudate nucleus, which wasin keeping with the distribution of the D1–D2 heteromers observed.Compared with saline-treated rats, SKF 83959-treated animalsshowed a 49 � 9% increase in the number of BDNF-positive cellsin nucleus accumbens core, whereas no increase was observed inSKF 83822-treated rats (Fig. 4D). Quantification of expression inindividual neurons by densitometry showed that, compared withsaline-treated animals, the level of BDNF increased in cells fromanimals treated with SKF 83959 (Fig. 4D Upper Right), whereas noincrease was observed in SKF 83822-treated animals, confirmingthat the activation of the calcium pathway, and not the cAMPpathway, increased both the number of cells expressing BDNF andthe concentration of BDNF within cells in the nucleus accumbenscore. The number of BDNF-positive neurons of control rats washigher in the nucleus accumbens shell region than core region (Fig.4D Left). This number was not significantly different from saline-treated rats in the shell region of SKF 83959- or SKF 83822-treatedrats (Fig. 4D Lower Left), although a small increase was observedin SKF 83959-treated animals. However, when BDNF expressionwas quantified by densitometry in individual shell neurons the levelof BDNF in cells from animals treated with SKF 83959 increased

D1-D2(IC)

Dopamine or SKF 83959RacloprideSCH 23390

GqYM254890

Gs Gi/o

Intracellular calcium mobilization

Cameleon (FRET)

IP3

Ca

DG + PC

TPG

PLC

2APB

SQNo effect

Translocation

CaMKII

pCaMKII

pCaMKII

Nucleus

Increased BDNF production

Modulation of gene expression

Neurotrophic effects:Growth, differentiationNeuroprotectionSurvival

D1-D2

AC

No effect PTX

IP3R++

Colocalization (ICC)Interaction (FRET)

ER

U73122

Cell surface

Fig. 5. Model of dopamine D1–D2 receptor heteromer signaling pathway.Activation of the D1–D2 receptor heteromer leads to intracellular calcium mobi-lization from IP3 receptor-sensitive stores through a cascade of events involvingrapid translocation of Gq to plasma membrane and activation of PLC. Calciummobilization triggers the activation of CaMKII� in the cytosol and the nuclearcompartment, either by a translocation of activated CaMKII� (pCaMKII�) or theactivation of a nuclear isoform of CaMKII�. In the nucleus, pCaMKII� triggersgene expression and protein synthesis, especially an enhanced BDNF expression.Through its trophic effects, BDNF would trigger the growth, maturation, anddifferentiation of striatal neurons.

21380 � www.pnas.org�cgi�doi�10.1073�pnas.0903676106 Hasbi et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

021

by a significant 26.8% compared with cells from saline-treatedanimals (Fig. 4D Lower Right), whereas no increase was observedin cells from animals treated with SKF 83822.

This finding suggests that the activation of the calcium pathway,and not the cAMP pathway, increased the production of BDNFwithin the nucleus accumbens core and shell neurons of adult ratbrain after D1–D2 receptor heteromer activation.

DiscussionWe describe a dopaminergic signaling pathway occurring inneonatal striatal neurons and adult rat brain, involving activationof the dopamine D1–D2 receptor heterooligomer-, Gq-, PLC-,and IP3 receptor-mediated intracellular calcium mobilization,resulting in CaMKII� activation in both cytosolic and nuclearcompartments, leading to enhanced BDNF production. Thisprocess resulted in enhanced neuronal maturation, differentia-tion, and growth. We also showed through confocal FRET theoccurrence of heteromers of endogenously expressed nativedopamine D1 and D2 receptors in striatal neurons in culture andin adult rat brain in situ.

We showed that �90% of the cultured striatal neurons coexpressboth D1 and D2 receptors. However, this number was lower in adultrat brain than in the cultured neurons and varied among the regions,with a high of �25% in the nucleus accumbens, and a lower degreeof colocalization in the caudate putamen (CPu) (�6%). Our resultsare consistent with the reported levels of D1 and D2 colocalizationin cultured neurons and adult rat striatum. In fact, the D1 and D2colocalization in the striatum has been a matter of long debate. Itis believed that D1 and D2 receptors are localized in two anatom-ically segregated sets of neurons, forming the striatonigral D1-enriched direct pathway and the striatopallidal D2-enriched indi-rect pathway (28–30). Data from BAC transgenic mice, usingEGFP driven by D1 or D2 receptor promoters, were consistent withthis view (30–35). BAC data analysis also showed that at least10–17% of medium spiny neurons (MSNs) coexpress both D1 andD2 receptors (30, 34, 35), with higher prevalence in ventral striatum(up to 17%) than in dorsal striatum (�1–6%) (34, 35). Evidence forneuronal colocalization of D1 and D2 receptors has also beenindicated by electrophysiological studies (36), immunohistochem-istry (24, 25), electron microscopy (37), and retrograde labeling(26). In dissociated, cultured striatal neurons from fetal (38),neonatal (39), and 2- to 3-week-old (40) rats D1 and D2 werecolocalized in a significantly higher number of neurons (60–100%)than in the adult striatum (15, 24, 27, 38, 39). The discrepancybetween the results observed in adult striatum versus those ob-served in cultured striatal neurons may be caused in part by the lackin the cultures of afferents and glial cells. Another explanation maybe a regulation of each receptor type during development, leadingto different levels of receptor colocalization within brain regions.

Our study indicates that these colocalized receptors were in closephysical proximity allowing the formation of receptor heterooli-gomers, both in striatal neurons in culture and in brain tissue.Furthermore, no known functional relevance has been attached tothe neurons where D1 and D2 receptors colocalized, and ourfindings indicate that the receptor heteromers have a uniquesignaling pathway linking dopamine and BDNF through a rapid risein calcium signaling and CaMKII activation. This finding mayindicate a more significant role for the dopamine-activated calciumsignaling cascade during striatal development, whereas it is rele-gated to have a more specialized and circumscribed role in the adultstriatum, largely confined to nucleus accumbens. These findingsalso suggest that in addition to the two separate direct and indirectpathways, another set of neurons form a third pathway, where bothreceptors interact to generate another specialized signal. This mayhelp to explain, at least in part, the cooperativity between D1 andD2 receptors observed by many (40, 41).

Dopamine-induced control of gene expression, which is impor-tant in long-term synaptic plasticity (42, 43), has been shown to

occur in striatum and other brain regions (43), but the molecularmechanisms of the information transfer from the cytoplasm to thenuclei of striatal neurons are still poorly understood (44). Ourresults showing activation of CaMKII� both in the cytoplasm andnucleus, as an immediate consequence of the D1–D2 heteromerstimulation and calcium release, represents an expedient mecha-nism by which this information could lead to rapid gene regulation.Whether activated CaMKII� was directly translocated to the nu-cleus or an intermediary component, such as the recently reporteddopamine-controlled inhibition of nuclear protein phosphatase-1(44), is responsible for the activation of preexisting nuclearCaMKII� remains to be determined.

After the increase in CaMKII� activity, we observed an increasein BDNF expression. We also observed, in rat brain, a significantincrease in the production of BDNF by neurons from the nucleusaccumbens, suggesting a regional specificity for the activation of thispathway in adult brain. It is notable that these results suggested thatBDNF could be synthesized locally by striatal neurons. Thus, thereexists a direct link between dopaminergic signaling and BDNFexpression through activation of the D1–D2 receptor heteromer.Repeated activation of the D1–D2 heteromer pathway led toaccelerated growth of the neurites and connections between striatalneurons, indicating enhanced maturation and differentiation ofstriatal neurons, with notably increased MAP2 expression. Theseresults are consistent with the ability of dopamine and BDNF topromote neuronal maturation, differentiation, and survival and tostimulate lengthening and arborization of neuronal processes (1–4).Our results, modeled in Fig. 5, suggest that the effects of dopaminein this particular case are mediated through the D1–D2 receptorheteromer-, Gq-, PLC-, IP3-dependent signaling pathway, usingcalcium as a second messenger, which is responsible for CaMKII�activation, notably in the nucleus, where it stimulates BDNFsynthesis, which in turn activates protein synthesis responsible forneuronal maturation and differentiation. In adult brain, this sig-naling pathway is region-specific and highly circumscribed, likelyconfined to the neurons expressing the D1–D2 receptor heteromer.

Because of the major roles played by both dopamine and BDNFin many aspects of neuronal maturation and survival, any disequi-librium in the presently described D1–D2 heteromer pathwaylinking dopamine to BDNF may have dramatic consequences thatundermine neuronal morphology, adaptation, and survival, poten-tially leading to neuropsychiatric disorders. Alterations in BDNFlevels caused by complications in the prenatal development period,early childhood events, or adult stress were associated with neuro-psychiatric disorders such as schizophrenia (45), depression (46), ordrug addiction (47). A neurotrophin hypothesis linking dysfunctionof BDNF to the emergence of symptoms of schizophrenia has beenpostulated (48), and there are reports of associations betweendysfunction of calcium signaling and its related proteins, includingGq, IP3, and CaMKII, with schizophrenia (49). A link between D1and D2 receptors was shown to be missing in postmortem striatafrom patients with schizophrenia and Huntington disease (50).

The importance of the different receptor signaling complexes inmediating specific dopamine functions is being revealed. Ulti-mately, alterations of one dopamine receptor in early developmen-tal stages, or even in the adult, may induce differences in the balanceof the heteromeric/homomeric complexes and be at the origin ofcomplicated diseases, such as schizophrenia, depression, or drugaddiction.

In summary, we have demonstrated a signaling pathway triggeredby the activation of the dopamine D1–D2 receptor heteromer thatbridges the action of dopamine to BDNF expression, with animportant physiological function of neuronal maturation andgrowth. A dopamine signaling pathway using calcium as a secondmessenger, targeting the nucleus through CaMKII activation andleading to enhanced BDNF production, may potentially be ofconsiderable importance in the postnatal development of striatalneurons and in nucleus accumbens function in the adult. The

Hasbi et al. PNAS � December 15, 2009 � vol. 106 � no. 50 � 21381

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

021

signaling cascade described here may indicate the presence of athird pathway consisting of neurons containing D1 and D2 receptorheteromers, in addition to the already well-characterized direct andindirect pathways. Finally, our findings also represent an opportu-nity for drug discovery strategies to target particular signalingpathways within the dopaminergic system.

Materials and MethodsConfocal Microscopy FRET and Data Processing. Paraformaldehyde-fixed striatalneurons or floating sections (10 �m) from rat brain were incubated for 24 h at 4 °Cwith primary antibodies highly specific to D1 and D2 receptors (15) and the specie-specific secondary antibodies conjugated to Alexa Fluor 488 and Alexa Fluor 350dyes, respectively. Anti-D2-Alexa Fluor 350 was used as a donor dipole, while anti-D1-Alexa Fluor 488 was used as acceptor dipole. The images were acquired with anOlympus Fluoview FV 1000 laser scanning confocal microscope with a 60�/1.4 NAobjective. The donor was excited with a krypton laser at 405 nm, while the acceptorwasexcitedwithanargonlaserat488nm.Theemissionswerecollectedwith430/20-and 530-nm LP filter. Other FRET pairs (488–568 and 568–647) were tested andshowed comparable results. Eleven images were acquired for each FRET analysis inaccordance with an algorithm (51) (Tables S1 and Table S2). The processed FRET(pFRET) images were then generated based on the described algorithm (51), inwhich: pFRET � UFRET � ASBT � DSBT, where UFRET is uncorrected FRET and ASBTand DSBT are the acceptor and the donor spectral bleed-through signals, respec-tively. The rate of energy transfer efficiency (E) and the distance (r) between the

donor (D) and the acceptor (A) molecules were estimated by selecting small regionsof interest (ROI) using the same images and software, based on the followingequations: E � 1 � IDA/[IDA� pFRET � ((�dd/�aa) � (Qd/Qa))], where IDA is the donorimageinthepresenceofacceptor,�dd and�aa arecollectionefficiencies inthedonorand acceptor channels, respectively, and Qd and Qa are the quantum yields. E isproportional to the sixth power of the distance (r) separating the FRET pair. r � R0

[(1/E) � 1]1/6. Ro is the Forster’s distance.

Calcium Measurements Using Cameleon YC6.1. Calcium mobilization was mea-sured by using cameleon YC6.1 (generous gift from M. Ikura, University ofToronto), an engineered calcium indicator based on the dependence of CaMconformation on elevations of calcium concentration (20). An increase in calciumbinding to CaM leads to a decrease in the distance separating the two flankingproteins,CFPandYFP,andresults inameasurableFRETchange(20).Usingasingleexcitation wavelength at 405 nm, which solely excites CFP, images and fluores-cence emissions data for both CFP and YFP were collected. The experiments wereperformed in live neurons in the absence of extracellular calcium, and the dataobtained from each individual cell were used to calculate the ratios, reflective ofthe energy transferred. The background signal was subtracted from the valuesobtained after drug injection.

Additional materials, methods, and related references are available in SI Text.

ACKNOWLEDGMENTS. The work was supported by a grant from the NationalInstitute of Drug Abuse. S.R.G. is the holder of a Canada Research Chair inMolecular Neuroscience.

1. Schmidt U, et al. (1996) Activation of dopaminergic D1 receptors promotes morpho-genesis of developing striatal neurons. Neuroscience 74:453–460.

2. Iwakura Y, Nawa H, Sora I, Chao MV (2008) Dopamine D1 receptor-induced signalingthrough TrkB receptors in striatal neurons. J Biol Chem 283:15799–15806.

3. Ivkovic S, Ehrlich ME (1999) Expression of the striatal DARPP-32/ARPP-21 phenotype inGABAergic neurons requires neurotrophins in vivo and in vitro. J Neurosci 19:5409–5419.

4. Ivkovic S, Polonskaia O, Farinas I, Ehrlich ME (1997) Brain-derived neurotrophic factorregulates maturation of the DARPP-32 phenotype in striatal medium spiny neurons:Studies in vivo and in vitro. Neuroscience 79:509–516.

5. Berton O, et al. (2006) Essential role of BDNF in the mesolimbic dopamine pathway insocial defeat stress. Science 311:864–868.

6. Guillin O, Abi-Dargham A, Laruelle M (2007) Neurobiology of dopamine in schizo-phrenia. Int Rev Neurobiol 78:1–39.

7. Monteggia LM, et al. (2004) Essential role of brain-derived neurotrophic factor in adulthippocampal function. Proc Natl Acad Sci USA 101:10827–10832.

8. Missale C, Nash SR, Robinson SW, Jaber M, Caron MG (1998) Dopamine receptors: Fromstructure to function. Physiol Rev 78:189–225.

9. Park SK, et al. (2005) Par-4 links dopamine signaling and depression. Cell 122:275–287.10. Beaulieu JM, et al. (2005) An Akt/�-arrestin 2/PP2A signaling complex mediates dopa-

minergic neurotransmission and behavior. Cell 122:261–273.11. Undie AS, Friedman E (1990) Stimulation of a dopamine D1 receptor enhances inositol

phosphates formation in rat brain. J Pharmacol Exp Ther 253:987–992.12. Tang TS, Bezprozvanny I (2004) Dopamine receptor-mediated Ca2� signaling in striatal

medium spiny neurons. J Biol Chem 279:42082–42094.13. Dearry A, et al. (1990) Molecular cloning and expression of the gene for a human D1

dopamine receptor. Nature 347:72–76.14. Pedersen UB, et al. (1994) Characteristics of stably expressed human dopamine D1a and

D1b receptors: Atypical behavior of the dopamine D1b receptor. Eur J Pharmacol267:85–93.

15. Lee SP, et al. (2004) Dopamine D1 and D2 receptor coactivation generates a novelphospholipase C-mediated calcium signal. J Biol Chem 279:35671–35678.

16. So CH, et al. (2005) D1 and D2 dopamine receptors form heterooligomers and coint-ernalize after selective activation of either receptor. Mol Pharmacol 68:568–578.

17. Rashid AJ, et al. (2007) D1–D2 dopamine receptor heterooligomers with uniquepharmacology are coupled to rapid activation of Gq/11 in the striatum. Proc Natl AcadSci USA 104:654–659.

18. Rashid AJ, O’Dowd BF, Verma V, George SR (2007) Neuronal Gq/11-coupled dopaminereceptors: An uncharted role for dopamine. Trends Pharmacol Sci 28:551–555.

19. Cheng A, Wang S, Yang D, Xiao R, Mattson MP (2003) Calmodulin mediates brain-derived neurotrophic factor cell survival signaling upstream of Akt kinase in embryonicneocortical neurons. J Biol Chem 278:7591–7599.

20. Truong K, et al. (2001) FRET-based in vivo Ca2� imaging by a new calmodulin-GFPfusion molecule. Nat Struct Biol 8:1069–1073.

21. Lepiku M, Rinken A, Jarv J, Fuxe K (1997) Modulation of [3H]quinpirole binding todopaminergic receptors by adenosine A2A receptors. Neurosci Lett 239:61–64.

22. Takasaki J, et al. (2004) A novel G�q/11-selective inhibitor. J Biol Chem 279:47438–47445.

23. Takeuchi Y, Yamamoto H, Miyakawa T, Miyamoto E (2000) Increase of brain-derivedneurotrophic factor gene expression in NG108–15 cells by the nuclear isoforms ofCa2�/calmodulin-dependent protein kinase II. J Neurochem 74:1913–1922.

24. Shetreat ME, Lin L, Wong AC, Rayport S (1996) Visualization of D1 dopamine receptorson living nucleus accumbens neurons and their colocalization with D2 receptors.J Neurochem 66:1475–1482.

25. Levey AI, et al. (1993) Localization of D1 and D2 dopamine receptors in brain withsubtype-specific antibodies. Proc Natl Acad Sci USA 90:8861–8865.

26. Deng YP, Lei WL, Reiner A (2006) Differential perikaryal localization in rats of D1 andD2 dopamine receptors on striatal projection neuron types identified by retrogradelabeling. J Chem Neuroanat 32:101–116.

27. Wong AC, Shetreat ME, Clarke JO, Rayport S (1999) D1- and D2-like dopamine recep-tors are colocalized on the presynaptic varicosities of striatal and nucleus accumbensneurons in vitro. Neuroscience 89:221–233.

28. Gerfen CR (2004) The basal ganglia. The Rat Nervous System, ed Paxinos G (Academic,New York), pp 455–508.

29. Le Moine C, Bloch B (1995) D1 and D2 dopamine receptor gene expression in ratstriatum: Sensitive cRNA probes demonstrate prominent segregation of D1 and D2mRNAs in distinct neuronal populations of the dorsal and ventral striatum. J CompNeurol 355:418–426.

30. Lee KW, et al. (2006) Cocaine-induced dendritic spine formation in D1 and D2 dopa-mine receptor-containing medium spiny neurons in nucleus accumbens. Proc Natl AcadSci 103:3399–3404.

31. Gong S, et al. (2003) A gene expression atlas of the central nervous system based onbacterial artificial chromosomes. Nature 425:917–925.

32. Lobo MK, Karsten SL, Gray M, Geschwind DH, Yang XW (2006) FACS-array profiling ofstriatal projection neuron subtypes in juvenile and adult mouse brains. Nat Neurosci9:443–452.

33. Shuen JA, Chen M, Gloss B, Calakos N (2008) Drd1a-tdTomato BAC transgenic miceforsimultaneous visualization of medium spiny neurons in the direct and indirectpathways of the basal ganglia. J Neurosci 28:2681–2685.

34. Bertran-Gonzalez J, et al. (2008) Opposing patterns of signaling activation in dopa-mine D1 and D2 receptor-expressing striatal neurons in response to cocaine andhaloperidol. J Neurosci 28:5671–5685.

35. Matamales M, et al. (2009) Striatal medium-sized spiny neurons: Identification bynuclear staining and study of neuronal subpopulations in BAC transgenic mice. PLoSOne 4:e4770.

36. Calabresi P, Mercuri N, Stanzione P, Stefani A, Bernardi G (1987) Intracellular studies onthe dopamine induced firing inhibition of neostriatal neurons in vitro: Evidence for D1receptor involvement. Neuroscience 20:757–771.

37. Hersch SM, et al. (1995) Electron microscopic analysis of D1 and D2 dopamine receptorproteins in the dorsal striatum and their synaptic relationships with motor corticos-triatal afferents. J Neurosci; 15:5222–5237.

38. Aizman O, et al. (2000) Anatomical and physiological evidence for D1 and D2 dopaminereceptor colocalization in neostriatal neurons. Nat Neurosci 3:226–230.

39. Iwatsubo K, et al. (2007) Dopamine induces apoptosis in young, but not in neonatal,neurons via Ca2�-dependent signal. Am J Physiol 293:C1498–C1508.

40. Capper-Loup C, Canales JJ, Kadaba N, Graybiel AM (2002) Concurrent activation ofdopamine D1 and D2 receptors is required to evoke neural and behavioral phenotypesof cocaine sensitization. J Neurosci 22:6218–6227.

41. Harsing LG Jr., Zigmond MJ (1997) Influence of dopamine on GABA release in striatum:Evidence for D1–D2 interactions and nonsynaptic influences. Neuroscience 77:419–429.

42. Nestler EJ (2001) Molecular basis of neural plasticity underlying addiction. Nat RevNeurosci 2:119–128.

43. Thoenen H (1995) Neurotrophins and neuronal plasticity. Science 270:593–598.44. Stipanovich A, et al. (2008) A phosphatase cascade by which rewarding stimuli control

nucleosomal response. Nature 453:879–884.45. Angelucci F, Brene S, Mathe AA (2005) BDNF in schizophrenia, depression, and corre-

sponding animal models. Mol Psychiatry 10:345–352.46. Kozisek ME, Middlemas D, Bylund DB (2008) Brain-derived neurotrophic factor and its

receptor tropomyosin-related kinase B in the mechanism of action of antidepressanttherapies. Pharmacol Ther 117:30–51.

47. Graham DL, et al. (2007) Dynamic BDNF activity in nucleus accumbens with cocaine useincreases self-administration and relapse. Nat Neurosci 10:1029–1037.

48. HashimotoT, et al. (2005) Relationship of brain-derived neurotrophic factor and itsreceptor TrkB to altered inhibitory prefrontal circuitry in schizophrenia. J Neurosci25:372–383.

49. Lidow MS (2003) Calcium signaling dysfunction in schizophrenia: A unifying approach.Brain Res Brain Res Rev 43:70–84.

50. Seeman P, Niznik HB, Guan HC, Booth G, Ulpian C (1989) Link between D1 and D2dopamine receptors is reduced in schizophrenia and Huntington diseased brain. ProcNatl Acad Sci USA 86:10156–10160.

51. Chen Y, Elangovan M, Periasamy A (2005) FRET data analysis: The algorithm. MolecularImaging: FRET Microscopy and Spectroscopy, eds Periasamy A, Day RN (Oxford UnivPress, New York), pp 126–145.

21382 � www.pnas.org�cgi�doi�10.1073�pnas.0903676106 Hasbi et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

021