Ligand-Induced Regulation and Localization of Cannabinoid CB₁ and Dopamine D_2L Receptor Heterodimers

Abstract

The cannabinoid CB₁ (CB₁) and dopamine D₂ (D₂) receptors are coexpressed in the basal ganglia, an area of the brain involved in such processes as cognition, motor function, and emotional control. Several lines of evidence suggest that CB₁ and D₂ receptors may oligomerize, providing a unique pharmacology in vitro and in vivo. However, limited information exists on the regulation of CB₁ and D₂ receptor dimers. We used a novel technique, multicolor bimolecular fluorescence complementation (MBiFC) to examine the subcellular localization of CB₁-D_2L heterodimers as well as D_2L-D_2L homodimers in a neuronal cell model, Cath. a differentiated cells. MBiFC was then used to explore the effects of persistent ligand treatment on receptor dimerization at the plasma membrane and intracellularly. Persistent (20-h) agonist treatment resulted in increased formation of CB₁-D_2L heterodimers relative to the D_2L-D_2L homodimers. The effects of the D₂ agonist quinpirole were restricted to the intracellular compartment and may reflect increased D_2L receptor expression. In contrast, treatment with the CB₁ receptor agonist (2)-cis-3-[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-trans-4-(3-hydroxypropyl) cyclohexanol (CP55, 940) produced increases in both membrane and intracellular CB₁-D_2L heterodimers independently of alterations in CB₁ receptor expression. The effects of CB₁ receptor activation were attenuated by the CB₁ antagonist 1-(2,4-dichlorophenyl)-5-(4-iodophenyl)-4-methyl-N-4-morpholinyl-1H-pyrazole-3-carboxamide (AM281) and were both time- and dose-dependent. The effects of CB₁ activation were examined further by combining MBiFC with a constitutively active CB₁ receptor mutant, CB₁T210I. These studies demonstrated that the expression of CB₁T210I increased intracellular CB₁-D_2L heterodimer formation. In summary, agonist-induced modulation of CB₁-D_2L oligomerization may have physiological implications in diseases such as Parkinson's disease and drug abuse.

Increasing evidence suggests that G protein-coupled receptors (GPCRs) may function in receptor dimeric or higher order oligomeric complexes (for review, see Milligan, 2008). One set of receptors that has received significant attention relevant to oligomerization is the CB₁ cannabinoid (CB₁) receptor and dopamine D₂ (D₂) receptor (for review, see Fuxe et al., 2008). It is thought that the cannabinoid system negatively modulates dopamine circuits as activation of the CB₁ receptor leads to an attenuation of dopamine signaling (Laviolette and Grace, 2006). The CB₁ receptor is widely expressed in the central nervous system, with great abundance in the basal ganglia (Herkenham et al., 1991). CB₁ receptors are located on striatal GABAergic neurons (Herkenham et al., 1991), and they are also found on dendrites in both the dorsal striatum and the nucleus accumbens (Pickel et al., 2006). The D₂ receptor exists as two splice variants, D_2S (short) and D_2L (long). The D_2S variant is highly expressed on presynaptic dopaminergic neurons, whereas the D_2L variant is found postsynaptically on dopaminergic neurons throughout the striatum (Khan et al., 1998; Usiello et al., 2000). ¹ These observations reveal that CB₁ and D_2L receptors have overlapping expression patterns in the striatum and also suggest that they are colocalized in neurons in the nucleus accumbens (see references within Kearn et al., 2005; Pickel et al., 2006).

It has been reported that CB₁ and D₂ receptors oligomerize, providing unique pharmacology in vitro and in vivo (Glass and Felder, 1997; Jarrahian et al., 2004; Kearn et al., 2005; Marcellino et al., 2008). For example, it was demonstrated in primary rat striatal neurons that concurrent activation of Gα_i/o-coupled CB₁ and D₂ receptors resulted in stimulation of cAMP accumulation (Glass and Felder, 1997). Subsequent experiments using recombinant CB₁ and D_2L receptors suggested that D_2L receptor activation promoted a switch in CB₁ receptor coupling from Gα_i/o to Gα_s (Glass and Felder, 1997). One proposed mechanism for D₂ receptor modulation of CB₁-G protein coupling may involve receptor oligomerization. This hypothesis was examined by demonstrating a physical interaction between CB₁ and D_2L receptors using coimmunoprecipitation (Kearn et al., 2005). The same investigators also revealed that the CB₁-D_2L receptor complex can be dynamically modulated by receptor agonists. More recent studies have examined CB₁-D_2L heteromers using FRET techniques (Marcellino et al., 2008). Using human embryonic kidney cells transiently transfected with fluorescently tagged CB₁ and D_2L receptors, a FRET interaction was detected. However, no significant changes in the FRET signal were detected after short-term exposure to CB₁ or D_2L receptor agonists (Marcellino et al., 2008). The ability of CB₁ and D_2L receptors to interact is consistent with suggestion of a CB₁-D_2L heterodimer. Additional behavior and biochemical data support further the physiological relevance of CB₁ and D₂ receptors heterodimers (Fuxe et al., 2008). However, limited information exists on the cellular localization and regulation of CB₁-D_2L receptor heterodimers. Despite the therapeutic potential of drugs targeting these receptors, the effect of persistent receptor activation on the dynamics of receptor oligomerization has not been explored.

The most common techniques currently being used to study the physical association of GPCRs include coimmunoprecipitation and traditional resonance energy transfer (FRET and BRET) techniques (Vidi and Watts, 2009). These techniques are typically limited to the study of a single protein-protein complex. In addition, coimmunoprecipitation does not allow for detection of an interacting protein complex within a living cell. To gain further insight into GPCR dimerization in live cells, we recently established the use of multicolor bimolecular fluorescence complementation (MBiFC) (Hu et al., 2002; Shyu et al., 2006) as a tool to investigate GPCR homo- and heteromer oligomerization (Vidi et al., 2008a,b). MBiFC allows for the detection of two separate protein-protein complexes in living cells by visualizing the fluorescence complementation of two distinct spectral variants of green fluorescent protein (Hu and Kerppola, 2003). Moreover, this technique can be used to measure the relative amounts of homodimer versus heterodimer formation in a cell region-specific manner (Vidi et al., 2008b).

The present study uses MBiFC to examine CB₁-D_2L heterodimers and D_2L-D_2L homodimers in Cath. a differentiated (CAD) cells. CAD cells are a neuronal cell model that express GAP-43, synaptotagmin, and synaptosome-associated protein of 25 kDa and upon differentiation, form neurite-like processes (Qi et al., 1997). The present results provide additional evidence for the existence of CB₁ and D_2L receptor oligomers. We also revealed that persistent agonist (i.e., dopaminergic or cannabinergic) treatment favors the formation of the CB₁-D_2L heterodimer relative to the formation of the D_2L-D_2L homodimer. The D₂ agonist-mediated effects were accompanied by an increase in D_2L receptor expression, whereas the CB₁ agonist-mediated changes in heterodimer formation appeared to involve primarily CB₁ receptor activation. These results provide further insight into the dynamic nature of CB₁-D_2L oligomerization.

Materials and Methods

Materials.

Human CB₁ and D_2L cDNAs were obtained from the Missouri S&T cDNA Resource Center (Rolla, MO). Growth media (Dulbecco's modified Eagle's medium), quinpirole, and sulpiride were obtained from Sigma-Aldrich (St. Louis, MO). Fetal bovine serum and bovine calf serum were purchased from Thermo Fisher Scientific (Waltham, MA). Penicillin/streptomycin/amphotericin B antibiotic/antimycotic was purchased from Invitrogen (Carlsbad, CA). Forskolin was purchased from Tocris Bioscience (Ellisville, MO). CP55,940 was a generous gift from Pfizer Pharmaceuticals (New York, NY). [³H]cAMP (25 Ci/mmol) was purchased from PerkinElmer Life and Analytical Sciences (Boston, MA). [³H]Spiperone (91 Ci/mmol) and [³H]SR141716A (42 Ci/mmol) were obtained from GE Healthcare (Chalfont St. Giles, Buckinghamshire, UK). Specific cellular compartment markers (mCherry-mem, YFP-ER, YFP-Endo, and YFP-Golgi) were gifts from Dr. Catherine Berlot (Weis Center for Research, Danville, PA).

Expression Vectors.

Full-length human CB₁ and D_2L cDNAs were amplified by polymerase chain reaction (PCR) using oligonucleotides with EcoRI, XbaI, or XhoI restriction sites and omitting the stop codons. The PCR products were digested with either EcoRI/XbaI or EcoRI/XhoI and ligated into the corresponding pBiFC vectors. These expression vectors contain nonfluorescent fragments of the N and C termini of the enhanced yellow fluorescent protein (Venus) and the enhanced cyan fluorescent protein (Cerulean). The N-terminal fragments (VN or CN) include residues 1 to 172, whereas the C-terminal fragment of Cerulean (CC) includes residues 155 to 238. This cloning strategy places the fragment on the C terminus of the receptors. In addition, the CB₁ and D_2L receptor PCR products were digested with either EcoRI/XbaI or EcoRI/XhoI and ligated into expression vectors containing the full-length Venus or Cerulean proteins resulting in the CB₁ and D_2L receptors tagged at the C terminus with either Venus or Cerulean. The CB₁ receptor mutant (CB₁T210I) was generated using the QuikChange kit according to the supplier's protocol (Stratagene, La Jolla, CA) in pcDNA3-CB₁ and then subcloned into the pBiFC vectors using EcoRI and XbaI restriction enzyme sites. All constructs were verified by DNA sequencing.

Cell Culture and Transient Transfections.

CAD cells were maintained as described previously (Vortherms and Watts, 2004). For microscopic evaluation of BiFC, CAD cells were grown to approximately 70% confluence in four-well LabTek chambered coverslips (Nalge Nunc International, Rochester, NY). Cells were transfected with 1 μl of Lipofectamine 2000 (Invitrogen) according to the manufacturer's recommendations. In MBiFC experiments, CB₁-VN (500 ng), D_2L-CC (300 ng), and D_2L-CN (300 ng) were transiently cotransfected with 20 ng of either mCherry-Mem, YFP-Endo, YFP-ER, or YFP-Golgi depending on the experiment. Twenty-four hours after transfection, the growth media and transfection reagent were replaced with 400 μl of warm phosphate-buffered saline (PBS), and images were taken using a charge-coupled device camera mounted on a TE2000-U inverted fluorescence microscope (Nikon, Melville, NY) equipped with a 100-W mercury lamp and band-pass filters (Chroma Technology Corp., Rockingham, VT) for Venus (excitation at 500/20 nm; emission at 535/30 nm), Cerulean (excitation, 430/25 nm; emission, 470/30 nm), or mCherry (Texas Red, excitation, 572/23 nm; emission, 625/25 nm). Fluorescent images were acquired using MetaMorph software (Molecular Devices, Sunnyvale, CA). For MBiFC experiments investigating the effects of receptor ligands on receptor dimer population, the cells were transfected as described above and 4 h after transfection, the appropriate drug treatment was added to the growth medium for an additional 20 h before image acquisition.

Quantitative Image Analysis.

Quantification of fluorescent signals was performed as described previously using ImageJ software (http://rsb.info.nih.gov/ij/; Hu et al., 2002; Supplemental Fig. 1). In each experiment, approximately 40 to 50 individual cells were quantified. Ten microscopic fields at 60× magnification were acquired as stacks of images from the YFP, CFP, and Texas Red channels corresponding to the fluorescent signals from Venus, Cerulean, and mCherry proteins, respectively. Background fluorescence intensity was measured in each channel in an area devoid of cells and subtracted from the fluorescent signals. The signals corrected for background fluorescence were then scaled to a factor equal to that of the inverse of the exposure time for each pixel intensity measurement. The images of the mCherry-Mem membrane marker signal were used to select cells for image analysis and to normalize BiFC signals (Supplemental Fig. 1). Cellular analysis of BiFC signals was performed in two parts. First, the fluorescent signal intensity maximum at the membrane was determined by drawing a perpendicular line through the membrane using the mCherry-Mem image. The maximum signal intensity was determined in all three channels, YFP, CFP, and Texas Red to estimate the BiFC signals at the membrane. The BiFC signal intensity in the intracellular space was determined by outlining the intracellular compartment (excluding the plasma membrane) and determining the average pixel intensity in all three channels, YFP, CFP, and Texas Red, to estimate the intracellular BiFC signals. Cells with saturated signals as well as cells with signals that were 1.2 times lower than background were not used for quantification. BiFC experiments assessing bleed-through/overflow of Cerulean or Venus in the opposite channels (i.e., YFP or CFP) revealed minimal cross-talk. Specifically, complemented Cerulean contributed less than 2% of the YFP signal and complemented Venus contributed less than 3% of the CFP signal (data not shown). Venus/Cerulean fluorescence ratios exhibit a non-Gaussian distribution; therefore, median values were calculated and averaged between experiments.

Cyclic AMP Accumulation Assays.

CAD cells were grown to 70% confluence in 24-well plates and were transiently transfected as described previously (Vidi et al., 2008a). CAD cells were either transfected with 300 ng/well D_2L constructs or 500 ng/well CB₁ constructs. All drugs were diluted in Earle's balanced salt solution assay buffer (Earle's balanced salt solution containing 2% bovine calf serum, 0.025% ascorbic acid, and 15 mM HEPES, pH 7.4) and added to the cells on ice. Determination of cAMP accumulation was performed by incubating the transfected CAD cells with forskolin (10 μM) in the absence and presence of either CP55,940 (10 μM) or quinpirole (10 μM) for 15 min at 37°C. All assays were performed in the presence of the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (500 μM) and terminated with ice-cold 3% trichloroacetic acid. Quantification of cAMP accumulation was determined using a competitive binding assay as described previously (Vortherms and Watts, 2004).

Radioligand Binding Assays.

Single point radioligand binding assays were used to estimate CB₁ and D_2L receptor densities after drug treatments as described previously (Vidi et al., 2008a). CAD cells were plated in a 12-well plate and were grown to 70% confluence before being transiently transfected with CB₁-VN, D_2L-CN, and D_2L-CC using 2 μl/well of Lipofectamine 2000. Four hours after transfection, the appropriate drug treatment was added in triplicate to the growth medium and transfection reagent. The cells were incubated for an additional 20 h before single point radioligand binding assays. Cells were washed three times with 500 μl of receptor binding buffer (50 mM Tris and 4 mM MgCl₂, pH 7.4). The cells were lysed with 500 μl of ice-cold lysis buffer (1 mM HEPES and 2 mM EDTA, pH 7.4) for 10 min on ice. The cells were removed from each well by trituration, and crude cell membranes were collected by centrifugation (30,000g for 15 min at 4°C). Membrane pellets were resuspended by mechanical homogenization in 1 ml of receptor binding buffer. For CB₁ receptor binding, the addition of 0.5% bovine serum albumin to the receptor binding buffer was used to decrease nonspecific binding. Crude cell membranes (approximately 30 μg in 150 μl) were added in duplicate to the assay tubes to determine both nonspecific and total binding. For CB₁ binding, nonspecific binding was defined by 10 μM nonradioactive SR141716A (essentially identical levels of nonspecific binding were obtained using 10 μM AM281; data not shown). All tubes contained a near-saturating amount of [³H]SR141716A (50 μl; final concentration, ∼5.0 nM) in a total volume of 500 μl. Likewise, for D₂ binding, nonspecific binding was defined with 5 μM (+)-butaclamol, with all reaction conditions containing a near-saturating amount of [³H]spiperone (50 μl; final concentration, ∼1.5 nM) in a total volume of 500 μl. The reaction was terminated by filtration onto FB glass fiber plates with ice-cold wash buffer (10 mM Tris and 0.9% NaCl) using a cell harvester (FilterMate; PerkinElmer Life and Analytical Sciences). Radioactivity was determined a TopCount scintillation counter (PerkinElmer Life and Analytical Sciences). Specific binding was determined as the difference between the average of the nonspecific and total binding conditions. The specific binding amount was normalized to the amount of protein using the bicinchoninic acid protein assay (Pierce Chemical, Rockford, IL) following the supplier's protocol. Under the transfection conditions used to explore the effects of drug treatments on BiFC, the following estimated K_d and B_max values were obtained via radioligand saturation binding experiments: [³H]SR141716A, K_d = 0.74 ± 0.18 nM and B_max = 204 ± 28 fmol/mg; and [³H]spiperone, K_d = 0.051 ± 0.02 nM and B_max = 3550 ± 200 fmol/mg.

Fluorescence Energy Transfer.

CAD cells were grown to 70% confluence in 12-well plates before transfection. Cells were transiently transfected with three general conditions depending on the receptor dimer species to be studied including: cells only expressing the FRET donor (Cerulean), cells only expressing the FRET acceptor (Venus), and cells expressing both the donor and acceptor. To normalize for protein expression in cells only expressing either the donor or acceptor, the total amount of DNA transfected was normalized with the untagged receptor. In each FRET assay, 750 ng/well of the donor (CB₁-Cerulean or D_2L-Cerulean) and 750 ng/well of the acceptor (CB₁-Venus or D_2L-Venus) were transiently transfected either alone or in combination 24 h before the experiment. Cells were washed with 500 μl of warm PBS and resuspended in 300 μl of warm PBS. Protein concentration was determined on the cell suspension using the bicinchoninic acid assay method (Pierce Chemical) and normalized to 200 ng/μl with PBS. CAD cells suspensions (40 μg) were transferred into a 96-well black plate (Nalge Nunc International), and fluorescence measurements were evaluated on the FUSION plate reader (PerkinElmer Life and Analytical Sciences). Determination of FRET signals was performed as described previously (Vidi et al., 2008b). In brief, FRET signals were measured using the sensitized acceptor method. Mock-transfected cells were used for background fluorescence. For each sample, Cerulean (C) and Venus (V) was measured using 430/25 nm and 500/20 nm excitation and 470/30 nm and 535/30 nm emission filters. FRET signals (F) were measured using excitation at 430/25 nm and emission at 535/30 nm. Bleed-through coefficients were calculated for the acceptor (a = F/V) and for the donor (d = F/C) in cells only expressing either Cerulean (donor) or Venus (acceptor) fusion proteins. The FRET signals were corrected (cFRET) for acceptor and donor bleed-through using the equation cFRET = F − aV − dC. The signals were then normalized to donor (C) and acceptor (Y) intensities as follows: nFRET = cFRET/√C × V.

Data and Statistical Analysis.

Data and statistical analyses were performed using Prism (GraphPad Software Inc., San Diego, CA). A p value <0.05 was considered significant.

Results

Functional cAMP accumulation assays were performed to verify the signaling properties of the BiFC-tagged CB₁ and D_2L receptors (Fig. 1). Because CB₁ and D_2L receptors couple to inhibitory G proteins (i.e., Gα_i/o), agonist-induced inhibition of forskolin-stimulated cAMP accumulation was used to evaluate receptor function. The BiFC-tagged D_2L receptors D_2L-CN and D_2L-CC were functional after stimulation with the D₂ agonist quinpirole (10 μM), revealing approximately 60% inhibition of forskolin-stimulated cAMP accumulation (Fig. 1A). Additional experiments confirming the functionality of the BiFC-tagged CB₁ receptors, CB₁-VN and CB₁-CC, were performed. Both constructs were functional after stimulation with the CB₁ receptor agonist CP55,940 (10 μM), yielding more than 35% inhibition of forskolin-stimulated cAMP accumulation (Fig. 1B). A modest but insignificant degree (approximately 10%) of inhibition was also observed in vector-transfected CAD cells. Receptor signaling in cells coexpressing the wild-type or BiFC-tagged receptors was also examined (Fig. 1, A and B). The D₂ agonist quinpirole robustly inhibited forskolin-stimulated cAMP accumulation in cells coexpressing CB₁ and D_2L receptors (CB₁ + D_2L or CB₁-VN + D_2L-CC) as well as cells cotransfected with D_2L-CN and D_2L-CC. Similar experiments revealed that CP55,940, inhibited forskolin-stimulated cAMP accumulation in cells coexpressing CB₁ and D_2L receptors (CB₁ + D_2L and CB₁-VN + D_2L-CC). These data suggest that the addition of a C-terminal tag (-VN, -CN, or -CC) and fluorescence complementation (see below) do not adversely affect agonist-mediated inhibition of cAMP accumulation.

Fig. 1.

Functional characterization of receptor-BiFC fragment fusion proteins by measurement of acute inhibition of forskolin-stimulated cAMP accumulation. CAD cells were transiently transfected as indicated. Cyclic AMP accumulation was measured after a 15-min incubation with forskolin (10 μM) in the presence of quinpirole (10 μM) (A) or CP55,940 (10 μM; B) as shown. All data are normalized to the percentage of forskolin-stimulated cAMP accumulation under matched transfection conditions. Each bar represents the mean ± S.E.M. of three to four independent experiments assayed in duplicate. *, p < 0.05 compared with forskolin-stimulated cAMP accumulation under vehicle conditions (one-sample t test).

MBiFC is novel technique that allows for the simultaneous study of two receptor dimer species within living cells (Fig. 2A; Vidi and Watts, 2009). Initial single color BiFC experiments used the fusion receptors to confirm interactions between CB₁ and D_2L receptors. Coexpression of either combination of BiFC constructs (CB₁-VN + D_2L-CC or D_2L-VN + CB₁-CC) in CAD cells produced a robust Venus signal (Supplemental Fig. 2A). Additional BiFC studies compared the CB₁-D_2L fluorescent signal with CB₁ or D_2L receptors in combination with the M₄ muscarinic receptor BiFC constructs (i.e., CB₁-VN + M₄-CC or D_2L-CC + M₄-VN). The CB₁-D_2L heterodimer displayed an enhanced fluorescent signal compared with the M₄-containing heterodimers (Supplemental Fig. 2B). The formation of CB₁-D_2L heterodimers supports previous studies demonstrating interactions between CB₁ and D_2L receptors (Kearn et al., 2005; Marcellino et al., 2008).

Fig. 2.

CB₁-D_2L and D_2L-D_2L dimers detected by MBiFC. A, schematic representing the MBiFC approach used in these studies. CB₁-D_2L dimerization reconstitutes the Venus fluorescent protein (yellow) and D_2L-D_2L dimerization reconstitutes the Cerulean fluorescent protein (cyan). B, representative images of the fluorescent signals observed in an MBiFC study as described in the schematic in A. CAD cells were transiently transfected and imaged as described under Materials and Methods. The merged image (overlapping signal in yellow) represents an overlap of the Venus signal (depicted in red) and the Cerulean signal (depicted in green). Scale bar, 5 μm. C, representative images of the expression patterns of CB₁-Venus and D_2L-Cerulean receptors after cotransfection. The merged image (overlapping signal in yellow) represents an overlap of CB₁-Venus (depicted in red) and D_2L-Cerulean (depicted in green). D, representative images of the expression patterns of CB₁-Venus (top) and D_2L-Cerulean (bottom) after individual transfections in the presence of mCherry-mem. The merge image (overlapping signal in yellow) represents the overlap of either CB₁-Venus or D_2L-Venus (green signal) with mCherry-mem (red signal).

One goal of the present study was to assess the dynamic nature of the CB₁-D_2L heterodimer in response to persistent drug treatment. This required the establishment of MBiFC as described previously for A_2A adenosine and D_2L dopamine receptors (Vidi et al., 2008a). Using this approach, CAD cells were transiently transfected with CB₁-VN, D_2L-CC, and D_2L-CN to simultaneously visualize CB₁-D_2L and D_2L-D_2L receptor dimers using fluorescence microscopy (Fig. 2A). The presence of a Venus signal is indicative of the CB₁-D_2L heterodimer, whereas a Cerulean signal corresponds to the D_2L-D_2L homodimer (Fig. 2B). CAD cells transfected with CB₁-VN, D_2L-CC, and D_2L-CN expressed both Venus and Cerulean signals consistent with the coexistence of CB₁-D_2L heterodimers and D_2L-D_2L homodimers (Fig. 2B). Fluorescent signals corresponding to the receptor dimers showed a similar pattern of distribution and were found at the plasma membrane as well as intracellularly. For comparison with the BiFC signals, the localization patterns of CB₁-Venus and D_2L-Cerulean were evaluated after coexpression (Fig. 2C). The CB₁-Venus signal showed significant intracellular localization, whereas the D_2L-Cerulean displayed localization at both the plasma membrane and intracellular compartments. Moderate overlap between the CB₁ and D_2L signals was also observed. For additional comparison, the individual expression patterns of CB₁-Venus and D_2L-Venus were examined (Fig. 2D). When expressed alone, the CB₁-Venus signal was primarily localized intracellularly demonstrated by the lack of overlap with the membrane marker (merge panel). Conversely, the D_2L-Venus expression was found primarily at the membrane and extensive overlap with the membrane marker was displayed (Fig. 2D).

We also attempted to perform MBiFC experiments to simultaneously examine D_2L-CB₁ and CB₁-CB₁ dimers. Unfortunately, the fluorescent signal of the CB₁-CB₁ dimer under MBiFC conditions was too low to reliably measure, restricting our MBiFC experiments to CB₁-D_2L and D_2L-D_2L receptor dimers. The lack of a CB₁-CB₁ dimer BiFC signal may reflect one of the disadvantages of BiFC. Specifically, the intensity of the fluorescence complementation signal is considerably weaker (2.5–5.5-fold) than the signal from the corresponding full-length fluorescent protein under similar transfection conditions (Vidi and Watts, 2009).

One advantage of BiFC is the ability to investigate the localization of the receptor dimers using epifluorescence. With the use of fluorescently tagged intracellular makers, the patterns of intracellular expression of the CB₁-D_2L and D_2L-D_2L receptor dimers were investigated using fluorescent microscopy (Fig. 3). CAD cells were transiently transfected with BiFC constructs that reconstitute Cerulean to either express the CB₁-D_2L heterodimer (CB₁-CN + D_2L-CC) or the D_2L-D_2L homodimer (D_2L-CN + D_2L-CC). In addition, these cells were transfected with the indicated YFP-tagged intracellular marker proteins (YFP-Endo, YFP-ER, or YFP-Golgi; Fig. 3). The endosome marker (YFP-Endo) is a fusion protein with RhoB, a known endosomal protein fused to YFP. The ER marker (YFP-ER) consists of YFP fused to the ER targeting sequence of calreticulin and the KEDL ER retrieval sequence. The Golgi marker (YFP-Golgi) is a YFP fusion protein with residues 1 to 81 of the β1,4-galactosyltransferase protein. Overall, both receptor dimers, D_2L-D_2L and CB₁-D_2L displayed moderate to extensive overlap with endosome and ER structures (Fig. 3, A and B). However, CB₁-D_2L and D_2L-D_2L receptor dimers demonstrated minimal to no overlap with the Golgi apparatus. These expression patterns are consistent with receptor dimer assembly at the ER (Herrick-Davis et al., 1997) and proper trafficking into endosomes (Leterrier et al., 2004). However, the additional possibility that receptors dimerize at the plasma membrane cannot be excluded in the absence of additional studies.

Fig. 3.

Intracellular localization patterns of the D_2L-D_2L homomers and CB₁-D_2L heteromers. CAD cells were transiently transfected with both D_2L-CC and D_2L-CN (cyan signal in A) or CB₁-VN and D_2L-CC (cyan signal in B) along with the indicated YFP fluorescent marker proteins (yellow signal). The merged image (overlapping signal in yellow) represents an overlap of the BiFC signal (depicted in red) and the fluorescent marker signal (depicted in green). Images are representative of three independent transfections.

The results demonstrating MBiFC in neuronal cells were further validated by examining dimerization of these receptors using FRET, which has been used previously to investigate interactions of CB₁ and D_2L receptors (Marcellino et al., 2008). CAD cells were transiently transfected with either CB₁-Venus + CB₁-Cerulean, CB₁-Venus + D_2L-Cerulean, or D_2L-Venus + D_2L-Cerulean (Fig. 4). A significant FRET signal was detected with all three receptor pairs compared with the mix control sample in which suspensions of cells only expressing the donor or acceptor was mixed in the FRET sample plate. These results provide further confirmation of our BiFC studies, supporting the hypothesis that CB₁ and D₂ form both homo- and heteromeric receptor oligomers in a neuronal-like cell model.

Fig. 4.

CB₁ and D_2L receptor form homo- and heteromeric receptor oligomers as measured by FRET. CAD cells were transiently transfected with 750 ng/well of either CB₁-Venus and CB₁-Cerulean, D_2L-Venus and D_2L-Cerulean, or CB₁-Venus and D_2L-Cerulean. Mix samples represent a mixture of CAD cell suspensions individually expressing the respective fluorescently tagged receptors of interest. Data represent the mean ± S.E.M. of three independent experiments assayed in triplicate. **, p < 0.01 compared with mixed samples (one-way analysis of variance followed by Dunnett's post hoc test).

Using MBiFC and FRET techniques, we have provided evidence that CB₁ and D₂ receptors participate in receptor dimer complexes. We next sought to investigate the effects of persistent ligand treatment on the formation of CB₁ and D_2L heterodimer and D_2L homodimers using MBiFC as a tool to monitor changes in relative receptor dimer population. CAD cells were transiently transfected with CB₁-VN, D_2L-CC, and D_2L-CN, and the presence of the CB₁-D_2L heterodimer (Venus) and D_2L-D_2L homodimer (Cerulean) was simultaneously measured. The fluorescent intensity ratio of Venus to Cerulean in both the plasma membrane and intracellular compartments was determined after drug treatment. Under the conditions used, an increase in the Venus-to-Cerulean ratio would be indicative of an increase in the formation of the CB₁-D_2L receptor dimer relative to the D_2L-D_2L receptor dimer compared with vehicle-treated cells.

Our previous work with D₂ and A_2A receptor ligands suggested that a 20-h drug treatment provided a robust BiFC signal in which drug-induced changes in A_2A-D_2L, D_2L-D_2L, and A_2A-A_2A dimers could be observed (Vidi et al., 2008a). In the present study, we completed MBiFC time course experiments with the CB₁ receptor ligand CP55,940 to verify that a similar treatment duration produced robust responses in the absence of a ceiling effect. The results of the time course study revealed that CP55,940 treatments shorter than 10 h (i.e., 5 h) had very low fluorescent signals and did not allow us to quantify an adequate number of cells for analysis (data not shown). However, robust YFP and CFP signals were evident after 10 h and the drug effects were time-dependent showing the greatest response at 30 h (Fig. 5). The time course study also suggested that the 20-h time point is on the dynamic portion of the temporal scale potentially allowing us to observe ratiometric changes in both directions as shown previously (Vidi et al., 2008a). Examination of the overall YFP and CFP intensities at 20 h indicated that the CP55,940-induced increase in the YFP/CFP ratio reflected a combined increase in the YFP signal (CB₁-D_2L) and a decrease in the CFP signal (D_2L-D_2L) compared with vehicle-treated cells. Specifically, the membrane showed an 11% increase in YFP and a 27% decrease in CFP intensity. Intracellularly, there was 33% increase in the YFP signal and a 15% decrease in the CFP signal (n = 4).

Fig. 5.

Time course examining the effects of persistent CP55,940 treatment on heteromer (D_2L-CB₁-Venus) and homomer (D_2L-D_2L-Cerulean) formation. CAD cells were transiently transfected with CB₁-VN, D_2L-CC, and D_2L-CN followed by quantitative image analyses of the Venus/Cerulean ratios for the membrane and intracellular compartments. A, cells were incubated with 10 μM CP55,940 for 10, 20, or 30 h before image analysis. Data represents the average median Venus-to-Cerulean ratio values normalized to percentage of vehicle treatment (±S.E.M.) in four independent experiments. B, images from 20-h time point depicting the effects of CP55,940 on Venus and Cerulean signals.

Drug-induced changes in the relative receptor dimer population were measured after treatment (20 h) with either D₂ (Fig. 6A) or CB₁ (Fig. 6B) receptor ligands. Persistent activation of the D_2L receptor with quinpirole (10 μM) resulted in a significant increase in the Venus-to-Cerulean ratio consistent with an increase in CB₁-D_2L heterodimers relative to D_2L-D_2L homodimers. However, this effect was only significant in the intracellular compartment. The effect of quinpirole was prevented by coapplication of the selective D₂ receptor antagonist sulpiride (1 μM). Treatment with sulpiride alone or in combination with quinpirole resulted in a significant decrease in the Venus to Cerulean ratio in both the membrane and intracellular compartments. Because the observed alterations in receptor dimer population may involve changes in receptor expression, single point radioligand binding experiments were used to estimate relative receptor densities after drug treatment. The results of these studies revealed that persistent treatment with quinpirole (10 μM), sulpiride (1 μM), or quinpirole + sulpiride significantly increased D_2L receptor density (118 ± 6, 149 ± 17, or 129 ± 9%; n = 5) compared with vehicle treatment (100%). These ligand-induced increases in D_2L receptor expression are consistent with our previous report (Vidi et al., 2008a) and work from others (Sibley and Neve, 1997). No significant changes in CB₁ receptor density were observed upon persistent treatment with either of the D₂ receptor ligands alone or the combination (data not shown).

Fig. 6.

Effects of persistent ligand treatment on heteromer (D_2L-CB₁-Venus) and homomer (D_2L-D_2L-Cerulean) formation. CAD cells were transiently transfected and imaged as described for Fig. 5. A, cells were incubated with 10 μM quinpirole (Quin), 1 μM sulpiride (Sulp), or quinpirole + sulpiride (Quin + Sulp) for 20 h. B, cells were incubated with 10 μM CP55,940 (CP), 10 μM AM281, or CP55,940 + AM281 (CP + AM281) for 20 h. Data represent the average median Venus-to-Cerulean ratio values normalized to percentage of vehicle treatment (± S.E.M.). *, p < 0.05 (compared with vehicle, one-sample t test, n = 5–8).

MBiFC experiments were also performed using CB₁ ligands. Persistent treatment with the CB₁ receptor agonist CP55,940 (10 μM) led to a significant increase in the Venus-to-Cerulean ratio in both the plasma membrane and the intracellular regions compared with vehicle-treated cells (Fig. 6B). The addition of the CB₁ receptor antagonist AM281 (10 μM) attenuated the CP55,940-induced increase in the Venus-to-Cerulean ratio. Dose-response experiments revealed that the average EC₅₀ values for CP55,940 increasing the YFP/CFP ratio were 320 and 210 nM for membrane and intracellular signals, respectively (Fig. 7). Subsequent single point radioligand binding experiments revealed that 20-h treatment with CP55,940 had no effect on CB₁ receptor density (106 ± 12%; n = 5); however, a modest decrease in D_2L receptor density (82 ± 3%; n = 5) was observed.

Fig. 7.

Dose-response analysis for CP55,940 modulation of the Venus/Cerulean ratio. CAD cells were transiently transfected and imaged as described for Fig. 5. Cells were incubated with increasing concentrations of CP55,940 for 20 h. Data represent the average median Venus-to-Cerulean ratio values normalized to percentage of vehicle treatment (±S.E.M.) in three independent experiments.

The observations described above suggest that persistent activation of the CB₁ receptor favors the formation of CB₁-D_2L heterodimers without alterations in CB₁ receptor expression. To investigate further the role of persistent activation on receptor dimerization, a constitutively active CB₁ receptor mutant was constructed for use in the MBiFC experiments. Threonine 210 of the CB₁ receptor was mutated to an isoleucine (CB₁T210I) to create a constitutively active receptor (D'Antona et al., 2006). The presence of an isoleucine at amino acid 210 disrupts the salt bridge in the DRY motif mimicking receptor activation, leading to enhanced agonist affinity and increased intracellular localization (D'Antona et al., 2006). We examined and compared the relative receptor heterodimer (CB₁-D_2L) and homodimer (D_2L-D_2L) populations in cells expressing either the wild-type (CB₁wt) or the constitutively active CB₁ receptor (CB₁T210I) using MBiFC (Fig. 8). The Venus- (CB₁-D_2L) to-Cerulean (D_2L-D_2L) ratios at the plasma membrane were similar in cells expressing the wild-type or constitutively active CB₁ (Fig. 8A). In contrast, expression of CB₁T210I resulted in a significant increase in the intracellular Venus-to-Cerulean ratio compared with the wild type CB₁ (Fig. 8A). The intracellular-to-membrane ratio of the Venus signal (i.e., CB₁wt-D_2L or T210I-D_2L dimer) in cells expressing the CB₁T210I mutant was also significantly increased (approximately 150%) compared with cells expressing CB₁wt (Fig. 8, A and B). The overlapping expression patterns of CB₁-D_2L and D_2L-D_2L dimers were markedly reduced in cells coexpressing CB₁T210I as indicated by a loss of white signal on the membrane in the merged images. Subsequent localization studies with the CB₁T210I-D_2L heterodimer revealed significant signal overlap with the endosomes and limited overlap in the ER consistent with enhanced endocytosis of the CB₁T210I mutant (D'Antona et al., 2006; Supplemental Fig. 3).

Fig. 8.

Effect of constitutively active CB₁ receptor (T210I) on relative dimer population at the plasma membrane or intracellular compartment. CAD cells were transiently transfected with either CB₁wt-VN or CB₁T210I-VN with D_2L-CC and D_2L-CN. A, left, quantitative image analyses of the Venus/Cerulean ratios were determined for the membrane and intracellular compartments as described under Materials and Methods. The Venus-to-Cerulean ratio induced by CB₁T210I was normalized to the Venus-to-Cerulean ratio measured in cells expressing CB₁wt receptor; *, p < 0.05 (compared with wild type, one-sample t test). Right, quantitative image analyses of the intracellular/membrane ratios were determined for the CB₁-D_2L-Venus signal in cells either expressing CB₁wt or CB₁T210I. *, p < 0.05 (compared with wild type, t test). Data for both analyses were generated from the same experiments and represent the average median ± S.E.M. from three independent experiments. B, representative images of CAD cells expressing either CB₁wt (top) or CB₁T210I (bottom) to reconstitute the BiFC signals CB₁(wt or T210I)-D_2L-Venus and D_2L-D_2L-Cerulean and the membrane marker (mCherry-mem). The merge panel represents overlap of the three channels and overlapping pixel intensity is presented in white.

Discussion

Evidence for the existence and functional significance of CB₁ and D_2L heterodimers has continued to evolve over the past 10 to 15 years. However, investigations examining the regulation of these heterodimers and their homodimer counterparts are just beginning as new technological advances for studying protein-protein interactions are developed (Vidi and Watts, 2009). In the present study, we have applied MBiFC as a novel technique to study the dimerization of CB₁ and D_2L receptors, and we reveal for the first time the localization patterns of these receptor heterodimers in a neuronal cell model.

Early studies of CB₁ and D_2L function were central to the development of the concept of CB₁-D_2L heterodimer (for review, see Glass et al., 1997). Several studies suggest that the CB₁-D_2L dimer possesses stimulatory properties toward adenylyl cyclase via the CB₁ receptor engaged in the heterodimer (Glass and Felder, 1997; Jarrahian et al., 2004; Kearn et al., 2005). However, conflicting conclusions from studies examining the regulation of CB₁ and D_2L receptor dimerization remain. One potential mechanism for regulating the CB₁-D₂ dimer is based on observations that the physical association of CB₁ and D_2L increases in the presence of acute coactivation of both receptors (Kearn et al., 2005). Activation of either CB₁ or D_2L receptor individually did not significantly increase the physical association, suggesting that coactivation of both receptors is necessary for enhanced receptor dimerization. It was also reported that expression (and not activation) of the D_2L receptor was sufficient to induce a switch in CB₁-G protein coupling to a stimulatory pathway, however, measurements of the CB₁-D_2L receptor dimer were not performed (Jarrahian et al., 2004). In addition, another study reported a lack of agonist-mediated increase in the FRET interaction between CB₁ and D₂ receptors under conditions of both single and concurrent receptor activation (Marcellino et al., 2008). The lack of consistency between the reports described above may reflect differences in the choice of receptor ligands, the model systems, technical approaches, or the complex pharmacology of the CB₁-D₂ dimer.

In the present study, we used MBiFC to show that persistent activation of either the CB₁ or D_2L receptor leads to the formation of more CB₁-D_2L heterodimers relative to the D_2L-D_2L homodimers. There are several differences between our study of CB₁-D_2L interactions and the previous work described above (e.g., cell type, methods to measure receptors, drug treatment); however, the drug treatment conditions and technology used to assess the receptor dimers probably have significant influence. Each of the drug treatments reported here represents an extended drug exposure (i.e., 10–30 h). Drugs were added 4 h after transfection and were present during the time of ongoing receptor biosynthesis and subsequent oligomerization. Therefore, the dimers observed in our studies probably involve mechanisms not reflected in shorter drug treatments or acute studies (Kearn et al., 2005; Marcellino et al., 2008). The present study used BiFC technology, which differs from FRET in that the complementation of fluorescent signal is essentially irreversible (Vidi and Watts, 2009). This property of MBiFC allows investigators to “capture” and subsequently measure drug-induced changes in receptor dimers over an extended period in which a sufficient signal can be collected.

Persistent D₂ agonist treatment with quinpirole favored the formation of CB₁-D_2L heterodimers versus D_2L-D_2L homodimers. This effect was accompanied by an increase in D_2L receptor expression and was prevented by the D₂ antagonist sulpiride. The increase in D_2L receptor expression may suggest a pharmacological chaperone effect on receptor dimer formation where D₂ ligands stabilize the receptor, somehow promoting CB₁-D_2L receptor interactions (Vidi et al., 2008b). However, treatment with the D₂ antagonist sulpiride also increased D_2L receptor density, but instead favored the formation of D_2L-D_2L homodimers. These opposing effects of D₂ agonists and antagonists on D_2L-D_2L versus CB₁-D_2L dimer formation argues against a simple role of increased D_2L receptor expression. One explanation for these differential effects may involve ligand-specific changes in receptor dimerization patterns (Vidi et al., 2008b). In addition to ligands, these dimerization patterns also appear to be influenced by the receptors under investigation. In a previous study of D_2L and A_2A receptor dimerization, quinpirole increased D_2L-D_2L homodimers relative to A_2A-D_2L heterodimers (Vidi et al., 2008b). The potential scenario gets increasingly complicated when considering a recent BiFC-BRET study providing evidence for a CB₁-D₂-A_2A receptor oligomer (Navarro et al., 2008). Linking the observations described above and the present results suggests a scenario where striatal neurons expressing D_2L, A_2A, and CB₁ receptors would be subject to a very complicated receptor regulation scheme. For example, persistent D₂ agonist treatment would increase overall D_2L receptor expression levels and perhaps promote the following pattern of receptor oligomers D_2L-CB₁ > D_2L-D_2L > A_2A-D_2L. The potential physiological and functional significance of these ligand-induced changes in heterodimers are intriguing and await biochemical and behavioral analysis (Marcellino et al., 2008). In addition to in vivo studies, new molecular tools to study these complex systems are becoming increasingly available as methods to study interactions of higher ordered GPCR oligomers (e.g., trimers and tetramers) such as BiLC-FRET, BiFC-FRET, and BiFC-BRET are developed (Vidi and Watts, 2009).

The ability of quinpirole to alter the formation of receptor oligomers involving D_2L receptors may provide insight into the disease states associated with persistent D₂ receptor activation, as in the treatment of Parkinson's disease with l-DOPA and D₂ dopamine receptor agonists (Hurley and Jenner, 2006). For example, persistent quinpirole treatment increases A_2A-A_2A homodimer formation and A_2A signaling (Vortherms and Watts, 2004; Vidi et al., 2008a). These observations may provide a molecular explanation for the beneficial clinical effects of A_2A antagonists in treating l-DOPA-induced dyskinesias (Morelli et al., 2007; Fuxe et al., 2008). The current results suggest that persistent treatment with D₂ receptor agonist drugs may promote the formation of CB₁-D_2L heterodimers. The increase in CB₁-D_2L dimer formation may allow the CB₁ receptor to have enhanced antagonistic effects over the D₂ receptor signaling (Marcellino et al., 2008). This scenario would provide for increased CB₁ signaling after a dopamine receptor-dependent increase in endocannabinoid release (Giuffrida et al., 1999; Piomelli, 2003). In addition, evidence linking the CB₁-D_2L heterodimer to a stimulatory pathway (Glass and Felder, 1997; Kearn et al., 2005) may provide a mechanism for CB₁ antagonism of D₂ signaling at the intracellular level (i.e., cAMP). Further in vivo investigations of CB₁ receptor and CB₁-D_2L heterodimer signaling after persistent D₂ receptor activation are warranted; however, studies suggest that the CB₁ receptor antagonists/inverse agonists may have beneficial effects in the management of Parkinson's disease. For example, the CB₁ antagonist 1-[7-(2-chlorophenyl)-8-(4-chlorophenyl)-2- methylpyrazolo[1,5-a]-[1,3,5]triazin-4-yl]-3-ethylaminoazetidine-3-carboxylic acid amide benzenesulfonate dose-dependently enhances the anti-Parkinson's activity of l-DOPA (Cao et al., 2007). Another study revealed that rimonabant, a CB₁ receptor inverse agonist, had beneficial effects in managing l-DOPA-induced dyskinesias (van der Stelt et al., 2005).

Similar to the D_2L receptors, the precise mechanism by which persistent activation of the CB₁ receptor favors the formation of CB₁-D_2L heterodimers relative to D_2L-D_2L homodimers remains largely unknown. Our observations suggest that the formation of the heterodimer is mediated by receptor activation and not alterations in CB₁ receptor expression. It is possible that the activated conformational state of the CB₁ receptor possesses enhanced affinity for the D_2L receptor and that persistent activation promotes CB₁-D_2L heterodimerization. This hypothesis is supported by the report that the CB₁ receptor increases the association with the D_2L receptor in a dose-dependent manner (Kearn et al., 2005). Furthermore, the present study demonstrated that expression of a constitutively active CB₁ mutant, CB₁T210I, promoted more CB₁-D_2L heterodimerization. Although the identification of a molecular mechanism awaits further study, it is tempting to consider that CB₁-D_2L interactions will represent a new CB₁ receptor signaling pathway that may be subject to functional selectivity (Glass and Northup, 1999; Mukhopadhyay and Howlett, 2005; Urban et al., 2007).

The physiological significance and functional consequences of CB₁ receptor-induced CB₁-D_2L dimers may have implications in the use of clinical cannabinoids to treat chronic pain as well as chronic marijuana use. Such conditions would involve persistent CB₁ receptor activation, providing an impetus to understand the molecular adaptations that occur in the nervous system (Cooper and Haney, 2008). Although we were able to study drug-induced changes of the CB₁-D_2L and D_2L-D_2L receptor dimers, a low BiFC signal between CB₁ receptors prevented us from examining the ratios of CB₁-CB₁ homodimers to CB₁-D_2L heterodimers. In the absence of CB₁-CB₁ studies, the CP55,940-induced increase in the CB₁-D_2L heterodimer may reflect a relative decrease in D_2L-D_2L homodimers and perhaps D_2L function. Consistent with this possibility we observed a modest CP55,940-induced decrease (approximately 15–25%) in D_2L receptor expression and D_2L-D_2L homodimers. These observations may suggest that persistent CB₁ receptor activation and subsequent CB₁-D_2L heterodimer formation could reduce D_2L receptor expression. In partial support of this hypothesis, it has been shown in rats and humans that chronic prenatal exposure to marijuana decreases the expression of dopamine D₂ receptors in the brain (Walters and Carr, 1986; Wang et al., 2004).

In the present report, we have visualized simultaneously the localization patterns of CB₁-D_2L heterodimers and D_2L-D_2L homodimers in living cells and provided evidence for agonist-regulated effects on receptor dimerization patterns. Recent studies propose that an increasing number of GPCRs may participate in higher order receptor oligomers or “receptor mosaics” and that these structures may mediate many signaling events (for review, see Fuxe et al., 2008). The present work and other recent studies are consistent with this concept (Carriba et al., 2008; Navarro et al., 2008). We anticipate the continued development of new technologies will allow investigators to examine these receptor mosaics in greater detail. Finally, the use of MBiFC provides a new tool to study drug-induced changes in receptor oligomerization and may offer an important asset relevant to the future of drug discovery in the area of receptor heterodimers.