|Home | About | Journals | Submit | Contact Us | Français|
Drugs of abuse such as morphine or marijuana exert their effects through the activation of G-protein-coupled receptors (GPCRs), the opioid and cannabinoid receptors, respectively. Moreover, interactions between either of these receptors have been shown to be involved in the rewarding effects of drugs of abuse. Recent advances in the field, using a variety of approaches, have demonstrated that many GPCRs, including opioid, cannabinoid, and dopamine receptors, can form associations between different receptor subtypes or with other GPCRs to form heteromeric complexes. The formation of these complexes, in turn, leads to the modulation of the properties of individual protomers. The development of tools that can selectively disrupt GPCR heteromers as well as monoclonal antibodies that can selectively block signaling by specific heteromer pairs has indicated that heteromers involving opioid, cannabinoid, or dopamine receptors may play a role in various disease states. In this review, we describe evidence for opioid, cannabinoid, and dopamine receptor heteromerization and the potential role of GPCR heteromers in pathophysiological conditions.
Morphine, an alkaloid extracted from the poppy Papaver somniferum, and Δ9-tetrahydrocannabinol (also known as THC or more commonly as marijuana), the main psychoactive extract from Cannabis sativum, are commonly abused drugs due to their mood-altering effects. Morphine is the most commonly used drug in the treatment of chronic pain, while THC is used in the alleviation of neuropathic pain and for the reduction of eye pressure in glaucoma patients.1–5 However, long-term use of these drugs leads to a number of undesirable side effects, including the development of tolerance and addiction. In the latter context, several lines of evidence demonstrate the involvement of the dopaminergic system in the rewarding effects of drugs of abuse (reviewed in Ref. 6). Both morphine and THC exert their physiological effects via activation of Gαi/o-coupled receptors, the opioid and cannabinoid receptors, respectively.1,3–5 In an effort to develop drugs that target these receptors and exert the same physiological effects as morphine or THC but without the undesirable side effects associated with chronic drug use, several groups have investigated whether protein–protein interactions, including receptor heteromerization, could modulate the function of individual receptors. Here, we describe heteromerization as associations between different subtypes of a receptor or between two different receptors. Although heteromers involving opioid, cannabinoid, or dopamine receptors have been described, not much is known about their physiological role. In this context, studies using peptides that disrupt specific heteromers, or antibodies that selectively block heteromer signaling, suggest that GPCR heteromers may play a role in a number of pathologies.7–11 While additional GPCR heteromers associated with other drugs of abuse have been reported, for the sake of brevity and conciseness, in this review we have focused only on heteromers that contain either opioid or cannabinoid receptors as one of the partners. We have also included dopamine receptors since they are the primary effectors of the action of stimulants such as cocaine and amphetamines. These are described below along with the putative role of these heteromers in the development of tolerance and other disease states.
Opioid receptors are members of the rhodopsin family of GPCRs and their signal transduction involves activation of Gαi/o proteins, downstream inhibition of adenylyl cyclase activity, the opening of inwardly rectifying K+ channels, and inhibition of Ca2+ channels.1 Molecular cloning studies have identified three types of opioid receptors μ-, δ-, and κ-opioid receptors (reviewed in Ref. 2).
Opioid receptors can be activated by endogenous peptides derived from the processing of the large precursor proteins, proopiomelanocortin, proenkephalin, and prodynorphin, by prohormone convertases via the classic peptide-processing pathway12 to generate mainly β-endorphin, Met- and Leu-enkephalin, and dynorphins.13 It is generally thought that enkephalins exert their biological effects primarily by activating δ opioid receptors (DOR), β-endorphin by activating μ opioid receptors (MOR), and dynorphins by activating κ opioid receptors (KOR).13,14 In addition, opioid receptors can also be activated by drugs of abuse such as morphine and heroin.15 Agonist-mediated activation of opioid receptors leads to the modulation of many biological functions including pain perception, motivation, locomotion, hormone secretion, and reward.1,2 Studies using genetic knockdown have further elucidated the role of each opioid receptor type. These studies have implicated MOR in the analgesic effects of morphine and in the rewarding effects resulting from other nonopioid drugs of abuse including cannabinoids,16–20 DOR in the regulation of emotional responses including anxiety and depression,21 and KOR in the dysphoric effects of opioids and cannabinoids as well as in stress-induced relapse.22,23 Together, these studies reveal a broad role for opioid receptors in normal physiology and disease.
Like opioid receptors, cannabinoid receptors are also members of the rhodopsin family of GPCRs and signal via the Gαi/o-signaling pathway, to activate similar downstream signaling cascades.3–5 The two main subtypes of cannabinoid receptors are CB1R and CB2R.3 Recently, the orphan receptor, GPR55, has been proposed as a putative cannabinoid receptor subtype (reviewed in Ref. 24).
CB1R and CB2R are activated by endogenous ligands derived from membrane lipids termed “endocannabinoids” which together with the receptors constitute the “endocannabinoid system” (for review, see Refs. 25–27). The two major lipid-based endocannabinoids are anandamide and 2-arachidonoylglycerol (2-AG)3 that are generated by phospholipases28–32 and degraded by fatty acid amide hydrolase,33,34 cyclooxygenases, lipooxygenases,35,36 and monoacylglycerol lipase.37 Recently, endogenous peptidic agonists of cannabinoid receptors have been identified.38–42 These peptides, RVD-hemopressin and VD-hemopressin, are derived from the α-chain of hemoglobin38 and appear to activate a signal transduction pathway distinct from that of classic CB1R ligands.38 However, very little information is available about the enzymatic steps involved in either the generation or degradation of RVD-hemopressin.
Studies show that the endocannabinoid system plays an important role in many physiological processes including pain perception, learning and memory, appetite control, motor coordination, lipogenesis, and modulation of immune responses (for reviews, see Refs. 25–27,43,44). Studies with knockout mice indicate that the CB1R subtype plays a role in the central effects of cannabinoids, including mood disorders, learning and memory, motor coordination, feeding, control of pain, and stress responses, and in the reinforcing and rewarding effects of drugs of abuse (reviewed in Refs. 45,46). Conversely, the CB2R subtype plays a role in the peripheral effects of cannabinoids, including the development and function of the immune system, embryonic development, bone homeostasis, and pain perception, and plays an antifibrogenic role in liver diseases (reviewed in Ref. 47). These studies imply a broad role for CB1R in central nervous system physiology and CB2R in immune system function.
Like opioid and cannabinoid receptors, dopamine receptors are members of the family A of GPCRs. Based on their ability to modulate cAMP levels, these receptors are classified as D1- or D2-like receptors (reviewed in Ref. 48). D1-like receptors couple to Gαs/olf proteins and are thought to be exclusively present at the presynapse (reviewed in Ref. 48). Activation of D1-like receptors leads to increases in cAMP levels, activation of cAMP-dependent protein kinases, increase in the phosphorylation levels of DARPP-32 (a molecule involved in the integration of cellular signaling responses to neurotransmitters) and ERK1/2 (the latter particularly in medium spiny neurons, MSNs), increases in L-type Ca2+ channel and NMDA receptor activity, inhibition of N- or P/Q-type Ca2+ channel and K+ channel activity, enhancement of AMPA currents and modulation of GABA currents (reviewed in Refs. 48–50). D2-like receptors couple to Gαi/o proteins and are present both at the pre- and postsynapse (reviewed in Ref. 48). Activation of D2-like receptors decreases cAMP levels and the activity of protein kinase A, decreases the phosphorylation levels of DARPP-32, inhibits ERK1/2 phosphorylation in striatum, decreases the activity of L-, N-, and P/Q-type Ca2+ channels, and activates K+ channels (reviewed in Refs. 48,49). There is some evidence suggesting that, in addition to Gαs/olf and Gαi/o-mediated signaling, dopamine receptors can also signal via Gαq- and G-protein-independent mechanisms (reviewed in Refs. 48–50). However, the mechanisms underlying Gαq-mediated signaling are not completely understood (reviewed in Refs. 48–50). Studies show that dopamine receptors play a role in the modulation of locomotor activity, olfaction, retinal processes, cardiovascular, renal, and immune functions, memory and learning, feeding, sleep regulation, and in reward behaviors, particularly those related to drugs of abuse (reviewed in Ref. 48). Five subtypes of dopamine receptor have been identified. While D1 and D5 dopamine receptors belong to the D1-like receptor type, D2, D3, and D4 receptors belong to the D2-like receptor type (reviewed in Refs. 48–50).
As the name suggests, the endogenous agonist of dopamine receptors is dopamine. Dopamine can be synthesized from phenylalanine and tyrosine by tyrosine hydroxylase (reviewed in Ref. 50). Following synthesis in the cytosol, dopamine is taken into storage vesicles by vesicular monoamine transporters (reviewed in Ref. 50). Dopamine is released at the synapse by Ca2+-mediated fusion of dopamine-containing vesicles with neuronal membranes (reviewed in Ref. 50). The synaptic activity of released dopamine is terminated by its reuptake into presynaptic nerve terminals by dopamine transporters (reviewed in Ref. 50). In the mammalian brain, four major dopaminergic pathways have been identified. These include the nigrostriatal, mesolimbic, mesocortical, and tuberoinfundibular systems (reviewed in Refs. 48–50). Interestingly, it has been shown that MSNs in the striatum and nucleus accumbens can be phenotypically distinguished depending on the expression of D1 or D2 dopamine receptors. D1 dopamine receptors (D1R) are expressed by MSNs that project into the medial globus pallidus and the substantia nigra pars reticulata, D2 dopamine receptors (D2R) are expressed by MSNs that project into the lateral globus pallidus (reviewed in Refs. 48–50), and a small percentage of MSNs coexpress D1R and D2R (reviewed in Refs. 48–50).
In order to elucidate the physiological roles of different dopamine receptor subtypes, knockout mice have been generated. Data with these mice have been the subject of excellent reviews (see Refs. 50–53) and are not described in detail here. Briefly, studies with D1R knockout mice suggest that the latter exhibit alterations in the motivational or attentive aspects of behavior.52 Interestingly, D1R knockout mice exhibit a decrease in the locomotor effects of cocaine.54 Studies with D2R knockout mice suggest that they may represent an animal model for Parkinson’s disease as they recapitulate the extrapyramidal symptoms of the disease.50 Interestingly, while the release of dopamine elicited by drugs of abuse such as morphine and cocaine is higher in D2R knockout mice than in wild-type animals (reviewed in Ref. 55), the rewarding effects of these drugs are attenuated in the knockout mice (reviewed in Ref. 55). D3R knockout mice exhibit increased activities in novel and exploratory environments, in rearing behavior, and deficits in spatial working memory compared to wild-type mice (reviewed in Ref. 50). These mice also exhibited a decrease in methamphetamine-mediated locomotion56 and behavioral sensitization to ethanol but not to amphetamines.56,57 Moreover, the D3R knockout mice exhibit attenuation in the development of tolerance and precipitated withdrawal to chronic morphine administration.58 Studies with D4R knockout mice show that this receptor subtype may be involved in the modulation of novelty-related exploration and avoidance behavior to unconditioned stimuli (reviewed in Ref. 50). With regard to drugs of abuse, these mice are more responsive to the locomotor effects of ethanol, cocaine, or amphetamines than wild-type mice (reviewed in Ref. 50). In addition, D4R alleles associated with novelty seeking behavior have been correlated with drug abuse, attention deficit hyperactivity disorder, and gambling (reviewed in Ref. 50). Studies show that D5R knockout animals exhibit hypertension suggesting that this dopamine receptor subtype may be involved in the regulation of blood pressure (reviewed in Ref. 50). Taken together, these studies reveal a role for dopamine receptors in normal physiology and in drug abuse.
In the past decade, the idea that GPCRs associate to form heteromeric complexes with novel pharmacology has emerged (for reviews, see Refs. 59–65). Initial evidence for GPCR heteromerization came from studies showing that activation of a GPCR modulated the function of another GPCR.66–70 The possibility that GPCRs could form heteromers was supported by elegant complementation studies. These included studies using chimeras of the α2c-adrenergic and m3 muscarinic receptors (α2c/m3 and m3/α2c) that, when expressed individually, could not bind to either radiolabeled adrenergic or muscarinic ligands but when coexpressed could.71 Similarly, coexpression of two different mutants of the angiotensin AT1 receptor that individually could not bind to angiotensin led to the restoration of angiotensin binding upon coexpression of these mutant receptors.72
The first direct evidence for GPCR heteromerization came from studies with the metabotropic GABAB receptors. The expression of functional GABAB receptors at the cell surface requires the coexpression of both the GABAB1 and GABAB2 protein subunits.73–76 Another example for the requirement for heteromerization for receptor function is provided by taste receptors where associations between T1R1 and T1R2 form a receptor that recognizes sweet tasting molecules, while those between T1R1 and T1R3 recognize “umami” tasting molecules.77–79
The availability of epitope-tagged GPCRs for immunoisolation of interacting complexes to examine whether two GPCRs are in close proximity for direct interaction has shown that a large number of GPCRs can form heteromers (for reviews, see Refs. 59–61). Scientists in the field have applied the following criteria in order to ascertain that two GPCRs form heteromers: (i) colocalization of the two interacting receptors in the same cellular and subcellular compartment; (ii) immunoisolation of interacting receptors, using receptor-selective antibodies, from wild-type animal tissue that expresses both receptors (but not from animals lacking one of the receptors); (iii) proximity-based assays showing that the two receptors are in close proximity to interact only in cells/tissues expressing both receptors and not in cells/tissues lacking individual receptors; and (iv) unique pharmacological properties as determined using highly selective reagents targeting the heteromer.
Early radiolabeled binding studies, as well as studies measuring the antinociceptive effects of selective opioid receptor ligands, suggested the presence of multiple opioid receptor subtypes.150,151 However, to date, the cDNAs for only MOR, DOR, and KOR have been identified.2,20 This suggested the possibility that the multiple opioid receptor subtypes detected by binding and antinociceptive studies could arise from the formation of heteromeric complexes between opioid receptor subtypes (MOR, DOR, KOR) or between the latter and other GPCRs. The availability of cDNAs for epitope-tagged receptors together with coimmunoprecipitation and/or proximity-based assays has enabled studies showing that opioid receptors can form heteromers with a number of GPCRs (see Table 9.1). These are described below.
The first opioid receptor heteromer to be detected and characterized involved DOR and KOR.82 Coimmunoprecipitation studies, using differentially epitope-tagged receptors, showed that DOR and KOR could form interacting complexes.82 BRET studies showed that both receptors were in close enough proximity (<100 Å) in live cells to directly interact.86 Analysis of the binding, signaling, and trafficking properties of DOR–KOR heteromers revealed that the latter exhibited properties distinct from that of the individual receptors. For example, the binding affinities of selective DOR and KOR ligands were reduced in the DOR–KOR heteromer; however, a combination of DOR and KOR ligands exhibited cooperativity in binding that in the case of agonists was translated into a potentiation in signaling responses.82 This binding cooperativity was also observed with a combination of DOR and KOR antagonists.82 Bivalent ligands consisting of DOR and KOR antagonists (naltrindole and 5′-guanidinonaltrindole (5′-GNTI), respectively) linked by spacers of variable length were used to target the DOR–KOR heteromer.152 One of these bivalent ligands, KDN-21, exhibited a higher affinity and selectivity for DOR–KOR heteromers compared to DOR and KOR homomers.153 Pharmacological and binding data indicate that KDN-21 binds to DOR–KOR heteromers of the DOR1–KOR2 phenotype.152,153 Interestingly, a study identified 6′-guanidinonaltrindole (6′-GNTI) as a DOR–KOR heteromer-selective agonist that induced antinociception when administered to the spinal cord but not to the brain.154 However, a recent study has suggested that 6′-GNTI is a KOR-biased ligand155; thus, it remains to be seen if 6′-GNTI represents a DOR–KOR heteromer-biased ligand. Studies that suggest a role for DOR–KOR heteromers in pain pathways are described in Section 3.
This is the most extensively studied opioid receptor heteromer due to its role in morphine-mediated antinociception and in the development of tolerance and dependence to opiates. A comprehensive body of evidence suggested the possibility of DOR–MOR heteromerization (reviewed in Ref. 156). Noteworthy are radiolabeled studies indicating that (i) there is an allosteric modulation of radiolabeled binding to DOR by MOR ligands157,158; (ii) intracerebroventricular potentiation of morphine-mediated antinociception by nonantinociceptive doses of DOR agonists which can be blocked by the DOR antagonists ICI174,864 and naltrindole-5′-isothiocyanate159; (iii) potentiation of acute morphine-mediated antinociception by DOR antagonists and inhibition of development of tolerance or reduction of already developed tolerance to morphine160; (iv) prevention of the development of tolerance and dependence in mice by pretreatment with DOR antagonists161; and (v) lack of development of tolerance to morphine in DOR knockout mice.162
Evidence that MOR and DOR form heteromeric complexes in vitro came from coimmunoprecipitation and BRET studies using differentially epitope-tagged receptors.93–95 In addition, coimmunoprecipitation studies using receptor-selective antibodies revealed the presence of DOR–MOR heteromers in spinal cord membranes from wild-type but not DOR knockout mice.94 More recently, antibodies that selectively recognize DOR–MOR heteromers were used to detect the presence of these heteromers in brain regions of wild-type but not MOR or DOR knockout mice.7
DOR–MOR heteromers exhibit distinct pharmacological properties from that of individual receptors.93–95 Low nonsignaling doses of DOR or MOR ligands potentiate the binding and signaling of MOR or DOR receptors, respectively.93,94,96 Moreover, the potency of highly selective synthetic MOR or DOR ligands is decreased, while the affinity for endomorphin-1 and DOR-selective agonists is increased in DOR–MOR heteromers as compared to MOR or DOR.95,97
Studies show that DOR–MOR heteromerization leads to changes in the signaling pathways activated by the heteromers compared to individual receptor homomers. In this context, a study showed that the DOR–MOR heteromer associates with pertussis-insensitive Gαz instead of the pertussis-sensitive Gαi and this interaction occurs in the endoplasmic reticulum.98,99 Another study reported that activation of a protomer in the DOR–MOR heteromer leads to a switch in signaling from Gαi- (as it occurs in receptor homomers) to β-arrestin-mediated signal transduction leading to changes in the spatiotemporal dynamics of ERK1/2 phosphorylation and ultimately differential activation of transcription factors.100 It has also been reported that while the MOR-selective agonist, DAMGO, stimulates Ca2+ signaling in cells expressing DOR–MOR heteromers by activating phospholipase C and releasing Ca2+ from intracellular stores, it inhibits Ca2+ signaling in GH3 cells expressing only MOR by inhibiting adenylyl cyclase activity and voltage-gated Ca2+ channels and by activating inwardly rectifying K137+. Taken together, these studies indicate that heteromerization increases the repertoire of signaling for opioid receptors.
A few studies have examined how DOR–MOR heteromerization modulates the trafficking of individual receptors. One study reported that treatment with receptor-selective agonists (DAMGO or deltorphin II but not DPDPE or DSLET) led to internalization of the DOR–MOR heteromer.99 However, another study examining receptor turnover in cells that expressed increasing levels of DOR with constant levels of MOR reported that the two receptors endocytosed independently of each other in the context of the heteromer.102 Finally, a recent study reported that methadone could promote the internalization of the DOR–MOR heteromer targeting it to the degradation machinery.103 Together, these data suggest that the internalization of the DOR–MOR heteromer could be a ligand type-dependent event. Recent studies suggest a role for these heteromers in pathways leading to development of tolerance to morphine; this is discussed in Section 3.
Early studies examining interactions between DOR and CB1R showed that (i) long-term exposure to a DOR agonist attenuated CB1R-mediated stimulation of G-protein activity in N18TG2 neuroblastoma cells that coexpress DOR and CB1R163; (ii) the DOR antagonist, naltrindole, blocked Δ9-THC-mediated anxiolytic responses105; (iii) CB1R levels as well as agonist-mediated G-protein activity are increased in the substantia nigra of mice lacking DOR as compared to wild-type animals164; and (iv) DOR activity is increased in the caudate putamen of CB1R knockout mice compared to wild-type mice.165
DOR–CB1R heteromers are detected in HEK-293 and in Neuro 2A cells coexpressing both receptors by coimmunoprecipitation and BRET assays.88,89 DOR and CB1R are colocalized in cortical neurons as examined by immunofluorescence studies with receptor-selective antibodies.88 Interestingly, heteromerization leads to a differential recruitment of trafficking (adaptor) proteins; CB1R associates with the adaptor protein AP-3 in the absence of DOR and with AP-2 in the presence of DOR.88 This, in turn, leads to differential localization of CB1R from intracellular vesicles, in the absence of DOR, to the cell surface in the presence of DOR.88 In addition, lentiviral-mediated knockdown of DOR in F11 cells that endogenously coexpress CB1R and DOR leads to a decrease in cell surface localization of CB1R.88 Heteromerization also alters CB1R signaling; the potency of the CB1R agonist, Hu-210, is decreased in cells expressing DOR–CB1R in assays that measure G-protein activity or ERK1/2 phosphorylation, and in these cells, knockdown of DOR leads to an increase in the potency of Hu-210.88 Moreover, DOR–CB1R heteromerization leads to a PLC-mediated recruitment of β-arrestin 3 to the heteromeric complex, and to the activation of novel signaling pathways that promote cell survival.88 Recent studies suggest a role for these heteromers in pain pathways; this is discussed in Section 3.
Given that both DOR and chemokine CXCR4 receptors are expressed in immune cells, coimmunoprecipitation and FRET assays were used to determine whether both receptors could form heteromers.90 These studies showed that DOR and CXCR4 form interacting complexes.90 Moreover, the studies showed that simultaneous activation of DOR and CXCR4 in cells coexpressing both receptors did not affect cell viability, chemokine binding, or receptor internalization but led to the formation of nonfunctional complexes.90 This suggests that DOR–CXCR4 heteromerization provides a means to silence CXCR4 function and could be important in the regulation of immune function.
Sensory neuron-specific receptors (SNSRs) are mainly expressed in dorsal root ganglia in the spinal cord, where they mediate pronociceptive responses.166 Given that DOR and SNSR-4 share localization in dorsal root ganglia, can be activated by the bovine adrenal medulla peptide 22 (BAM22), and their functions are antagonistic to each other, led to studies examining whether both receptors could form heteromeric complexes. BRET assays in HEK-293 cells coexpressing differential epitope-tagged receptors indicated that both receptors were in sufficiently close proximity to directly interact in live cells.104 In cells expressing DOR–SNSR-4 heteromers, activation of SNSR-4 alone led to Gαq-mediated signaling, while activation of DOR alone resulted in Gαi-mediated signaling. However, coactivation of both receptors by receptor-specific agonists led to a preferential signaling via Gαq.104 Interestingly, the endogenous ligands for SNSR are derived from the proteolytic processing of the proenkephalin precursor that also generates the antinociceptive endogenous DOR ligands, Leu- and Met-enkephalin.167 Moreover, one of these peptides, BAM22, has been shown to bind to both DOR and SNSR-4.167 Activation of DOR–SNSR-4 heteromers with BAM22 leads to signaling via Gαq. This is the first demonstration of an endogenous peptide as a bivalent agonist of a GPCR heteromer.
Interactions between DOR, a Gαi-coupled receptor, and the β2 adrenergic receptor (β2AR), a Gαs-coupled receptor, were suggested by studies showing that in the isolated isovolumic rat heart the DOR agonist, Leu-enkephalin, significantly inhibited β2AR-mediated positive ionotropic effects and stimulation of cAMP levels, effects that were blocked by the opioid antagonist, naloxone, and by pertussis toxin treatment.168 Moreover, in isolated rat ventricular myocytes, nanomolar concentrations of Leu-enkephalin could block β2AR-mediated increases in sarcolemmal L-type Ca2+ currents, cytosolic Ca2+ transients, and contraction.169 Coimmunoprecipitation and BRET assays using differentially epitope-tagged receptors demonstrated the formation of DOR–β2AR heteromers in HEK-293 cells.86,87 While the binding and signaling properties of individual protomers in the DOR–β2AR heteromer were not changed by heteromerization, individual receptor trafficking was, in that DOR agonists could internalize β2AR only in cells coexpressing both receptors.87
A number of studies have suggested interactions between DOR and α2A adrenergic receptors (α2AAR). These include studies showing that (i) a selective imidazole/α2AR agonist synergizes with the DOR agonist, deltorphin II, to block substance P elicited nociception170; (ii) the synergy between agonists for the two receptors was observed in α2CAR and in MOR knockout mice but not in α2AAR knockout mice suggesting an involvement of α2AAR and of DOR in these interactions171–173; and (iii) cotreatment of synaptosomes isolated from the spinal cord or of spinal slices with α2AAR and DOR agonists led to potentiation of inhibition of K+-stimulated release of neuropeptides.84,174
Studies using differentially epitope-tagged receptors expressed in heterologous cells together with coimmunoprecipitation and BRET assays show that α2AAR and DOR are in close proximity (<100 Å apart) to form interacting complexes.85 Not much is known about how DOR–α2AAR heteromerization modulates the pharmacological, signaling, and trafficking properties of this heteromer pair. One study suggested the involvement of protein kinase C in heteromer-mediated synergy in inhibition of calcitonin gene-related peptide release from spinal cord slices,84 while another indicated that the presence of α2AAR is sufficient to potentiate DOR-mediated neurite outgrowth in Neuro 2A cells.85 The latter study indicated that just the presence of α2AAR is sufficient to modulate DOR activity; whether the converse also occurs and how ligand occupancy of both protomers in the DOR–α2AAR heteromer modulates individual receptor function need to be further evaluated.
Interactions between DOR and D1R were suggested by studies showing (i) that alterations in dopaminergic neurotransmission modulate the endogenous levels of opioid peptides175–177; (ii) attenuation of DOR-mediated signaling by the D1R agonist, SKF 82958178; and (iii) enhancement of climbing behavior in mice by a combination of DOR and D1R agonists.179
Immunoelectron microscopy detected the presence of DOR and D1R immunoreactivity in a subset of dendrites in the dorsolateral striatum and nucleus accumbens of rats, indicating that both receptors were present in the same neuron.91,92 Moreover, withdrawal from cocaine, a drug that blocks dopamine transporter activity and thereby alters dopamine neurotransmission by increasing and prolonging signaling by postsynaptic dopamine receptors caused a redistribution of DOR in the nucleus accumbens.91 Although these colocalization studies indicate that DOR–D1R could form heteromers, this needs to be further evaluated by studies using epitope-tagged receptors, coimmunoprecipitation, and proximity-based assays.
Very few studies have examined heteromerization between MOR and KOR. An early study using coimmunoprecipitation of differentially epitope-tagged receptors did not detect the presence of interacting MOR–KOR complexes.82 This does not rule out the formation of MOR–KOR heteromers if the latter are not stable under the detergent conditions used in the coimmunoprecipitation studies. Another study used BRET to show that MOR and KOR are in close proximity to interact in live cells.114 A recent study reported the formation of MOR–KOR heteromers in spinal cord membranes of female, but not male, rats and showed that it is dependent on the stage of the estrous cycle.115 Interestingly, this study identified dynorphin 1–17 as an MOR–KOR heteromer-selective ligand.115
Studies describing synergistic interactions between the antinociceptive effects of MOR and α2AAR agonists180,181 and with α2AAR knockout mice that exhibit a reduction in the analgesic potency of morphine173 suggested the possibility of MOR–α2AAR heteromerization.
Differential epitope-tagged receptors together with coimmunoprecipitation, BRET, or FRET assays demonstrated the formation of MOR–α2AAR heteromers.107,108 Studies examining heteromer-mediated signaling showed that while the presence of α2AAR is sufficient to enhance MOR signaling, coactivation of both receptors in the heteromeric complex leads to a significant attenuation of signaling responses.107 Proximity-based assays (FRET, BRET) showed that, in cells coexpressing MOR–α2AAR heteromers, treatment with morphine induces conformational changes in norepinephrine occupied α2AAR leading to a decrease in Gαi-mediated signaling.108
Finally, a study examining the trafficking of MOR–α2AAR heteromers showed that treatment with the agonist to one receptor protomer did not lead to cointernalization of the heteromeric partner receptor.109 However, given that, in addition to heteromers, the cells express individual receptor homomers, gives rise to the possibility that the agonists are in fact causing internalization of the homomer rather than the targeted receptor in the heteromer. Thus further studies are required to evaluate the trafficking properties of MOR–α2AAR heteromers.
Several studies have described functional interactions between MOR and CB1R (reviewed in Ref. 43). These include studies showing that (i) Δ9-THC enhances the potency of morphine in animal models4; (ii) conditioned place preference to Δ9-THC is not observed in MOR knockout mice17; and (iii) CB1R knockout mice exhibit a reduction in the reinforcing properties of morphine as well as in the severity of withdrawal symptoms to this opioid.182
Coimmunoprecipitation, BRET, and FRET assays using differentially epitope-tagged receptors detected the presence of MOR–CB1R heteromers in heterologous expressing systems.89,110 While the presence of unliganded CB1R does not affect MOR-mediated signaling, ligand binding to CB1R decreases MOR-mediated signaling.89 Conversely, ligand binding to MOR decreases CB1R-mediated signaling.89 These interactions had physiological relevance as observed by inhibition of neurite outgrowth in Neuro2A cells that coexpress MOR and CB1R following cotreatment with agonists to both receptors.89 Electrophysiological studies with oocytes coexpressing MOR and CB1R and where one of the receptors is fused to a chimeric Gqi5 showed that agonists to either receptor elicited Ca2+-activated Cl− currents thus demonstrating the formation of a functional receptor heteromer.110 However, this study did not examine the effect of coadministration of agonists to both receptors on Ca2+-activated Cl− currents.
Studies have described the presence of MOR in immune cells and that receptor activation by morphine increases the expression of chemokine receptor 5 (CCR5) suggesting interactions between these two receptors.183,184 Coimmunoprecipitation studies using differentially epitope-tagged receptors showed that MOR–CCR5 interacting complexes could be detected only in cells coexpressing both receptors.111,112 Examination of the properties of the MOR–CCR5 heteromer showed that treatment with either the MOR agonist, DAMGO, or the CCR5 agonist, RANTES, induced chemotaxis in cells coexpressing both receptors.111 However, pretreatment with DAMGO inhibited RANTES-induced chemotaxis, increased CCR5 phosphorylation, and decreased G-protein coupling to CCR5.111 The converse was observed when cells were pretreated with RANTES111 suggesting cross-desensitization of MOR and CCR5 within the heteromeric complex.
Studies showing that (i) MOR agonist administration induces itching (also known as pruritus)185–187; (ii) scratching responses to itch-inducing stimuli including morphine administration are abrogated in mice lacking gastrin-releasing peptide receptor (GRPR)113,188,189; and (iii) colocalization of MOR and GRPR in the dorsal horn of the spinal cord113 led to the examination of interactions between the MOR isoform, MOR1D (comprising exons 1–3 and 8–9 of the Oprm gene) and GRPR.113
Direct interactions between MOR1D and GRPR were suggested by coimmunoprecipitation studies carried out in heterologous cells expressing epitope-tagged receptors or with spinal cord membranes that endogenously express both receptors.113 Examination of the signaling properties of the MOR1D–GRPR heteromer showed that morphine-induced intracellular Ca2+ release only in cells coexpressing both receptors.113 Interestingly, treatment with morphine enhanced GRPR internalization, while treatment with the GRPR agonist had no effect on MOR internalization in cells coexpressing both receptors.113
Interactions between the orphanin FQ/nociceptin receptor (OFQ/N) and MOR have also been described. These include studies showing that an OFQ/N receptor antagonist potentiated the analgesic effects of morphine both in naïve mice and in mice that were tolerant to this opioid.190,191 Coimmunoprecipitation studies using differentially epitope-tagged receptors showed that MOR and OFQ/N receptors could form interacting complexes in cells coexpressing both receptors.117,118 MOR–OFQ/N heteromers exhibited an enhanced affinity for MOR-selective ligands.117 Interestingly, MOR–OFQ/N receptor heteromerization attenuated the potency of MOR-mediated inhibition of adenylyl cyclase activity and ERK1/2 phosphorylation.118 Moreover, treatment with the OFQ/N receptor agonist, nociceptin, led to desensitization of both MOR- and OFQ/N receptor-mediated signaling, while treatment with the MOR agonist, DAMGO, desensitized only MOR-mediated signaling.118 These studies suggest that MOR–OFQ/N receptor heteromerization may play a role in modulating MOR receptor function.
Substance P exerts its effects through the activation of neurokinin-1 (NK-1) receptors. Both MOR and NK-1 receptors have been detected in dorsal horn neurons of the spinal cord as well as in brain regions implicated in mood disorders, stress responses as well as those that mediate the rewarding effects of drugs of abuse.192,193 In addition, treatment with morphine leads to an increase in the expression of NK-1 receptors in cortical neurons leading to a greater substance P-mediated Ca2+ mobilization than in untreated neurons.194 Moreover, activation of NK-1 receptors modulates neuronal morphine-mediated endocytosis and MOR desensitization.195 These observations together with those showing that the rewarding effects of morphine are not present in NK-1 receptor knockout mice led to studies examining whether these two receptors formed heteromeric complexes.196 Coimmunoprecipitation and BRET assays using differentially epitope-tagged receptors demonstrate the presence of MOR–NK-1 receptor heteromers in HEK-293 cells coexpressing both receptors.116 MOR–NK-1 receptor heteromerization did not lead to changes in the binding and signaling properties of individual protomers.116 However, treatment with the MOR-selective agonist, DAMGO, led to phosphorylation and cointernalization of NK-1 receptors, while treatment with substance P induced the phosphorylation and cointernalization of MOR.116 Moreover, in cells coexpressing MOR and NK-1 receptors but not in cells expressing only MOR, cotreatment with substance P and DAMGO-induced cointernalization of both receptors along with β-arrestin into the same endosomal compartment leading to a delay in the recycling and resensitization kinetics of MOR.116
Another receptor pair investigated for the possible formation of heteromeric complexes involved MOR and somatostatin 2A receptors (sst2a) based on the fact that the latter exhibited 38% homology with the amino acid sequence of MOR.119 Coimmunoprecipitation studies using differentially epitope-tagged receptors demonstrated the presence of interacting MOR–sst2a receptor complexes in HEK-293 cells coexpressing both receptors.119 MOR–sst2a receptor heteromerization did not affect protomer-mediated ligand binding and signaling; however, it modulated agonist-mediated receptor trafficking.119 In this context, in cells coexpressing MOR and sst2a, the sst2a receptor agonist promoted the internalization of both receptors, while the MOR agonist promoted only the internalization of MOR.119 Although it is possible that MOR–sst2a heteromers do not undergo internalization following treatment with MOR agonists (the observed internalization being that of MOR homomers), it is to be noted that MOR agonists desensitize sst2a receptor signaling and sst2a receptor agonists desensitize MOR-mediated signaling.119
Both KOR and β2AR are present in the heart and treatment with the β2AR agonist, isoproterenol, has been shown to increase radiolabeled U69,593 binding to KOR present in rat cardiac sarcolemma.197,198 This suggested interactions between KOR and β2AR. Coimmunoprecipitation studies using differentially epitope-tagged receptors expressed in HEK-293 cells indicated that KOR and β2AR formed interacting complexes.87 While KOR–β2AR heteromerization did not affect the pharmacology of each protomer in the complex, it did modulate trafficking and signaling as examined by observing the phosphorylation of ERK1/2 in cells coexpressing both receptors.87 For example, treatment with a β2AR agonist reduced β2AR internalization in cells coexpressing KOR–β2AR, but could induce internalization in cells expressing only β2AR.87 Moreover, β2AR-mediated increases in ERK1/2 phosphorylation are significantly attenuated in cells coexpressing KOR–β2AR heteromers. Together, these results indicate an important physiological role for KOR–β2AR heteromerization in modulation of β2AR function.
The apelin receptor (APJ) binds to the apelin peptide and plays a role in the modulation of several functions including cardiovascular functions, neuroprotection, pain perception, and regulation of food and water intake.199,200 Given that both KOR and APJ exhibit a similar distribution and function in the cardiovascular and nervous system, studies examined the possible heteromerization between these two receptors.106 Coimmunoprecipitation, colocalization, and BRET assays show that KOR and APJ can form interacting complexes in HEK-293 cells.106 KOR–APJ heteromerization leads to potentiation of ERK1/2 phosphorylation following treatment with either the APJ or KOR agonist compared to what is observed in cells expressing only homomers for either receptor.106 Moreover, KOR–APJ heteromerization leads to a decrease in the activity of protein kinase A, an increase in the activity of protein kinase C, and an increase in cell proliferation.106
Compared to opioid receptors, fewer studies have examined heteromerization of cannabinoid receptors. While heteromerization between opioid receptor subtypes and CB1R has been described in Section 2.1, in this section we describe cannabinoid receptor heteromers that do not involve opioid receptors (see Table 9.2).
Several studies suggested interactions between CB1R and adenosine 2A receptors (A2AR). These include studies showing that (i) activation of A2AR blocks, while inhibition of A2AR promotes, CB1R-mediated effects on evoked glutamate release210; (ii) the rewarding and aversive effects of Δ9-THC were significantly reduced in A2AR knockout mice211; (iii) A2AR antagonist treatment blocks the synergy between MOR and CB1R for CRE-mediated gene transcription212; (iv) locomotor behavior induced by administration of A2AR antagonists is attenuated by CB1R antagonists or in CB1R knockout mice213; and (v) low doses of A2AR antagonists attenuated while higher doses potentiated the self-administration of Δ9-THC and anandamide.214
The formation of interacting CB1R–A2AR complexes in rat striatum was detected by coimmunoprecipitation studies using receptor-selective antibodies.201 BRET using differentially epitope receptors demonstrated that CB1R and A2AR are in close proximity to interact in HEK-293 cells.201 Studies carried out in SH-SY5Y neuroblastoma cells, that endogenously express CB1R and A2AR, show that CB1R activation inhibits forskolin-mediated increases in cAMP levels, and that the latter is blocked by an A2AR antagonist.201 Moreover, A2AR-mediated increases in cAMP levels are blocked by a CB1R antagonist.201
Early studies demonstrating an upregulation of CB1R in liver cells that have resident AT1 angiotensin receptors (AT1R) prompted the examination of interactions between these two receptors.215–218 The heteromerization of CB1R and AT1R was demonstrated by coimmunoprecipitation, colocalization, and BRET assays using differentially epitope-tagged receptors.10 Examination of the signaling properties of the CB1R–AT1R heteromer shows that, in cells coexpressing both receptors, the AT1R agonist induces a rapid, robust dose-dependent increase in ERK1/2 phosphorylation that is drastically reduced either by RNAi-mediated knockdown of CB1R or by inhibiting the activity of diacylglycerol lipase, the enzyme responsible for the synthesis of the endocannabinoid 2-AG.10 In addition, CB1R agonists potentiate while antagonists block AT1R-mediated increases in ERK1/2 phosphorylation.10 Interestingly, an antibody that selectively recognizes the CB1R–AT1R heteromer blocks AT1R-mediated ERK signaling in cells coexpressing both receptors.10 Moreover, in cells expressing the CB1R–AT1R heteromer, CB1R-mediated ERK1/2 phosphorylation is only observed following activation of AT1R by a very low dose of agonist that by itself does not induce ERK1/2 phosphorylation.10
Examination of receptor G-protein coupling shows that, in cells expressing the CB1R–AT1R heteromer, AT1R-mediated ERK1/2 phosphorylation occurs via Gαi instead of Gαq. Moreover, AT1R-mediated intracellular Ca2+ mobilization occurs via Gαq but requires the presence of CB1R since it is significantly attenuated by RNAi-mediated knockdown of CB1R.10 Interestingly, activation of CB1R did not increase intracellular Ca2+ mobilization, although the CB1R agonist could potentiate AT1R-mediated effects.10 Taken together, these studies show that CB1R–AT1R heteromerization provides a mechanism for modulation of AT1R function by CB1R that could be of clinical relevance in pathologies involving these receptors (see Section 3).
Based on coexpression of CB1R and β2AR in several tissues and on inhibition of noradrenergic neurotransmission by CB1R a study examined whether both receptors formed interacting complexes.202 BRET assays using differentially epitope-tagged receptors showed that both receptors were in close proximity to form complexes in live cells.202 The BRET signal was increased in the presence of a CB1R antagonist suggesting that the latter either caused a change in receptor conformation (bringing in close proximity the acceptor and donor pairs) or caused an increase in the surface levels of this heteromer.202 Heteromerization between CB1R and β2AR was shown to lead to alterations in the signaling properties of individual receptors in that it led to an increase in CB1R-mediated ERK1/2 phosphorylation and a decrease in CREB phosphorylation compared to cells expressing only CB1R.202 Moreover, β2AR-mediated ERK1/2 phosphorylation was attenuated by an inverse agonist but not by a neutral agonist to CB1R.202 CB1R–β2AR heteromerization modulated the trafficking properties of individual receptors in that treatment with a β2AR receptor agonist induced CB1R internalization in cells expressing both receptors.202
A recent report has described heteromerization between CB1R and CB2R in vivo.148 The authors used BRET assays with differentially epitope-tagged receptors in HEK-293 cells to show that CB1R and CB2R are in close proximity in live cells.148 They also carried out a proximity ligation assay in SH-SY5Y neuroblastoma cells (endogenously express CB1R) transfected with epitope-tagged CB2R, in pinealocytes obtained from rat pineal glands, and in rat nucleus accumbens slices to show that both receptors are in close proximity (<17 nm) to form interacting complexes.148 The authors examined the signaling properties of CB1R–CB2R heteromers in SH-SY5Y cells transfected with CB2R and reported that a CB2R agonist induces Akt/PKB phosphorylation only in cells that coexpressed both receptors, and that this is significantly attenuated by cotreatment with the CB1R agonist.148 Interestingly, this translated into a physiological response given that CB2R-mediated neurite outgrowth in cells expressing CB1R–CB2R heteromers was blocked by the CB1R agonist.148 These results suggest that the interactions between CB1R and CB2R are antagonistic in nature.
A number of studies have reported interactions between CB1R and D2R. These include studies showing that (i) the D2R agonist, quinpirole, mediated motor behavior is stimulated by the CB1R antagonist, SR141716A203; (ii) G-protein coupling to CB1R is modulated by D2R in cells coexpressing both receptors204,205; and (iii) in primary striatal neuronal cultures treatment either with the CB1R agonist, Hu-210, or with the D2R agonist, quinpirole, leads to inhibition of forskolin-stimulated cAMP levels,204 whereas coadministration of Hu-210 and quinpirole resulted in increases in cAMP levels mediated via CB1R stimulation of Gαs.204 Interestingly, while the presence of D2R is required for CB1R coupling to Gαs, persistent activation of D2R switches the coupling from Gαs to Gαi.205 The formation of CB1R–D2R heteromeric complexes was demonstrated by coimmunoprecipitation studies using differentially epitope-tagged receptors that show that cotreatment with subsaturating concentrations of agonists to both receptors causes an increase in the levels of CB1R–D2R heteromers.206
Coimmunoprecipitation studies along with homogenous TR-FRET assays using SNAP- and CLIP-tagged substrates were used to show that CB1R and orexin-1 receptors (OX1R) form multimeric complexes.207 Examination of the trafficking properties of this heteromer showed that the OX1R agonist was more efficient at promoting heteromer internalization compared to OX1R internalization.207 In heterologous cells coexpressing CB1R and OX1R, a Gq/11-coupled receptor, treatment with the OX1R agonist led to a potent and long-lasting increase in ERK1/2 phosphorylation compared to cells expressing only OX1R, and that this increase was blocked by a CB1R antagonist or by pertussis toxin pretreatment.208 Moreover, coadministration of CB1R and OX1R agonists did not result in additive effects suggesting that just the presence of CB1R was sufficient to modulate OX1R signaling.208 Given the involvement of CB1R and OX1R in feeding, targeting this heteromer pair could lead to the identification of novel therapeutics for the treatment of obesity.
Above, we described studies showing heteromerization between CB1R and D2R and between CB1R and A2AR. In addition, D2R–A2AR heteromerization has been reported.219 This led to the hypothesis that the three receptors could associate to form an oligomeric complex. This was tested with the sequential use of BRET and FRET assays.209 BiFC assay was combined with BRET to show that the three receptors were in close proximity to interact.129 Although this finding is exciting, further studies are needed to determine whether these heterotrimers are present endogenously, and if so, what is their tissue distribution and physiological role.
Several studies have examined heteromerization between dopamine receptor subtypes and between the latter and other GPCRs. Since heteromerization between dopamine receptors and opioid receptor subtypes or CB1R have been described in Sections 2.2 and 2.3, we describe dopamine receptor heteromers that do not involve opioid or cannabinoid receptors in this section (see Table 9.3).
Interactions between D1R and D2R were suggested by the colocalization of D1R and D2R in MSNs in the striatum (reviewed in Refs. 48–50). Further evidence for direct interaction was suggested by the findings that (i) there is a requirement of D1R activation for D2R-mediated locomotor activity241; (ii) enhancement of the acute effects of cocaine requires activation of D1R and of D2R242; and (iii) potentiation in the release of arachidonic acid is observed upon coactivation of D1R and D2R in cells expressing both receptors.243
Evidence for direct interactions between D1R and D2R came from coimmunoprecipitation and confocal FRET studies carried out with rat striatal membranes, primary rat striatal neuronal cultures or with cells coexpressing epitope-tagged receptors.220–224 In addition, protein complementation studies where coexpression of D1R fused to a fragment of luciferase and of D2R fused to a complementary luciferase fragment resulted in the reconstitution of active luciferase supported D1R–D2R heteromerization.125 Signaling assays showed that while activation of D1R led to Gαs/olf-mediated signaling and of D2R to Gαi-mediated signaling, coactivation of both receptors led to increases in intracellular Ca2+ levels.220,221 Since the increase in intracellular Ca2+ levels is not observed in cells that express only D1R or D2R this suggested that the D1R–D2R heteromer can couple to Gαq to induce Ca2+-mediated signaling.220,221 Interestingly, occupancy of D1R in cells expressing D1R–D2R heteromers results in rapid desensitization of heteromer-mediated Ca2+ signaling that is regulated by both the catalytic and RGS domains of GRK2.223
The D1R agonist, SKF83959, has been proposed as a D1R–D2R heteromer-selective ligand based on the finding that (i) it did not increase adenylyl cyclase activity in cells coexpressing D1R–D2R or expressing only D1R; (ii) it increased intracellular Ca2+ release in cells coexpressing D1R–D2R, and this effect was potentiated by coadministration of quinpirole; and (iii) it increased [35S]GTPγS binding to Gαq and to Gαi in cells coexpressing D1R–D2R.221 Finally, D1R–D2R-mediated signaling by SKF83959 in primary striatal neurons via Gαq was found to lead to activation of both cytosolic and nuclear Ca2+/calmodulin-dependent kinase IIα resulting in increased expression of BDNF and ultimately rapid neuronal maturation and differentiation.222 Moreover, treatment with SKF83959 increased grooming behavior in rats and decreased the phosphorylation of the GluR1 subunit of AMPA receptors via Ca2+/calmodulin-dependent kinase IIα in the nucleus accumbens.224
Interactions between D1R and D3R were suggested by studies showing that (i) D1R agonists could induce the expression of mRNA and increase D3R levels, and that this could be blocked by a D1R antagonist244,245; (ii) the coactivation of D1R and D3R caused a potentiation in the expression of substance P mRNA246,247; (iii) D3R knockout mice exhibit a decrease in D1R-mediated expression of the immediate early gene, c-fos248; and (iv) cross talk between D1R and D3R may play a role in rearing habituation and reward behavior to low doses of cocaine.249
Evidence for direct interactions came from coimmunoprecipitation studies showing that D1R and D3R form interacting complexes in cells coexpressing both receptors and in rat striatal membranes.225 BRET assays showed that the two receptors were in close proximity in living cells.225 Treatment with dopamine increased the BRET signal, and this effect could be blocked by expression of a dominant negative mutant of dynamin leading to the suggestion that the increased signal could be due to internalized D1R–D3R heteromers.225 Binding studies with a radiolabeled D1R antagonist showed that dopamine binds with higher affinity to D1R in the presence of D3R.225 This is translated into more potent dopamine-mediated stimulation of adenylyl cyclase activity in cells coexpressing D1R and D3R than in cells expressing only D1R.225 Examination of the trafficking properties of D1R–D3R heteromer showed that treatment with either a D1R or a D3R agonist did not induce heteromer internalization; however, cotreatment with agonists to both receptors induced the internalization of the D1R–D3R heteromer.225
Functional complementation studies suggested that D2R and D3R could form heterodimers.227 For example, expression of D2R with a mutant D3R receptor (that cannot signal) restores the ability of D3R agonists to inhibit adenylyl cyclase activity.227 Coimmunoprecipitation assays with differentially epitope-tagged receptors showed that D2R and D3R could form interacting complexes.227 Signaling assays show that preferential D3R versus D2R agonists (pramipexole, ropinirole, or S32504) were more potent at inhibiting the activity of a chimeric adenylyl cyclase (does not respond to D3R activation) in cells coexpressing D2R and D3R compared to cells expressing only D2R (reviewed in Ref. 228). In contrast, preferential D2R agonists such as pergolide showed a marginal increase in potency in cells coexpressing the D2R–D3R heteromer while no changes were observed for the full D2R/D3R agonist, quinpirole (reviewed in Ref. 228). Interestingly, partial D2R/D3R agonists with antipsychotic activity did not inhibit adenylyl cyclase activity in cells coexpressing both receptors and also could block quinpirole-mediated inhibition of cAMP levels (reviewed in Ref. 228). This suggests that in cells expressing D2R–D3R heteromers partial D2R/D3R agonists can function as antagonists (reviewed in Ref. 228).
Interactions between D1R and A1AR receptors were suggested by studies showing that (i) A1AR activation leads to a decrease in D1R-mediated locomotion and oral dyskinesia in animals,250 while A1AR antagonists enhance D1R-mediated locomotion251; (ii) radiolabeled D1R antagonist affinity is decreased by A1AR agonist252; (iii) A1AR agonists modulate D1R-mediated electroencephalographic arousal11; and (iv) in cells coexpressing A1AR and D1R, A1AR antagonists cause an increase in both basal- and dopamine-mediated increases in cAMP levels while an A1AR agonist blocks dopamine-mediated effects.253
Coimmunoprecipitation studies using receptor-selective antibodies detected the presence of D1R–A1AR interacting complexes in cells coexpressing both receptors.226 In fibroblast cells that coexpress D1R and A1AR, pretreatment with agonists to both receptors decreases D1R-mediated increases in cAMP levels, while pretreatment with either receptor agonist does not. Additional studies using proximity-based assays to confirm the formation of D1R–A1AR heteromers in vivo as well as studies to evaluate whether the D1R–A1AR heteromer exhibits novel properties compared to individual receptor homomers need to be carried out.
Interactions between D2R and A2AR are suggested by studies showing that (i) D2R and A2AR are colocalized in GABAergic striatopallidal neurons254; (ii) selective A2AR agonists decrease D2R binding and signaling229,255,256; (iii) Blockade or knockdown of D2R impairs locomotive behavior, and this effect can be attenuated by A2AR antagonists257,258; and (iv) antagonists of D2R produce catalepsy and this is blocked by A2AR antagonists.259 The formation of interacting D2R–A2AR complexes was demonstrated by coimmunoprecipitation studies and by proximity-based assays such as BRET and FRET.219,230–232 Investigation of the pharmacological properties of the D2R–A2AR in cells coexpressing both receptors showed that D2R agonists caused a decrease in the affinity for radiolabeled A2AR agonist binding.229 Moreover, signaling assays showed that D2R agonists decreased the efficacy and potency of A2AR agonists in cells coexpressing both receptors.229 In addition, studies showed that treatment with either A2AR or the D2R agonists causes internalization of the D2R–A2AR heteromer.219,260
Bivalent ligands comprising of a D2R agonist joined by spacers of different lengths (26–118 atoms) to an A2AR antagonist have been used to show that in striatal membranes coexpressing both receptors, these ligands cause a greater displacement of radiolabeled ligand binding to either D2R or A2AR than the corresponding monovalent ligands.261
A recent study examined D2R and ghrelin receptor (GHSR1a) interactions because of their robust colocalization to a subset of hypothalamic neurons.233 Time-resolved FRET carried out in heterologous cells and with hypothalamic membrane preparations demonstrated that the two receptors are in close proximity to form heteromers.233 In cells coexpressing D2R and GHSR1a, activation of D2R led to increases in intracellular Ca2+ release and not to Gαi-mediated signaling and this could be blocked by a D2R or a GHSR1a antagonist.233 Moreover, cotreatment with agonists to both receptors led to additive effects on intracellular Ca2+ release.233 In addition, cross-desensitization of Ca2+ release is observed in cells coexpressing both receptors in that a 30-min pretreatment with agonist to one receptor attenuated intracellular Ca2+ release mediated via the other receptor.233 Studies with wild-type and ghrelin receptor knockout mice showed that administration of a D2R agonist caused a reduction in food intake in wild-type but not in knockout mice.233 This reduction in food intake can be blocked by a GHSR1a antagonist.233 These results indicate that D2R–GHSR1a heteromerization could play an important role in the modulation of food intake.
Interactions between dopamine and neurotensin receptors were demonstrated by studies showing that (i) type 1 neurotensin receptors (NTS1) are expressed in substantia nigra and ventral tegmental area dopaminergic neurons262; (ii) NTS1 has excitatory effects on dopaminergic neurons present in the midbrain262; (iii) neurotensin causes a modest decrease in the affinity of D2R for receptor agonists262; (iv) D2R autoreceptor function is decreased by NTS1 activation via a protein kinase C- and Ca2+-dependent mechanism262; and (v) activation of D2R attenuates NTS1-mediated Ca2+ mobilization.262
A combination of immunoprecipitation of the long isoform of D2R (D2RL) with fluorescent spectroscopy for detection of the presence of C-terminally epitope tagged (with enhanced CFP) NTS1 in the immunoprecipitate indicated that the two receptors formed interacting complexes only in cells coexpressing both receptors.234 Radiolabeled binding experiments with membrane preparations from cells coexpressing D2RL and NTS1 showed that the NTS1 agonist decreased the affinity of radiolabeled agonist binding to D2RL.234 However, dopamine ligands (agonist or antagonist) had no effect on radiolabeled agonist binding to NTS1.234 Signaling assays showed that the potency of D2RL-mediated signaling is increased in the presence of neurotensin.234 Internalization assays showed that in cells coexpressing D2RL and NTS1 treatment with NTS1 agonist increased D2RL internalization via a PKC-dependent mechanism.235
Studies showing (i) the involvement of striatal oxytocin receptors (OTR) in pair bond formation in monogamous prairie female voles and in regulating social and emotional behaviors in animals (rodents, sheep, humans)263–267; (ii) the involvement of ascending dopaminergic neurons in social attachment behavior1,2; (iii) that a D2R antagonist can block partner preference268,269; and (iv) administration of a D2R agonist into the nucleus accumbens can induce partner preference and this is blocked by an OTR antagonist268,269 suggest interactions between D2R and OTR.
BRET assays in cells coexpressing both receptors showed that D2R and OTR are in close proximity in live cells.236 In situ proximity ligation assays detected the presence of D2R–OTR heteromers in the neuropil of nucleus accumbens and in the dorsal striatum of rats.236 Binding assays with nucleus accumbens membrane preparations showed that a low, but not high, concentration of the OTR agonist increase the maximal number of binding sites of a radiolabeled D2R antagonist and this is blocked by an OTR antagonist.236 Signaling assays with nucleus accumbens membrane preparations showed that the OTR agonist increases the efficacy and potency of D2R-mediated G-protein activity.236 These studies indicate that D2R–OTR heteromers may play a role in modulating social and emotional behavior.
Interactions between dopamine and somatostatin receptors were suggested by studies showing that (i) a decrease of somatostatin levels attenuated dopamine receptor-mediated stereotypy as well as locomotor responses to amphetamine270; (ii) dopamine enhances somatostatin receptor-mediated inhibition of adenylyl cyclase activity271,272; and that (iii) a dopamine receptor agonist attenuates the chronic effects of a long-acting somatostatin analogue.273 This led investigators to examine heteromerization between D2R and somatostatin receptor subtypes.
Direct interactions between D2R and somatostatin receptor 2 subtype (SSTR2) was examined by coimmunoprecipitation and pbFRET assays that demonstrated that these two receptors formed complexes in heterologous cells coexpressing both receptors as well as in primary striatal neuronal cultures.237 Binding studies showed that in cells coexpressing D2R and SSTR2 a combination of agonists to both receptors displaced radiolabeled antagonist binding to D2R more efficiently than the D2R agonist alone.237 Interestingly, in these cells, the SSTR2 agonist could dose-dependently displace radiolabeled antagonist binding to D2R.237 Also, the total number of binding sites for the radiolabeled ligand binding to SSTR2 was increased in the presence of dopamine.237 Examination of signaling properties of the D2R–SSTR2 heteromer showed that D2R-mediated inhibition of cAMP levels was lower for some agonists (dopamine) and not others (quinpirole) than that observed in cells that expressed only D2R.237 Moreover, the presence of the SSTR2 agonist caused an increase in the potency of quinpirole and in both the potency and efficacy of dopamine in inhibiting cAMP levels.237 Examination of the trafficking properties of the D2R–SSTR2 heteromer showed that treatment with D2R agonist enhanced agonist-mediated internalization of SSTR2.237
Heteromerization between D2R and SSTR5 was suggested by the functional rescue of a C-tail deletion mutant of human SSTR5.144 Only when coexpressed with D2R does agonist activation of the mutant SSTR5 lead to inhibition in forskolin-mediated increases in cAMP levels.144 pbFRET supported the formation of D2R–SSTR5 heteromers.144 Interestingly, treatment with agonists to either receptor or cotreatment with agonists to both receptors led to an increase in FRET efficiency, while a decrease in the latter was observed with D2R antagonist.144 This would suggest that either agonist treatment leads to an increase in the level of D2R–SSTR5 heteromers or, alternatively, it changes receptor conformation leading to a decrease in the distance between the fluorescent probes.144 The pharmacological properties of the D2R–SSTR5 heteromer were found to be distinct from those of individual receptor homomers in that the heteromers exhibited a greater affinity for the binding of each protomer agonist.144 Moreover, occupancy of one of the receptors in the D2R–SSTR5 heteromer led to an increase in the binding affinity and signaling by the agonist to the second receptor.144
Studies showing that typical antipsychotic drugs used to treat schizophrenia are D2R antagonists and that some symptoms of schizophrenia are similar to those observed with hallucinogenic serotonin receptor agonists led to the examination of heteromerization between D2R and serotonin 5-HT2A receptors (5-HT2AR).274,275
Coimmunoprecipitation studies show that D2R and 5-HT2AR form interacting complexes.238 Examination of the pharmacological properties of the D2R–5-HT2AR heteromer in cells or in striatal membranes coexpressing both receptors show that a D2R agonist increased the binding affinity of a 5-HT2AR agonist, and this is blocked by a D2R antagonist.238 The presence of D2R had no effect on serotonin-mediated inositol phosphate production, but it increased the potency of a hallucinogenic 5-HT2AR agonist.238 Interestingly, D2R activation had no effect on serotonin-mediated inositol phosphate production, but decreased the potency of the hallucinogenic 5-HT2AR agonist.238 Behavioral studies with 5-HT2AR knockout mice show that the antipsychotic effects of the D2R antagonist, haloperidol, require the presence of 5-HT2AR.238
Heteromer formation between D4R and α adrenergic receptors (αAR) was examined based on studies showing that the D4R gene is the most prominently expressed dopamine receptor gene in the rat pineal gland, and that its mRNA expression is circadian and controlled by adrenergic signaling.276
Coimmunoprecipitation and BRET studies with differentially epitope-tagged receptors show that D4R and α1BAR form interacting complexes.239 Pharmacological characterization of D4R–adrenergic receptor heteromers shows that the D4R agonist decreases the ability of an α1BAR agonist to displace radiolabeled antagonist binding to α1BAR.239 Signaling assays in cells expressing the D4R–α1BAR heteromer or in pineal gland preparations show that pretreatment or cotreatment with a selective D4R agonist leads to an inhibition of α1BAR-mediated phosphorylation of ERK1/2 and of Akt/PKB.239 Interestingly, a D4R receptor antagonist could also block α1BAR-mediated signaling and the converse also occurred in that adrenergic antagonists could block D4R-mediated signaling.239 Given that D4R expression is circadian, and that α1BAR plays a role in modulating melatonin synthesis and release from the pineal gland, the formation of D4R–α1BAR heteromers provides a means for regulating melatonin production and activity.239
Studies examining D4R–αAR heteromerization also examined whether D4R and β1 adrenergic receptors (β1AR) formed heteromers. Coimmunoprecipitation and BRET studies with differentially epitope-tagged receptors show that D4R and β1AR form interacting complexes.239 Pharmacological characterization of D4R–β1AR heteromers shows that the D4R agonist decreases the ability of a β1AR agonist to displace radiolabeled antagonist binding to β1AR.239 Signaling assays in cells expressing either the D4R–β1AR heteromer or in pineal gland preparations show that pretreatment or cotreatment with a selective D4R agonist leads to an inhibition of β1AR-mediated phosphorylation of ERK1/2 and of Akt/PKB.239 Interestingly, a D4R receptor antagonist could also block β1AR-mediated signaling and the converse also occurred in that adrenergic antagonists could block D4R-mediated signaling.239 These studies suggest that like the D4R–α1BAR heteromer the D4R–β1AR heteromer could also play a role in regulating melatonin production and activity.239
Based on studies showing (i) colocalization of D2R-like receptors with β-AR in rat primary cortical neurons and in prefrontal cortex277 and that (ii) D2R-like receptor activation inhibited β-AR-mediated increases in cAMP levels,277 heteromerization between D4R and β2-AR was examined.
Coimmunoprecipitation, BRET, and binding assays with β2-AR immunoprecipitates, obtained from detergent solubilized mouse brain extracts using alprenolol coupled to Sepharose beads, show that D4R and β2-AR form interacting complexes and are in close proximity in cells or tissues coexpressing both receptors.240 However, further studies are needed to evaluate whether the D4R–β2-AR heteromer exhibits novel properties compared to individual receptor homomers.
The possibility of formation of heteromers involving D2R, metabotropic glutamate receptor 5 (mGluR5), and adenosine2A receptors (A2AR) was suggested by studies showing (i) similar localization of these receptors in striatopallidal neurons278–280; (ii) that a mGluR5 agonist inhibited D2R-mediated contralateral turning behavior, and this was potentiated by an A2AR agonist and attenuated by an A2AR antagonist281; and (iii) that a mGluR5 agonist reduced the high affinity state for the D2R agonist, and this was potentiated by an A2AR agonist.281
Coimmunoprecipitation studies carried out with cells coexpressing D2R, A2AR, and mGluR5, and also with rat striatal membranes, show that these receptors form interacting complexes.127 BRET assays with cells expressing a constant ratio of A2AR tagged with luciferase and D2R tagged with YFP show that increasing amounts of mGluR5 can disrupt the BRET signal.127 Similarly, increasing amounts of A2AR could disrupt the BRET signal between D2R tagged with luciferase and mGluR5 tagged with YFP.127 This suggests that the three receptors are in close proximity to interact.127 The formation of D2R–A2AR–mGluR5 was confirmed by a combination of BRET and BiFC assays.127 For this, cells were transfected with A2AR tagged with luciferase, mGluR5 fused with an N-terminal fragment of YFP and D2R fused with a C-terminal fragment of YFP.127 A BRET signal was observed only when all three receptors were expressed in a cell suggesting that these receptors are in close proximity to form heterotrimers.127 Further studies are needed in order to elucidate whether the D2R–A2AR–mGluR5 heterotrimer exhibits novel properties compared to individual receptor homomers or heterodimers.
Although it has become increasingly evident that GPCRs can form heterodimers,61,63,64,282 very littleinformation is available aboutthe physiological role of GPCR heteromers. Pharmacological and coimmunoprecipitation studies using receptor-selective antibodies suggest that some receptor heteromers may play a role in the disease state. However, care has to be taken in the interpretation of such data given that such studies do not rule out a contribution of receptor homomers to the disease state. The recent development of antibodies that selectively recognize receptor heteromers, but not receptor homomers, and of TAT-based peptides that can disrupt heteromer formation, has provided us with tools to study the role of GPCR heteromers in the disease state.7–10 In the following sections, we describe a few cases where a role for GPCR heteromers involving opioid, cannabinoid, and dopamine receptors in pathology has been implicated.
Antibodies that selectively recognize the MOR–DOR heteromer detect increased heteromer levels in brain regions involved in pain processing following chronic morphine administration under a paradigm that leads to the development of tolerance.7 Taken with the fact that MOR–DOR heteromers signal via the β-arrestin-mediated signaling cascade100 and that β-arrestin or DOR knockout mice do not develop tolerance to morphine162,283 would suggest that MOR–DOR heteromers play a role in the development of tolerance to this opioid. This is supported by a study showing that the administration of a MOR transmembrane 1 TAT peptide fusion protein, which selectively disrupts the MOR–DOR heteromer, resulted in an enhanced analgesic response to morphine with a concomitant decrease in tolerance to the drug.8
Studies have also examined whether the DOR–MOR heteromer could be a target for the development of therapeutics that are as efficacious as morphine but lack the associated adverse side effects. In this context, a study showed that low doses of the DOR antagonists, TIPPψ or naltrindole, could potentiate intrathecal morphine-mediated analgesia.94,160 Moreover, DOR antagonist administration to animals exhibiting tolerance to morphine led to a restoration of the analgesic response and potency of morphine.160 Based on similar observations, bivalent ligands targeting the DOR–MOR heteromer were developed.284 These bivalent ligands comprised a MOR agonist pharmacophore, oxymorphone, joined to a DOR antagonist pharmacophore, naltrindole, by spacers of variable lengths (16–21 atoms).284 The bivalent compounds exhibited antinociceptive activity.284 Moreover, compounds with spacer lengths between 19 and 21 atoms did not exhibit antinociceptive tolerance and physical dependence and were less rewarding than morphine.284,285
An involvement of DOR–KOR heteromers in thermal allodynia was suggested by studies demonstrating the presence of these heteromers in peripheral sensory neurons.286 Moreover, the antiallodynic effects of the KOR agonist, 6′-GNTI, to prostaglandin E2-induced thermal allodynia were blocked by either DOR or KOR antagonists.286 This, together with the finding that the antiallodynic effects of the DOR agonist, DPDPE, were potentiated by a DOR–KOR heteromer-selective antibody supports a role for the DOR–KOR heteromer in modulating thermal allodynia.286
A recent study has implicated the CB1R–DOR heteromer in neuropathic pain.9 This study found that levels of CB1R and DOR are increased in distinct brain regions following peripheral nerve lesion.9 An antibody selective for the CB1R–DOR heteromer detected increased heteromer levels in the cortex, hypothalamus, and striatum of lesioned animals.9 In addition, CB1R–DOR heteromerization led to a decrease in DOR activity in the cortex of lesioned animals9 and this decrease could be attenuated by low, nonsignaling doses of CB1R ligands.9 Thus, the CB1R–DOR heteromer is a novel target for the treatment of neuropathic pain.
Antibodies that selectively recognize the CB1R–AT1R heteromer detect a significant upregulation of this heteromer in hepatic stellate cells of rats chronically treated with ethanol.10 This leads to increased AT1R-mediated signaling that can be blocked by a CB1R antagonist or by inhibiting the synthesis of the endocannabinoid, 2-AG.10 In addition, the AT1R-mediated increase in profibrogenic markers in hepatic stellate cells of rats chronically treated with ethanol was completely blocked by a CB1R antagonist.10 These results suggest that the profibrogenic activity of AT1R in ethanol-induced liver fibrosis requires CB1R activity. Thus, the CB1R–AT1R heteromer represents a novel therapeutic target for the treatment of this disease.
An involvement of dopamine receptor subtype heteromers in major depression has been suggested by studies showing an increase in D1R–D2R interacting complexes in the striatum of postmortem individuals that suffered from major depression.287 Interestingly, bilateral infusion into the rat prefrontal cortex of a TAT peptide that disrupts D1R–D2R heteromers results in a decrease in the levels of D1R–D2R heteromers and leads to antidepressant-like effects in the forced swim test.287 This suggests an involvement of D1–D2 receptor heteromers in the pathology associated with major depression.
Involvement of D1R–D2R heteromers in schizophrenia was suggested by studies showing that clozapine, a drug used to treat this condition, binds to both D1R and D2R but that the high affinity binding to D1R decreases in the presence of D2R.288,289 Moreover, FRET assays showed that clozapine appears to modulate the formation of D1R–D2R heteromers in that high concentrations of the drug increase the FRET signal between D1R and D2R while low concentrations attenuate this signal.288,289 Further support for the involvement D1R–D2R heteromers in schizophrenia came from studies showing that acute treatment with a heteromer-selective agonist increased the striatal expression of brain-derived neurotrophic factor, a molecule that has been linked to schizophrenia (for review, see Ref. 290).
An involvement of D2R–5-HT2AR heteromers in schizophrenia is suggested by studies showing that coexpression of D2R in cells increases the signaling potency of a hallucinogenic 5-HT2AR agonist, but activation of D2R decreases it.238 Moreover, the antipsychotic effects of the D2R antagonist, haloperidol, are decreased in 5-HT2AR knockout mice238 suggesting that the D2R–5-HT2AR heteromer is a target of this drug.
An involvement of CB1R–D2R–A2AR heteromers in schizophrenia has been suggested based on studies showing upregulation of CB1R, D2R, and A2AR in the postmortem brains of schizophrenics.291–293 Moreover, in animal models of schizophrenia, treatment with antipsychotics decreases CB1R levels in the nucleus accumbens.294 These reports together with studies describing CB1R–D2R and CB1R–A2AR interactions (described in Section 2) and studies showing that these three receptors can form heteromers in live cells suggest that the CB1R–D2R–A2AR heteromer could be a novel therapeutic target in the treatment of schizophrenia.
The involvement of D2R–A2AR heteromers in Parkinson’s disease is suggested by studies showing that the antagonistic D2R–A2AR interactions were increased in the striatal membranes from rat models of Parkinson’s disease (reviewed in Ref. 295). In addition, studies indicate that A2AR antagonists have anti-Parkinsonian effects (reviewed in Ref. 295). For example, A2AR antagonist administration could attenuate dopamine-mediated neuronal degeneration in animal models of Parkinson’s disease (reviewed in Ref. 295). Given that D2R–A2AR heteromers are constitutively formed, it is possible that A2AR antagonists targeting the heteromer could increase D2R-mediated signaling leading, in turn, to an attenuation of Parkinson’s disease symptoms (reviewed in Ref. 295).
The heterotrimer involving D2R–mGluR5–A2AR could also be a novel therapeutic target in the treatment of Parkinson’s disease (reviewed in Ref. 295). This is supported by studies showing that mGluR5 antagonist-mediated induction of motor activity requires the presence of both D2R and A2AR and that a combination of mGluR5 and A2AR antagonists could potentiate locomotor behavior in reserpinized mice (reviewed in Ref. 295).
Another heterotrimer that could be a novel target for the treatment of Parkinson’s disease involves CB1R–D2R–A2AR (reviewed in Ref. 295). Low doses of CB1R and A2AR antagonists could synergize to potentiate D2R-mediated anti-Parkinsonian effects during the early stages of Parkinson’s disease when a substantial release of dopamine still occurs (reviewed in Ref. 295).
Recent evidence indicates that some GPCR heteromers including those associated with drugs of abuse are upregulated under pathophysiological conditions. However, the physiological role of these heteromers in pathology is not clearly understood due to the lack of tools to distinguish between heteromer and homomer-mediated effects. For this, heteromer-specific ligands, agonists and/or antagonists, are needed to elucidate the role of GPCR heteromers in normal and pathophysiological conditions. Although heteromer-specific ligands for a few GPCR heteromers have been reported they are mostly receptor agonists and need to be extensively evaluated for their heteromer selectivity in both in vitro and in vivo assays. Moreover, the availability of heteromer-selective antagonists would help in the evaluation of heteromer levels in the normal and diseased states as well as in studies examining heteromer pharmacology. The recent development of antibodies that selectively recognize a GPCR heteromer in endogenous tissue, and of TAT peptides that can specifically disrupt the heteromer of interest, could help to elucidate the role of GPCR heteromers in pathological conditions.7–10 This is clearly shown in the case of MOR–DOR heteromers where heteromer-selective antibodies demonstrate that these heteromers are increased in brain areas involved in pain perception following chronic treatment with morphine under a paradigm that leads to the development of tolerance.7 In addition, studies with TAT peptides that selectively disrupt the MOR–DOR heteromer suggest a role for the heteromer to keep morphine-mediated signaling via MOR in the desensitized state.8
Studies demonstrating that GPCR heteromers are upregulated during a disease state suggest that these heteromers could be novel targets for the development of therapeutics to treat these diseases. If heteromers are disease-specific and/or exhibit unique tissue specificity, they would serve as ideal drugs for the development of therapeutics with potentially lesser side effects than the ones that are currently available.
We thank Jonathan Wardman and Steven D. Stockton, Jr. for critical reading and helpful comments. L. A. D. is supported by NIH grants DA008863 and DA019521.