GPCR heteromers: Beyond the monomeric paradigm in neuropsychiatric drug discovery

Abstract

G protein-coupled receptors (GPCRs) constitute the largest druggable superfamily of membrane proteins and underlie approximately 35% of all approved medicines. Yet this conventional paradigm leaves a substantial minority of neuropsychiatric patients with incomplete efficacy, treatment resistance, or dose-limiting adverse effects. Converging evidence from structural biology, single-molecule biophysics, and molecular pharmacology has dismantled the monomeric model. GPCRs function in vivo as dynamic homomeric and heteromeric complexes in brain regions directly implicated in disease, including the striatum, prefrontal cortex, hippocampus, and extended amygdala. Here, we advance the central proposition that the effective drug target in the central nervous system is not an isolated receptor monomer but a context-dependent macromolecular assembly whose ligand-binding, signaling, and trafficking properties arise from specific protein–protein interactions. We synthesize evidence for exemplary heteromers across disease-relevant circuits, A2AR-D2R in the striatum, 5-HT2AR-mGlu2R in the prefrontal cortex, D1R-D2R and D1R-mGlu5R in the striatum, D2R-5-HT2AR, and higher-order A2AR-D2R-mGlu5R assemblies together with Sigma-1 receptor-mediated modulation. We then assess translational strategies, including bitopic ligands, nanobodies, and cryo-EM-guided structure-based design, and propose a roadmap centered on native-tissue validation, state-selective assays, and circuit-level behavioral readouts. This perspective provides a conceptual framework for achieving anatomical selectivity, signaling precision, and the uncoupling of therapeutic efficacy from adverse effects.

Introduction

The human genome encodes approximately 800 G protein-coupled receptors (GPCRs), which constitute the largest and most pharmacologically exploited superfamily of membrane proteins [1]. In neuropsychiatry, GPCRs have provided the molecular foundation for therapeutic interventions across schizophrenia, major depressive disorder, addiction, anxiety, and chronic pain [2]. The dopamine hypothesis of psychosis gave rise to antipsychotics targeting dopamine D2 receptors (D2R) [3], whereas the monoamine theory of depression spawned successive generations of serotonergic and noradrenergic modulators [4]. The opioid receptor family remains the principal target for analgesia despite its well-documented liabilities [5]. GPCR-directed agents account for roughly 35% of all approved drugs, a figure that underscores both the biological prominence and the clinical dependency on this receptor class [6].

Yet this legacy masks an escalating crisis in neuropsychiatric drug development. The conventional “one drug, one receptor” paradigm, dominant in pharmacological thinking for decades, is increasingly recognized as a reductionist oversimplification that fails to capture the biological reality of receptor function in the central nervous system (CNS) [6]. The clinical consequences are unmistakable: incomplete efficacy in a substantial subset of patients; dose-limiting adverse effects that erode adherence; profound inter-individual variability in treatment response; and treatment-resistant populations that remain underserved by existing pharmacotherapies. These limitations are not merely artifacts of imperfect drug design; rather, they reflect a fundamental conceptual gap in how receptor pharmacology is understood in vivo.

We therefore advance the following central proposition: the effective drug target in the CNS is not an isolated receptor monomer but a context-dependent macromolecular assembly whose ligand-binding, signaling, and trafficking properties arise from specific protein–protein interactions. This proposition carries three testable implications. First, the pharmacological unit may be spatially restricted to particular cell types or brain regions, opening a route to anatomical selectivity. Second, the conformational and transductional output of the complex can differ from that of either protomer alone, creating opportunities for signaling bias. Third, the assembly state itself is dynamic, such that targeting a complex may engage pathways inaccessible to a monomeric reference. This review synthesizes the evidence for this claim and outlines a translational roadmap.

From monomers to heteromers: A conceptual Shift

Over the past two decades, converging evidence from structural biology, single-molecule biophysical imaging, and molecular pharmacology has dismantled the monomeric model of GPCR function [7,8]. GPCRs assemble into dynamic homodimers, heterodimers, and higher-order oligomeric structures that exhibit conformational landscapes, signaling signatures, and trafficking behaviors distinct from those of their constituent protomers. The proposition that GPCRs function as dimers or higher-order oligomers is not new; it has been debated since the discovery of obligatory class C GPCR dimers. However, the weight of evidence now supports the view that heteromerization is a general feature of receptor biology rather than an exception confined to a handful of receptor families [9].

Importantly, GPCR heteromers are not merely artifacts of heterologous overexpression systems. A growing body of in vivo evidence has documented the existence of specific heteromeric species in brain regions directly implicated in neuropsychiatric disease, including the striatum, prefrontal cortex, hippocampus, and extended amygdala [8]. These spatially restricted heteromeric populations offer an elegant solution to a longstanding problem in CNS drug development, how to achieve anatomical selectivity without sacrificing target engagement. A compound that recognizes a heteromer-enriched interface may spare the same receptor monomer or homodimer expressed elsewhere, thereby limiting off-target effects. The pharmacological implications are threefold. (1) Novel binding pockets. Heteromeric assemblies can generate ligand-binding pockets with allosteric properties absent from either protomer, enabling selective engagement of receptor complexes in specific cell types or brain regions. (2) Signaling bias. They can bias signal transduction toward specific pathways, offering a route to disentangle therapeutic efficacy from adverse effects. (3) Altered trafficking kinetics. They can modify receptor internalization and desensitization, potentially mitigating the tachyphylaxis and tolerance that plague chronic neuropsychiatric pharmacotherapy [10]. These three features jointly redefine what constitutes a “drug target,” shifting the focus from individual receptor genes to context-dependent receptor complexes.

However, a critical conceptual distinction must be drawn between physical heteromers and functional heteromers. A physical heteromer (heterodimer, heterotrimer, or higher-order complex) comprises two or more receptors held in direct spatial proximity; such complexes represent only a small and highly dynamic fraction of the total receptor population (e.g., μ- and κ-opioid receptor heteromers) [11]. Because the interface is structurally circumscribed, a small molecule can in principle be designed to engage the complex selectively while sparing the far more abundant monomeric receptors, opening a route to anatomical and signaling selectivity. By contrast, a functional heteromer refers to a receptor pair that displays coordinated or reciprocal signaling through shared downstream pathways without necessarily residing in a stable physical complex; a selective ligand acting on such a pair retains the pharmacodynamics and adverse-effect profile of its cognate receptor. Every physical heteromer can manifest as a functional entity, but the converse is not true. Accordingly, the two categories demand fundamentally different drug-development strategies: physical heteromers are best pursued with bitopic or allosteric ligands that exploit a discrete quaternary interface, whereas functional heteromers may require combination therapies or pathway-selective (biased) ligands to achieve a therapeutic advantage.

Exemplary heteromers in neuropsychiatric circuits

The A2AR-D2R heteromer in the striatum

The adenosine A2A–dopamine D2 receptor (A2AR–D2R)heteromer is a functionally relevant complex in which the two GPCRs engage in cross-talk via shared signaling pathways, although the extent of stable physical association remains under investigation. As one of the most extensively characterized GPCR heteromers, it is abundantly expressed in the striatum, where it regulates motor function, motivational salience, and reward processing [12]. Within this heteromer, A2AR agonists allosterically reduce D2R affinity for dopamine and inhibit D2-mediated signaling, whereas A2AR antagonists potentiate D2R function [12]. This antagonistic receptor–receptor interaction has been validated in native striatal tissue, including striatal astrocytes, using proximity ligation assays and co-immunoprecipitation, confirming that the A2AR–D2R interface is not merely an overexpression artifact [10].

Despite this well-defined interaction, a key challenge persists: selective A2AR or D2R ligands retain their receptor-specific pharmacodynamics and adverse effects. In other words, compounds that target only one protomer of the heteromer may not fully exploit the therapeutic potential of modulating the complex as a whole. Nevertheless, the molecular antagonism between A2AR and D2R has direct therapeutic relevance. A2AR antagonists are under investigation as adjunctive therapies for Parkinson's disease (PD), aiming to enhance dopaminergic neurotransmission without the dyskinesias associated with direct D2R agonism. Notably, the A2AR antagonist istradefylline is already approved as an add-on treatment for “off” episodes in levodopa-treated PD patients. Targeting the A2AR–D2R heteromer may also hold promise for ameliorating negative symptoms in schizophrenia and for treating psychostimulant addiction [13,14]. The confirmed existence of this heteromer in native striatal tissue validates the concept that receptor complexes, rather than isolated D2R monomers, represent the true physiological targets of dopaminergic pharmacology in basal ganglia circuits.

The 5-HT2AR-mGlu2R heteromer in the prefrontal cortex

A landmark discovery was the identification of a physical and functional complex between the serotonin 5-HT2A receptor (5-HT2AR) and the metabotropic glutamate receptor 2 (mGlu2R) in the prefrontal cortex, a region centrally implicated in schizophrenia pathophysiology and in the mechanisms of action of both antipsychotic and psychedelic drugs [15]. This heteromer serves as a molecular integrator of serotonergic and glutamatergic signaling. Classic hallucinogens such as lysergic acid diethylamide (LSD) and psilocybin exert their psychoactive effects, at least in part, through this 5-HT2AR-mGlu2R complex [16]. Conversely, mGlu2R agonists and positive allosteric modulators can attenuate 5-HT2AR-mediated hallucinogenic responses, suggesting that allosteric modulation across the heteromeric interface may underlie antipsychotic efficacy.

Fribourg and colleagues decoded the signaling logic of the 5-HT2AR-mGlu2R heteromer and proposed that it constitutes a unifying molecular target for antipsychotic drug action, thereby linking the serotonergic and glutamatergic hypotheses of schizophrenia through a single macromolecular entity [17]. The balance index—the difference between changes in Gi/o and Gq/11 signaling—predicts the pro- or antipsychotic behavioral effects of ligands acting at the complex: atypical antipsychotics such as clozapine and risperidone produce a high balance index, whereas psychedelics shift it toward a pro-psychotic pattern [17]. Subsequent studies identified specific transmembrane residues essential for heteromerization; disruption of the 5-HT2AR-mGlu2R interface abolishes hallucinogen-specific behavioral responses in preclinical models [18].

The heteromeric framework also accommodates conflicting experimental observations. Although Delille and colleagues confirmed the physical association of 5-HT2AR and mGlu2R receptors, they reported that heterocomplex formation did not invariably produce functional cross-talk in all recombinant cellular contexts, highlighting the importance of cellular background and assay conditions in detecting heteromer-specific pharmacology [19,20]. This is instructive rather than discouraging: it identifies the cell-type- and assay-dependent conditions under which heteromer-specific pharmacology becomes detectable, a methodological caveat that any drug-discovery program must address.

D1R-D2R and D1R-mGlu5R heteromers: non-canonical signaling in the striatum

The dopamine receptor family itself is a rich source of heteromeric diversity. The D1R–D2R heteromer, co-expressed in a subpopulation of striatal medium spiny neurons, couples to Gq/11 proteins and phospholipase C to mobilize intracellular calcium, a signaling profile entirely distinct from the canonical Gs (D1R) or Gi/o (D2R) coupling of the individual protomers [21]. This non-canonical calcium signaling cascade links the D1R-D2R heteromer to calcium/calmodulin-dependent protein kinase IIα (CaMKIIα) activation, brain-derived neurotrophic factor (BDNF) production, and neuronal growth, implicating it in neuroplasticity mechanisms relevant to schizophrenia and mood disorders [21].

Similarly, the D1R forms heteromeric complexes with the mGlu5R, and this assembly mediates non-canonical dopamine signaling that becomes dysregulated in PD. Sebastianutto and colleagues demonstrated that D1R-mGlu5R nanocomplexes are strongly upregulated in the dopamine-denervated striatum in rodent models of PD, where they exacerbate phospholipase C (PLC) signaling and intracellular calcium release, thereby causally linking D1R-mGlu5R-dependent PLC signaling to l-DOPA-induced dyskinesia [22]. Targeting this specific complex may therefore offer a path toward addressing the motor and cognitive symptoms of PD with greater precision than current dopamine replacement strategies.

The D2R-5-HT2AR heteromer: convergence of psychotomimetics

Adding further complexity, the dopamine D2R and the serotonin 5-HT2AR, both primary targets of antipsychotic medications, also form functional heteromers. Borroto-Escuela and colleagues reported that hallucinogenic 5-HT2AR agonists such as LSD and DOI enhance the recognition and signaling of the D2R protomer within D2R–5-HT2AR heteroreceptor complexes, as demonstrated by in situ proximity ligation assay in the ventral and dorsal striatum [23]. This provides a molecular mechanism for the cross-talk between serotonergic hallucinogens and dopaminergic systems, and may contribute to the psychotic actions of these compounds.

Albizu and colleagues further demonstrated functional cross-talk and heteromerization between 5-HT2AR and D2Rs, with implications for the balance between antipsychotic efficacy and extrapyramidal adverse effects [24]. In particular, the relative affinity of an antipsychotic for 5-HT2AR versus D2R, operating through this complex, may help explain why some 5-HT2AR inhibitors possess antipsychotic properties while structurally related compounds do not. These findings illustrate how heteromeric assemblies can serve as convergence points for multiple neurotransmitter systems, and why drugs designed against monomeric receptors may inadvertently miss, or inappropriately engage, integrated signaling nodes.

Higher-order oligomers: the A2AR-D2R-mGlu5R complex

The heteromeric landscape extends beyond pairwise interactions to include higher-order oligomeric assemblies. Cabello and colleagues provided evidence that adenosine A2ARs, dopamine D2Rs, and mGlu5Rs can co-assemble into higher-order oligomers in living cells, using bimolecular fluorescence complementation and sequential resonance energy transfer [25]. High-resolution immunoelectron microscopy confirmed that the three receptors co-distribute within the extrasynaptic plasma membrane of the same dendritic spines at putative glutamatergic striatal synapses, and co-immunoprecipitation in rat striatal homogenates demonstrated their native association [25].

Such complexes challenge the very notion of discrete receptor “targets” and suggest that the effective pharmacological unit in densely innervated brain regions may be a multi-receptor signaling hub rather than a single polypeptide. The A2AR–D2R–mGlu5R assembly integrates adenosinergic, dopaminergic, and glutamatergic inputs within a spatially defined postsynaptic compartment, precisely the kind of integrative unit that conventional reductionist pharmacology is poorly equipped to address.

Beyond canonical GPCRs: the sigma-1 receptor as a non-classical modulator

The heteromeric framework of receptor signaling extends far beyond canonical class A GPCRs, encompassing non-classical entities that function as allosteric architects of neurotransmission. Among the most conceptually important of these is the sigma-1 receptor (S1R), a ligand-regulated molecular chaperone that lacks the seven-transmembrane architecture typical of classical GPCRs. Instead, S1R adopts a trimeric organization, with each protomer contributing a single transmembrane domain. This unusual topology underpins its role as a dynamic integrator of receptor and ion channel signaling at two critical cellular loci: the plasma membrane and mitochondria-associated endoplasmic reticulum (ER) membranes (MAMs) [26].

S1R engages in physical and functional complexes with dopamine D1 and D2 receptors [27,28], μ-opioid receptors [29], 5-HT receptors [30,31], and N-methyl-d-aspartate (NMDA) receptors (NMDARs) [32], among others [33], acting as an allosteric modulator that fine-tunes the signaling output of its partners. Under basal conditions, S1R is sequestered at the ER, bound to the chaperone BiP/GRP78. However, upon cellular stress, such as oxidative injury, ER calcium depletion, or bioenergetic crisis S1R rapidly dissociates from BiP/GRP78, translocates to the plasma membrane, and stabilizes interactions with its receptor partners through direct protein–protein interactions [26,34].

It is important to distinguish S1R-containing complexes from classical GPCR heteromers: because S1R is not a canonical seven-transmembrane GPCR, the terms S1R-containing receptor complexes, heterocomplexes, or receptor–chaperone complexes are more appropriate than equating every S1R interaction with a GPCR heterodimer. With this qualification, a growing body of structural and biochemical evidence supports genuinely physical S1R-containing complexes. Several lines of evidence support the formation of S1R-containing receptor complexes. Co-immunoprecipitation and proximity ligation assays have demonstrated physical associations between S1R and D1R in striatal neurons, where S1R acts as a positive allosteric modulator of D1R-mediated signaling [27]. Similarly, S1R-D2R heterocomplexes have been identified in the striatum, where cocaine-bound S1R allosterically inhibits D2R signaling, suggesting a mechanism for cocaine's modulation of dopaminergic neurotransmission [28]. In the case of 5-HT1ARs, S1R has been shown to form heterocomplexes in the medial prefrontal cortex (mPFC), where S1R activation enhances 5-HT1AR-mediated neuroplasticity and produces faster antidepressant-like effects [30,31]. Additionally, S1R interacts with NMDARs in the hippocampus, where S1R activation promotes NMDAR trafficking to the plasma membrane, thereby enhancing glutamatergic neurotransmission [32].

Pharmacological modulation of S1R via these receptor interfaces offers substantial therapeutic potential. The selective S1R agonist SA4503, for example, potentiates NMDAR function and promotes neuroprotection, indicating potential applications in neurodegenerative disorders [35,36]. Another selective agonist, YL-0919, maintains normal protein function and regulates neuroinflammatory homeostasis within the brain [33,37,38]. Furthermore, fluvoxamine, an antidepressant with high affinity for S1R, may exert its therapeutic effects in part through S1R-mediated modulation of 5-HT1ARs and other receptor complexes [39,40]. These findings illustrate how S1R-containing heterocomplexes can serve as targets for neuropsychiatric drug development, with S1R ligands capable of fine-tuning the signaling output of multiple receptor systems simultaneously.

The inclusion of S1R within the heteromeric landscape underscores a broader conceptual point: the boundaries of what constitutes a “receptor target” are blurring. Effective neuropsychiatric pharmacology will increasingly require an integrative view that encompasses GPCR–GPCR, GPCR–ion channel, and even GPCR–kinase assemblies. The S1R example is particularly instructive because it demonstrates that modulatory capacity need not reside in a canonical seven-transmembrane receptor at all; instead, it can emerge from a chaperone-like protein that reorganizes the signaling output of an entire receptor ensemble. This realization expands the druggable landscape and suggests that future therapeutics may succeed not by targeting receptors in isolation, but by selectively reshaping the higher-order complexes that govern neural circuit function.

Toward heteromer-targeted drug discovery

Translating the heteromeric paradigm into clinically viable therapeutics presents formidable but increasingly surmountable challenges. Heteromers lack standardized high-throughput screening assays; their expression in native tissues is often low-abundance and cell-type-specific; and their conformational dynamics are substantially more complex than those of monomers, requiring sophisticated biophysical approaches to map allosteric binding sites and signaling bias. A further complication is that the relevant assembly state may be transient or stabilized only under specific physiological conditions, rendering it difficult to capture in conventional screening formats.

Nevertheless, the field is advancing rapidly along complementary axes (Table 1). Bitopic ligands designed to engage orthosteric and allosteric sites simultaneously within heteromeric interfaces are entering preclinical development [41]. Nanobodies and conformationally selective antibodies that stabilize specific heteromeric states offer alternative modalities with potentially superior selectivity profiles [42]. Although cryo-electron microscopy structures of heteromeric GPCRs remain sparse, they are beginning to provide the atomic-resolution templates necessary for structure-based drug design. Finally, the integration of single-cell transcriptomics with spatial proteomics is mapping heteromeric expression patterns with unprecedented resolution [43].

Table 1.

Translational strategies for heteromer-targeted drug discovery.

Strategy	Principle	Current status
Bitopic ligands	Simultaneously engage orthosteric and allosteric sites within heteromeric interfaces	Entering preclinical development
Nanobodies/conformationally selective antibodies	Stabilize specific heteromeric states; potentially superior selectivity	Alternative modality under investigation
Cryo-EM structures	Atomic-resolution templates for structure-based drug design	Sparse but growing
Single-cell transcriptomics + spatial proteomics	Map heteromeric expression patterns; enable patient stratification by target presence	Increasingly high-resolution

The convergence of these approaches enables patient-stratification strategies rooted in target presence rather than symptomatology alone, a shift that may prove decisive for disorders in which the same clinical phenotype arises from distinct molecular substrates. A patient in whom a therapeutically relevant heteromer is absent or below a functional threshold may derive little benefit from a heteromer-selective agent, thereby justifying a stratified rather than universal indication. Importantly, drug development strategies must be tailored to the type of heteromer being targeted. For physical heteromers, bitopic ligands that simultaneously engage orthosteric and allosteric sites within the heteromeric interface offer the potential for highly selective modulation of the receptor complex while sparing monomeric receptors [11]. Allosteric modulators that bind to the composite binding pocket formed at the protomer–protomer interface represent another promising approach. In contrast, functional heteromers, which may not form stable physical complexes, may be better addressed through combination therapies using selective ligands for each protomer, or through biased ligands that preferentially activate specific signaling pathways downstream of the functionally interacting receptors [44]. This distinction underscores the importance of rigorous characterization of heteromer type before embarking on drug discovery campaigns.

Discussion and Outlook

The transition from a monomeric to a heteromeric understanding of GPCR biology represents more than an incremental refinement of receptor theory; it constitutes a paradigm shift with the potential to reinvigorate neuropsychiatric drug discovery. By recognizing that drug targets in the CNS are not single genes but context-dependent macromolecular assemblies, we open avenues for achieving anatomical selectivity, signaling precision, and therapeutic efficacy, goals that the monomeric model has long promised but consistently failed to deliver. The striatal A2AR-D2R heteromer illustrates that motor adverse effects and antipsychotic efficacy can be pharmacologically uncoupled by targeting receptor complexes rather than monomers. The prefrontal 5-HT2AR-mGlu2R heteromer reveals that serotonergic and glutamatergic hypotheses of schizophrenia are not competing frameworks but complementary facets of a single molecular interface. Higher-order assemblies such as A2AR-D2R-mGlu5R suggest that the relevant unit of pharmacology may extend to multi-receptor hubs. The sigma-1 receptor further illustrates that the repertoire of modulatable targets extends beyond canonical seven-transmembrane receptors.

The field must also contend with legitimate methodological concerns. Heteromer detection remains technically demanding: co-immunoprecipitation and resonance energy transfer can yield false positives, and dependence on recombinant overexpression systems has led some heteromeric claims to be questioned when tested in native tissue. The observation by Delille and colleagues that 5-HT2AR–mGlu2R cross-talk is not invariant across cellular contexts is instructive rather than discouraging; it identifies the cell-type- and assay-dependent conditions under which heteromer-specific pharmacology becomes detectable. A second caveat concerns stoichiometry and dynamics: the physiological abundance of a given heteromer, its temporal stability, and its confinement to specific neuronal subpopulations remain poorly quantified for most complexes. Without such quantitative grounding, “heteromer-targeted” drug discovery risks becoming a purely conceptual exercise. A third concern is the distinction between direct allosteric modulation and indirect functional cross-talk: not every coordinated response between two receptors implies a physical heteromer, and rigorous evidence must satisfy accepted criteria for physical association, a unique functional property, and in vivo relevance.

We propose three priorities. First, establish native-tissue validation standards for each candidate heteromer, including endogenous co-expression, proximity-based interaction evidence, and functional readouts in primary neurons. Second, develop assays that report heteromeric state selectively, rather than relying on bulk measures that cannot distinguish monomeric, homodimeric, and heteromeric contributions. Third, integrate circuit-level and behavioral validation early in the discovery pipeline, so that pharmacological selectivity translates into meaningful changes in physiology and behavior. Meeting these standards will determine whether the heteromeric paradigm moves from compelling molecular biology to a genuinely new class of neuropsychiatric therapeutics. The requisite tools, advanced imaging, single-particle cryo-EM, genome editing, and spatially resolved omics are now converging. The remaining challenge is to apply them to heteromers under native physiological conditions rather than in isolation.

Author contributions

PR, YF-L, and JY-W conceptualized the perspective. PR and JY-W performed the literature synthesis and drafted the manuscript. JY-W and YF-L revised the manuscript and takes responsibility for the integrity of the work. All authors read and approved the final version.

Funding

This work was supported by grants from the Shandong Provincial Natural Science Foundation (No. ZR2025QC813), the National Natural Science Foundation of China (No. 82602976), the National Key Research and Development Program of China (STI2030-Major Projects, No. 2021ZD0200900), the Innovative Drug Research and Development - National Science and Technology Major Project (No. 2026ZD1807500), and Research Incubation Fund of Shandong Provincial Hospital (2024FY069). The funding sources had no involvement in study design, data collection, analysis, writing, or the decision to submit.

Declaration of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.