Advances in yeast synthetic biology for human G protein–coupled receptor biology and pharmacology

Abstract

G protein–coupled receptors (GPCRs) are the largest family of transmembrane receptors in humans. Over 800 GPCRs regulate the (patho)biology of every organ, tissue, and cell type. Consequently, GPCRs are the most prominent therapeutic targets in medicine. Although over 30% of current U.S. Food and Drug Administration-approved drugs target GPCR signaling, most receptors remain understudied and therapeutically underutilized. Challenges include an incomplete understanding of GPCR signaling, pharmacology, structural biology, and the multiplicity of endogenous GPCR ligands, in addition to a scarcity of biological and pharmacological tools for elucidating GPCR-mediated cellular processes beyond initial signaling events. Various mammalian, insect, and yeast cell models currently address some of these needs. Here, we review recent advances in yeast synthetic biology that are helping to catalyze new and unexpected conceptual and technical breakthroughs in GPCR-based medicine and biotechnology.

Advantages of human G protein–coupled receptor studies in yeast models

Overview of G protein–coupled receptor signaling

A cell’s ability to sense and respond to environmental stimuli is essential to all life. Over a billion years ago, the seven transmembrane (7TM)-fold emerged as an elegant and prototypical solution to this need. Early prokaryotes used the 7TM architecture of bacteriorhodopsin to detect and convert light energy to chemical energy [1,2]. Since then, the 7TM fold has evolved to meet the increasingly complex demands of higher organisms, from pheromone sensing in yeast to neurotransmission in humans. Over 800 GPCRs are ubiquitously expressed in humans to mediate myriad (patho)physiological processes, including metabolism, inflammation, immune responses, and nociception [3]. Consequently, 34% (445) of FDA-approved drugs regulate 108 receptors, making GPCRs the most common drug target in the clinic. Given only 12% of GPCRs are currently druggable, they continue to be a primary focus of pharmaceutical development, with 321 GPCR drugs currently in clinical trials [4].

As illustrated in Figure 1, an activated GPCR interacts with and regulates three canonical signal transducers (G proteins) and desensitizers (G protein–coupled receptor kinases [GRKs] and β-arrestins, β-Arr1 and β-Arr2). With G protein signaling, receptor activation causes the exchange of Guanosine diphosphate (GDP) for Guanosine triphosphate (GTP) on the Gα subunit, driving dissociation from the Gβγ heterodimer. Both subunits can control diverse downstream signaling events that ultimately regulate second messenger levels (e.g. cyclic AMP and Ca²⁺) and other effectors (e.g. small GTPases and ion channels). Following receptor stimulation, GRKs phosphorylate active conformations of many GPCRs on their intracellular loops and C-terminal tails. These phosphorylated receptor states recruit β-arrestin proteins that orchestrate receptor-dependent desensitization, internalization by endocytosis, recycling to the cell membrane, or degradation. β-arrestin recruitment to activated GPCRs also leads to kinase scaffolding , suggesting arrestin can both positively and negatively regulate GPCR signaling beyond initial signaling events. Other noncanonical aspects of GPCR signaling are emerging. For example, it is now known that GPCR signaling occurs from intracellular membranes, where the local environment (e.g. pH) and effector repertoire often vary from that at the cell surface [5–9].

Figure 1.

Overview of canonical GPCR signaling in humans. GPCRs interact with three canonical signal transducers (G proteins) and desensitizers (GRKs and β-arrestins). GPCR activation by extracellular ligands causes the exchange of GDP for GTP on the Gα subunit, driving dissociation from the Gβγ heterodimer. This initiates an intracellular signaling cascade via second messengers, such as cyclic AMP. Following receptor stimulation, GRKs phosphorylate active conformations of many GPCRs. This promotes the recruitment of β-arrestin proteins that orchestrate receptor-dependent desensitization and internalization, leading to receptor recycling or degradation.

From complex to relatively simple models of G protein–coupled receptor signaling

Given that any human cell has the potential to express hundreds of GPCRs and many Gα (16), Gβ (4), and Gγ (12) subtypes (Figure 2a), the combinatorial nature of GPCR signaling networks adds to the inherent complexity of GPCR signaling. Consider a modest example. In theory, a human cell expressing 50 GPCRs and 10 Gα, 2 Gβ, and 4 Gγ subunits can assemble 4000 different GPCR-Gαβγ quaternary complexes. Including other known GPCR effectors, such as GRKs and β-arrestins, adds further combinatorial complexity. Although the convoluted nature of GPCR signaling networks is essential to cell biology and systems physiology, it may hinder more basic, high-throughput experiments aimed at GPCR ligand and drug discovery. As we will discuss, human GPCR studies in yeast avoid these complexities to provide highly interpretable insight.

Figure 2.

Modeling human GPCR signaling in yeast. GPCR signaling is far more complex in humans (a) than yeast (b). Humanized versions of the yeast MP (c) link human GPCR signaling to transcriptional reporters via a chimeric Gα subunit. (d) Using CRISPR, human GPCRs, chimeric Gαs, transcriptional reporters, and genetically encoded ligands can be integrated into the yeast genome to build stable, barcoded strains for various applications. See text for further details.

From an evolutionary perspective, canonical GPCR signaling arose in eukaryotes and increased in complexity from lower to higher organisms. In this sense, microorganisms, such as yeast, exemplify GPCR signaling, version 1.0. Within the fungal kingdom, some yeast species, such as Saccharomyces cerevisiae (hereafter yeast), exist in diploid and haploid states as single cells. In the haploid state, they can reproduce asexually by budding or sexually by combining with the opposite mating type (MATα or MATa). A GPCR-regulated mating pathway (MP) controls the latter (Figure 2b). In the MP, a single GPCR (Ste2/Ste3 in MATα/MATa cells) is agonized by a peptide hormone secreted by the opposite mating type (α-factor or a-factor) [10]. Intracellular Gα (Gpa1), Gβ (Ste4), and Gγ (Ste18) subunits amplify this signal by triggering a MAP kinase cascade to initiate a mating program comprising ~200 pheromone-responsive genes [11]. Notably, in contrast to GPCR signaling in higher organisms, the Gβ subunit Ste4, as opposed to the Gα subunit Gpa1, controls the primary signaling events in the MP (Figure 2b,c).

Advantages of using the yeast mating pathway to study human G protein–coupled receptor signaling

Despite the evolutionary divergence of yeast and humans occurring ~1 billion years ago, the simple yeast MP has provided much insight into the fundamentals of human GPCR signaling. Discoveries such as the MAP kinase cascade, regulators of G protein signaling (RGS proteins), and endosomal signaling are prime examples [9,12–14]. The MP has also provided an invaluable platform for studying human GPCRs in humanized yeast strains [15] (Figure 2d). Although the details vary, these strains are typically engineered in three steps (see Box 1) and assayed in liquid media. Because these strains proliferate rapidly (typical doubling time of 90 min) and do not require biosafety cabinets, they can be assayed in various lab settings and instruments to accelerate the discovery cycle. Finally, the reduced signaling complexity and cross-talk afforded by the MP are the ultimate advantages of the humanized yeast model, as it is the only in vitro pathway that provides a 1:1:1 correspondence between a single, isolated human GPCR, a single Gα subtype, and robust transcriptional reporting.

Box 1. Engineering humanized yeast reporter strains for GPCR studies.

First, introduce a fluorescent (GFP or mTurqoise2), luminescent (NanoLuc), or growth-selective (His) transcriptional reporter in place of a strongly expressed pheromone-responsive gene, such as FIG1 or FUS1. Second, engineer a yeast-human Gα chimera in which the last five residues of a human Gα subunit replace the equivalent residues of Gpa1. Because some C-terminal protein sequences of the 16 human Gα subunits are degenerate, only 10 chimeric yeast strains are needed to cover all possible human GPCR–Gα coupling combinations. For strains built in the MATa background, deletion of the RGS protein SST2, cell cycle arrest factor FAR1, and endogenous GPCR STE2 results in strains with enhanced pathway sensitivity (SST2Δ) that avoid cell cycle arrest (FAR1Δ) and express only one GPCR (STE2Δ). Alternatively, an episomal plasmid or genome-integrated landing pad expresses a human GPCR that may or may not be codon optimized for yeast. Finally, MP refactoring can provide additional advantages [16]. Microplate readers, microscopes, FACS, and point-of-contact detection devices quantify activated GPCR–ligand interactions using individual/pooled strain formats [17–21]. See Figure 2b–d for additional details.

A key advantage of GPCR screening experiments in yeast is their extreme tolerance to various ligand/drug vehicle solutions and a wide range of pH values. Dimethyl sulfoxide, ethanol, and methanol vehicle concentrations between 0.1% and 10% are compatible with exploratory GPCR screening studies. Additionally, unlike human cells, yeast cells maintain a stable intracellular pH (pH_i) of 7.2 over a wide range (pH 2–7) of extracellular pH values (pH_e) [22]. This pH resilience holds for acute and chronic acid exposure. As discussed ahead, this feature has proven essential for advancing pH studies of human GPCR function and pharmacology in yeast [22–24]. Conversely, glucose limitation enables exquisite control of yeast pH_i between 6.0 and 7.2 while maintaining a constant pH_e of 7 [25,26]. This ability to set pH_i is essential to pH sensor discovery. The finding that yeast and human Gα subunits are pH sensors exemplifies this approach [26]. While this review focuses on GPCRs, the benefits of stably manipulating yeast pH_e and pH_i are broadly applicable and uniquely limited to the yeast cell model.

Finally, CRISPR/Cas9 genome editing has dramatically accelerated the speed and scale of yeast genetic engineering [27–29]. Unlike mammalian cells, which repair double-stranded DNA breaks (DSBs) primarily by nonhomologous end joining, yeast uses homologous recombination (HR). Decades of yeast genetic engineering have exploited HR to delete, mutate, and introduce genes in yeast. However, these traditional approaches have been limited in throughput by two factors: the exhaustion of selective auxotrophic markers and the infrequency of spontaneous DSBs near the target genome locus. The CRISPR/Cas9 system overcomes these barriers by providing a DSB at a precise genome locus using a plasmid carrying the single-guide RNA sequence, Cas9 endonuclease, and a renewable URA3 auxotrophic marker. Co-transforming the Cas9 plasmid with a DNA payload flanked by ~60 bp of genome-targeting homology implements the genome edit. The result is a stable yeast line with a scarlessly modified and barcoded genome ideally suited for multiplexed GPCR ligand and drug discovery studies [19].

Exemplary applications: conceptual and technical advances

From parallel to pooled screening to study G protein–coupled receptor ligand/drug multiplicity

GPCRs detect myriad endogenous ligands in humans, ranging from neurotransmitters to lipids. However, most biologically relevant ligand–GPCR interactions are yet unknown. Only about 40% of human GPCRs have at least one known ligand that (de)activates or (de-)enhances GPCR signaling by binding to primary ‘orthosteric’ and secondary ‘allosteric’ receptor sites [4]. Adding to this uncertainty is the likelihood that multiple ortho- and allosteric chemical inputs multimodally regulate most receptors. Ligand promiscuity, polypharmacology, and coincidence detection likely add further layers of complexity [19,22,30–33]. For these reasons, traditional screening of individual GPCRs is unlikely to reveal their ligand multiplicity. Instead, pooled experiments combining many GPCRs and ligands are more likely to advance the field and bring wet-lab experimental throughput closer to the throughput of in silico screening approaches.

Recent studies are beginning to address the need for pooled GPCR profiling technologies. The DCyFIR (Dynamic Cyan Induction by Functionally Integrated Receptor) method exemplifies these efforts [19]. In the DCyFIR platform, yeast strains are barcoded with a genome-integrated GPCR and chimeric Gα subunit and express a fluorescent reporter, mTurquoise2, in response to GPCR activation. Following ligand treatment, activated strains are collected by fluorescence-activated cell sorting (FACS), and the identity of each GPCR–ligand pair is determined by quantitative PCR. DCyFIR screening of known GPCR ligands and a library of 320 human metabolites against a combined pool of 300 GPCR–Gα combinations identified 36 new GPCR–ligand interactions for 15 unique receptors. As anticipated, pooled profiling identified instances of ligand promiscuity and polypharmacology. For example, the inflammatory metabolite kynurenic acid, a known agonist of GPR35, was discovered to also polyagonize the pharmacologically dark HCAR3 [19,34]. Given the low expense and high scalability of such pooled approaches, they will likely feature prominently in continued efforts to map the human GPCR–ligand interactome, build GPCR-based biosensors, and profile increasingly expansive libraries of synthetic and natural products.

Putting the pH into pHarmacology

Although pH regulates GPCRs in various cellular and (patho)physiologic contexts, studying the effects of pH on GPCR signaling in human/mammalian cell models is challenging. Recent yeast models now provide this previously inaccessible biological, biophysical, and pharmacological insight. These methods build on the discovery that yeast maintains a stable pH_i in increasingly acidic media [22], a key conceptual and technical breakthrough. For example, pH-dependent GPCR signaling from the yeast cell membrane is an apt model of human GPCR endomembrane signaling. Experimentally varying yeast media pH from 7 to 5 mimics the lumenal acidification of maturing early, middle, and late endosomes. The outward orientation of a human GPCR in the yeast cell membrane exposes the receptor’s ligand-binding pocket to extracellular pH and the intracellular G protein–binding to a constant intracellular pH. In humans, the pH exposure of an endocytosed GPCR has the same directionality, with the ligand-binding pocket facing the acidified lumen and the G protein–binding pocket facing the neutral cytosol. This conceptual framework mirrors other (patho)biologic scenarios/processes, including neurotransmission, inflammation, ischemic vasculature, and tumor microenvironments. It enables the identification and detailed characterization of receptors (in)capable of endomembrane signaling and receptor–drug interactions (in)sensitive to (patho)physiologic pH changes.

Within the past several years, yeast pH models have catalyzed two significant advances in GPCR signaling [22,23] and motivated the inclusion of pH considerations in other GPCR work [24]. For example, a yeast pH model has been used to show that the proton-sensing receptors GPR4, GPR65, and GPR68 evolved the ability to sense pH by acquiring a triad of buried acidic residues [23]. Lowering pH causes protons to bind these acidic residues to elicit signaling. Additionally, this work refutes the long-held belief that extracellular histidine residues activate proton-sensing GPCRs and shows that GPR132, a related GPCR widely thought to be a proton sensor, lacks the conserved acidic triad and, thus, the ability to sense pH. Finally, the 400+ CRISPR-engineered strains built for this effort, which covered all known GPR4, GPR65, and GPR68 point mutations in the literature, further exemplify the high-throughput capabilities of yeast-based models for GPCR studies and receptor engineering.

Another breakthrough catalyzed by a yeast pH model is the discovery of proton-gated coincidence detection [22]. In contrast to proton sensing by GPR4, GPR65, and GPR68, the direct activation of GPCRs by low pH and proton binding is relatively rare. Alternatively, changes in extracellular pH may indirectly modulate GPCRs by regulating receptor–ligand interactions through ligand–pH coincidence detection. Recent studies in yeast confirmed this hypothesis by showing that proton signals regulate most GPCRs by synergistically modulating ligand binding and receptor activation [22]. This work tested a panel of 28 GPCRs covering 280 possible GPCR-Gα combinations to show that proton gating positively and negatively regulates all modes of GPCR pharmacology — ligand efficacy, potency, and cooperativity — often in unexpected ways. For example, antagonists for the adenosine receptor 2A (A2A) were increasingly ineffective at lower pH values. Because traditional ligand/drug screening campaigns are limited to neutral pH, this hidden feature of A2A antagonism likely extends to other approved drugs. The latest yeast pH models can unmask these lost insights on an unprecedented scale.

Proton-gated coincidence detection has several other features that will guide the development of GPCR therapeutics and pharmacological tools. The general trend is that the activity of most GPCRs is lower as pH is decreased from 7 to 5. However, counterexamples, such as LPAR4 and FFAR2, signal more at increasingly acidic pH. In both scenarios, pH regulation can be Boolean-like or graded. Boolean-like proton coincidence detection cooperatively turns GPCR signaling on/off at low/high pH and vice versa. In contrast, graded proton coincidence detection is less cooperative but still highly sensitive to pH. Both modes of pH regulation have implications for cell biology and ligand/drug design/discovery. Again, using the example of endocytosis, Boolean-like and graded pH regulation will help dictate the effects of lumenal endosomal pH on GPCR desensitization, recycling, and degradation. Furthermore, exploiting Boolean-like and graded pH regulation for drug discovery enables the design of pH-intelligent ligands that (in)activate GPCRs at precise pH values. Proof-of-concept for this approach lies in the first-generation nonaddictive μ-opioid agonists designed to function exclusively in acidified inflammatory zones caused by chronic analgesia [35,36], and the discovery that the benzodiazepine drug lorazepam is a nonselective positive allosteric modulator of acid-activated GPR68 [37]. As we look to the future, yeast pH models and the phenomenon of GPCR proton gating provide an actionable path forward for building and testing pH-intelligent modulators for many more human GPCRs, opening a new frontier in GPCR biology and pharmacology.

The growing diversity of yeast-based approaches related to human G protein–coupled receptor signaling

Since its early development and use in GPCR drug discovery [15,38], the diversification of the yeast MP has spawned a growing number of applications in medicine and biotechnology. Figure 3 illustrates a representative list of recent examples. Various GPCR inputs and readouts are now widely established. Principal input examples include small molecules, such as protons, natural compounds, synthetic drugs, metabolites, and lipids, and exogenously added or genetically encoded and secreted peptides and protein ligands (Figures 2d and 3). Standard readouts include cell growth and luminescent/fluorescent transcriptional reporters. In the case of fluorescent reporters, signal detection may depend on instrument availability, as microscope and FACS readouts typically offer more sensitivity than microplate readers.

Figure 3.

Representative receptor inputs, readouts, and applications. Receptor inputs are categorized as ions, small molecules, and peptides. The ‘all’ category covers individual studies that include all three inputs. Pathway readouts are categorized as absorbance, luminescence, and fluorescence. The ‘all’ category covers individual studies that include all three readouts. Triple dashes indicate inputs and readouts that appear absent in the literature.

One of the earliest applications of the yeast MP was GPCR biosensor design. In 2007, the Roth group used directed evolution to create the first Designer Receptor Exclusively Activated by Designer Drugs [39]. In this work, they evolved human acetylcholine receptors to selectively sense a biologically inert compound, clozapine-N-oxide [39]. These DREADDs are now widely used in the community to chemogenetically stimulate tissue- and cell-specific signaling through specific G protein pathways. This tradition of building GPCR biosensors in yeast continues today. Recently, the Jensen group developed chemical biosensors for serotonin receptors, and the Isom, Ellis, and Kampranis groups have created biosensor strains for cannabinoids [19,20,40] and various metabolites [19]. The Peralta Yahya lab has also offered a perspective on the potential of building chemical sensors using olfactory receptor chassis [41]. As the field grows, it will continue to provide innovative approaches addressing the most vexing challenges in GPCR biology, pharmacology, and GPCR-based biosensor technologies.

Outlook

This review aimed to concisely explain the conceptual and technical framework and rationale that underlie continuing translational studies of human GPCRs in yeast models. Based on our efforts and the work of others, we estimate that 50– 100 human GPCRs express functionally in yeast. We anticipate that it will be possible to study most human GPCRs in yeast in the next decade and that other human GPCR signaling components, such as β-arrestins and GRKs, will be incorporated into more sophisticated models. Additionally, we anticipate that efforts to engineer GPCR-based chemical sensors will expand to include more exotic receptors drawn from various lower and higher organisms. Finally, we predict that the throughput of genetic engineering and scalability of yeast models will integrate with emerging artificial intelligence methods for protein and drug design to advance many aspects of GPCR biology and GPCR-driven biotechnologies.

Data Availability

No data were used for the research described in the article.