The superfamily of G protein-coupled receptors (GPCRs), characterized by a similarity in structure consisting of seven transmembrane domains, bind a wide variety of ligands that range from small biogenic amines and lipids to large complex proteins . Upon ligand binding a conformational change occurs in the GPCR, leading to activation of heterotrimeric G proteins via a catalytic exchange of GTP for GDP on the a subunit and dissociation of the a subunit from the Py complex . The free a subunit and the Py complex modulate the activity of a variety of effector proteins, resulting in alterations in second messenger molecules or alterations of cell physiology and/or signal transduction that lead to the cellular response. These effector proteins include adenylyl cyclase, phospholipase CP, G protein-coupled Ca2+ and K+ channels, phosphatases, sodium/hydrogen exchangers, and the mitogen activated protein kinase (MAP kinase) signal transduction pathways. First recognized in mammalian cells, GPCR mediated signal transduction pathways have functional homologs in evolutionarily distant organisms like insects, nematodes, plants, and yeast. Recent studies demonstrate that yeast is likely to be a useful model system to study components of the GPCR signaling pathways because of the high level of conservation between the elements of the yeast pheromone response pathway and mammalian GPCR-coupled MAP kinase signaling systems. Advances in GPCR expression in yeast and coupling to the pheromone response pathway suggest that this approach may be particularly useful in examining aspects of structure and function . Furthermore, haploid yeast cells can be altered by the introduction of specific mutations and reporter gene constructs that make them useful as host cells for the develop ment and implementation of sensitive HTS assays designed to identify novel ligands that modulate the activity of the GPCRs [66-68].
Haploid S. cerevisiae cells detect the presence of cells of opposite mating type through binding of peptide mating pheromones to G protein-coupled receptors. Thus, a cells secrete a factor and express the a factor receptor, Ste2p, while a cells secrete a factor and contain the a factor receptor, Ste3p [69,70]. When a cells and a cells come into close proximity, mating pheromone is detected by receptors, which initiates the mating process by activating intracellular hetero-trimeric G proteins. Dissociation of the a subunit, Gpa1p, from the complex of P (Ste4p) and y (Ste18p) subunits allows the Py complex to activate downstream elements of the pheromone response pathway. Ste20p, a p20 activated kinase (PAK) homolog, stimulates a MAP kinase cascade that consists of the sequential activation of Ste7p (MAP kinase kinase or MEK), Ste11p (MEK kinase), and the MAP kinases Fus3p and Kss1p [71,72]. Upon activation of the pathway, cells undergo a series of changes that prepare the yeast cell to mate with a cell of the opposite mating type. These changes include cell cycle arrest, activation of transcription of pheromone-responsive genes, and formation of mating-related cell structures.
Elements of the yeast pheromone response pathway are remarkably similar to mammalian GPCR signaling systems. This similarity has proved useful for analysis of mammalian GPCRs and G proteins, since yeast GPCRs and G proteins may be functionally replaced with their homologous mammalian counterparts. The yeast system permits analysis of these proteins in isolation, which is not possible using other expression systems. It is predicted that at least 400 GPCR genes may be present in the human genome, and this estimate is as high as 1000 or more if odorant and pheromone receptors are included . Since only about 300 GPCRs have had their cognate ligands identified, a large number of orphan GPCRs for which cognate ligands are not known remain to be characterized. The yeast system can be a valuable tool for the analysis of these receptors, both to study their structure and function and to identify ligands that modulate their activity.
Early studies indicated that heterologous expression of GPCRs in S. cerevisiae and other fungal cells resulted in the presence of functional antagonist binding sites in membrane fractions. King et al. reported that the activation of a hetero-logous mammalian P-adrenergic receptor expressed in yeast could be coupled to the pheromone response signal transduction pathway, suggesting that this yeast expression approach could be successfully employed . The coupling was dependent on coexpression of a mammalian Gas protein in yeast cells lacking the endogenous G protein a subunit, Gpa1p. Binding of the P-agonist, isoproterenol, resulted in activation of the pheromone response pathway, including expression of a pheromone-responsive reporter gene, apparent cell-cycle arrest, and formation of mating specific cell structures.
A number of alterations introduced into the yeast expression system were required to increase its usefulness and flexibility in HTS applications and allow for genetic selections in the presence of agonist [66,75,76]. The terminal cell-cycle arrest response of haploid yeast cells to mating pheromone was eliminated by deletion of the FAR1 gene, which encodes a negative regulator of G1 cyclins and is thought to serve as the primary interface between the pheromone-response pathway and the cell-cycle regulatory machinery [77,78]. Agonist stimulation of farl mutant cells results in activation of the pathway and transcription of phero-mone-responsive genes without affecting the cell's ability to grow and divide. A second important modification was introduced to enable yeast cells to grow only in response to an agonist. A pheromone-responsive reporter gene was constructed by placing the gene encoding His3p, an enzyme required for histidine biosynthesis, under the control of the pheromone induced FUS1 promoter . Hence his3 far1 mutant yeast cells will grow on media lacking histidine only when agonist is applied to the cells and the GPCR is activated (Fig. 1).
Additional changes have been made to improve the sensitivity of the expression system, including the elimination of desensitization pathways that promote recovery from cell-cycle arrest by reducing the signal transmitted through the pheromone-response pathway. One desensitization pathway, which is induced in response to chronic pheromone stimulation of Gpalp, allows cells to adapt and continue to grow in the presence of pheromone. This response is mediated by Sst2p (supersensitive), a member of the RGS (regulator of G protein signaling) family of GTPase activating proteins that play an important role in the desen-sitization of GPCR signaling pathways . Yeast cells lacking Sst2p exhibit pheromone hypersensitivity and are unable to recover from pheromone induced cell-cycle arrest. A second desensitization response, initiated by the pheromone receptors themselves, acts via a poorly understood mechanism to reduce agonist induced signaling [81,82]. In yeast cells optimized for HTS, deletion of the sst2 and ste2 genes in MATa cells serves to increase greatly the sensitivity of the yeast cell response to GPCR agonists.
The state of the art in pharmaceutical drug discovery requires mechanism-based screening assays of high selectivity, sensitivity, and throughput. This is achieved by using cloned gene targets in a robust and miniaturizable system with low background (i.e., a high signal-to-noise ratio). Given these criteria, yeast strains that functionally express heterologous GPCRs are ideal for HTS applications. The diversity of GPCRs successfully expressed in yeast so far indicates that the technology will have applicability to a broad range of therapeutic targets. Beyond this, the yeast system has also proven to be flexible with regard to important practical considerations in pharmaceutical drug discovery. For instance, yeast screening assays can be performed on agar plates or in microtiter trays (liquid format), each with specific advantages. Test compounds spotted on yeast cells imbedded in agar will diffuse radially, thus effectively displaying a response over a large concentration gradient. On the other hand, liquid assays can be performed robotically in 96-well or higher formats with fixed test compound concentrations. Another practical consideration is the ability of a screen to test compounds accurately from different sources: organic chemicals dissolved in solvents to natural extracts to synthetic combinatorial libraries. Finally, screens must be designed to identify antagonists vs. agonists at a particular target site. The choice of reporter genes in the yeast GPCR system has broadened its utility for such considerations. For instance, the HIS3 reporter gene cannot be used in screening natural products, many of which contain histidine. Here, an antibiotic resistance reporter gene (e.g., G418R) would be used . To screen for GPCR antagonists, a CANI (canavanine sensitivity) reporter induces a toxic response to an agonist until blocked (rescued) by an antagonist .
An example of a successful yeast GPCR assay is the adenosine A2a receptor assay . For this target, with potential uses for agonists in both agriculture and medicine, known purine compounds were difficult and expensive to synthesize. Using an agar plate assay, the A2a receptor expressed in yeast (coupled to the endogenous Ga subunit, Gpalp) was quickly screened at low cost against a conventional compound library. Of 55,000 compounds tested, 44 hits (0.08%) were retested, of which 12 (0.02%) were positive in a radioligand competition binding secondary assay, indicating that the active compounds bound to the receptor at the known agonist binding site. Among these were nonpurines with submicromolar binding constants, and up to 100 fold selectivity for A2a vs. A1 receptors. In this screen, the rapidity of screening (<1 week) and its low cost (about 1 cent/sample disposables; <10 cents/sample labor/overhead) contributed to the program's success.
Another successful GPCR yeast assay is the somatostatin subtype 2 receptor (SST2) assay. Here antagonists, with potential uses in both agriculture and medicine, were sought. Somatostatin (SRIF, a 14 amino acid peptide hormone) is inhibitory, causing reduced cAMP levels via interaction with Gai. Follow-up assays useful for demonstrating the effect of somatostatin analogs, e.g., inhibition of cAMP accumulation, are difficult to perform because the system must be artificially stimulated before SRIF activity can be detected. Since the yeast assay responds directly to SRIF, agonists can be measured directly. Thus, antagonists were efficiently detected using agar plates containing 10 nM SRIF, where zones of growth inhibition were measured [75,83]. This assay proved to be useful in a conventional analog program, where small, subtype selective peptides were tested individually for agonist and antagonist potency . The yeast assay was also very powerful for screening complex mixtures that required extreme sensitivity to detect activity. A synthetic combinatorial random peptide library containing 160,000 peptides per sample would be virtually impossible to test using conventional assays due to lack of sensitivity. Using the yeast assay, however, a combinatorial library of this kind was successfully screened in a stepwise, iterative fashion. The first round of screening resulted in faint, but discernible, zones of inhibition for antagonists. Each successive round of screening gave incrementally stronger signals, and served to further define the structure of a lead peptide. The final result of these studies was a novel antagonist peptide that showed potent effects in vitro and in vivo .
These approaches can be implemented for the increasingly large number of heterologous GPCRs that couple directly to the pheromone-response pathway through the yeast G protein alpha subunit. The coupling of receptors that do not functionally interact with the yeast a subunit will be facilitated by coexpression of cognate mammalian Ga proteins. In addition, the Ga proteins can be modified by the introduction of mutations that improve functional coupling to the GPCR and Py subunit complex. This type of analysis is facilitated by the recently deter mined crystal structure, which can be used to identify domains of the protein and individual residues that are critical for interaction [85,86].
Agonists that interact with orphan GPCRs can also be identified using assays based on a growth phenotype. Yeast strains expressing orphan GPCRs can be screened with selected agonists, compounds from chemical files and/or combinatorial libraries, to identify surrogate ligands that allow cell growth on selective media. This approach was used to identify an agonist for edg-1, an orphan GPCR thought to be a member of the lysophosphatidic acid (LPA) receptor family . Ligand binding specificity was difficult to demonstrate in cultured mammalian cells due to the ubiquitous presence of LPA receptor subtypes. In this case a yeast expression system was used to demonstrate that phosphatidic acid acts as a high-affinity agonist. This analysis was made possible because of the distinct advantage of expressing a single GPCR subtype in a yeast cell and the ability to couple the receptor to the yeast pheromone-response pathway. As a potentially useful alternative to screening compounds applied to the cell, yeast expression libraries designed for secretion of random small peptides were constructed. The plasmid library is expressed in yeast along with a GPCR to identify cells that express agonists or antagonists of the receptor being expressed. An autocrine loop is established that results in the growth of cells that express an active peptide . Using this scheme, novel agonists and antagonists of the yeast a-mating pheromone receptor  and surrogate ligands for the orphan GPCR, FPRL1 were identified .
D. GPCR Analysis and Screening with the Yeast Two-Hybrid System
Certain classes of GPCRs, including secretin and growth hormone releasing hormone receptors, possess ligand binding determinants in a large N-terminal extracellular domain that can be used in the yeast two-hybrid system to examine GPCR/ligand interactions. The interaction of the GHRH receptor N-terminal domain with GHRH was evaluated using this system by fusing the complete N-terminus of the human GHRH receptor to one half of the two-hybrid Gal4p protein, and fusing GHRH to the other half . In the two-hybrid system, the expression of a reporter gene that allows growth on selective media occurs only when a protein-protein interaction is formed (see section below). The proteinprotein interaction formed between the GHRH receptor domain and GHRH was sufficient to promote growth of yeast cells on selective media, and this interaction was disrupted when specific mutations known to interfere with GHRH binding were introduced into GHRH . This approach may be extended to other members of the secretin class of GPCRs and provides a potentially useful alternative method for investigating receptor-ligand interactions, as well as for high-throughput screen design.
The heterologous yeast expression system for GPCRs can also be used to examine structure-function relationships that are difficult to study genetically using other systems, including elucidation of ligand binding sites and GPCR interactions with heterotrimeric G proteins, agonist activation of GPCR activity, and the response to surrogate agonists. Mutations in GPCRs or their signaling pathways can be identified using genetic selections based on the growth phenotype. Constitutively active and dominant-negative mutants of the yeast a factor receptor were identified using genetic approaches that could be employed to identify mutants of heterologous GPCRs [91,92]. In addition, the yeast expression system was used to identify amino acid residues involved in melatonin receptor activation of hetero-trimeric G proteins . Similar genetic approaches should be useful for the dissection of the interactions within the heterotrimeric G protein complex, with RGS proteins, and with downstream effector enzymes. Yeast cells also express adenylyl cyclase (CYR1), phospholipase C (PLC1), high-affinity potassium uptake transporter (TRK1), and potassium channel (TOK1). In principle, entirely synthetic signal transduction pathways can be constructed in yeast cells by complementing conditional phenotypes with the corresponding mammalian genes.
V. MACROMOLECULAR INTERACTION TARGETS: YEAST TWO-HYBRID SYSTEMS
A therapeutic target is most often envisioned to be a receptor or an enzyme, where the therapeutic agent constitutes a surrogate agonist, antagonist, or active site inhibitor. However, targets can also be proteins that work in concert with other factors in a complex, in a cascade where other enzymes are the target of enzyme activity, or as integral membrane components transducing a signal. The therapeutic agent can then be viewed as a small molecule that leads directly or indirectly to alterations in protein-protein interactions. Can small molecules that affect interactions of large proteins be detected? Nature offers many examples of such molecules. Benzimidazoles, among other antifungal and antitumor compounds, exert their action by disruption of tubulin protein assembly . Small molecules can also promote association of proteins; for example, taxol and related compounds have this effect on microtubules . Binding of the drugs cyclospo-rin, FK-506 , and Rapamycin  to their corresponding immunophilins facilitates the interaction with target proteins.
Yeast two-hybrid technology  has evolved into a standard method for detecting protein-protein interactions for gene discovery. It was realized early on that the same methodology could be applied to detect a variety of intracellular interactions and, by extension, for compound discovery. Induction of dimeriza-
tion [99,100], small molecule-protein interactions , and disruption of known protein-protein interactions [102,103] have been demonstrated. In addition, higher order protein interactions , as well as protein-RNA [105,106] and protein-DNA [107,108] interactions can be studied with two-hybrid or similar techniques. A recent example of a small molecule inducing protein-protein interactions is relevant to receptor surrogate agonist discovery. In a screen for activators of granulocyte-colony-stimulating factor receptor, a small nonpeptide molecule that presumably mimics the receptor dimerization characteristics of the peptide hormone was detected . Alarcon and Heitman  showed that reporter expression induced by hybrid FKBP12 (Fpr1p) and aspartokinase (Hom3p) interactions can be reversed by application of FK-506 to the cells. Young et al.  identified small molecule inhibitors of human N-type calcium channels by screening for disruption of hybrid proteins encoding portions of the a1B and the P3 subunits of the channel.
A large number of variations on the two-hybrid theme have emerged in recent years, reviewed in Refs. 111 and 112, and any of these systems could be amenable to compound discovery. One consideration for choosing among these technologies is the characteristic of the compound(s) one desires to detect. If the drug target normally exists in the nucleus, then two-hybrid or one-hybrid screens will be appropriate. But examples exist for nuclear two-hybrid interactions where one might not have expected success. For example, plasma membrane receptors and their cognate soluble ligands have been shown to interact in this system . If there is concern that the yeast nucleus is a poor environment for the particular target interaction, there are a number of techniques that do not depend on transcription for a detectable output. Alternative methods exist for detection, and disruption, of interactions occurring in the cytoplasm  or in membranes [115,116]. The latter may be important for discovery of a surrogate ligand for a cell surface receptor where cell permeability of such a compound is not expected, or is not desirable. Alternative protein interaction screens include mammalian and bacterial two-hybrid systems [117-119] and a recent report using fluorescence resonance energy transfer between green fluorescence protein (GFP) fusion proteins .
One of the most important considerations for screen design, whether of the two-hybrid type, the conventional enzyme-based type, or other technology, is the method for detection of a hit. Microbial and mammalian cell-based protein interaction screens have been designed using many available reporting systems. The ease of genetic manipulation of yeast allows one to take advantage of a variety of reporting systems as simple as rescue of growth inhibition. Common fluorescence assays, designed for high sensitivity, can also be used in the yeast systems [121,122]. In addition to the reporter itself, the sensitivity of the tran-scriptional (or other) inducing system should be such that weakly or moderately active compounds will still be detected. See Ref. 123 and the review by Golemis and Brent for a discussion of this topic .
One of the most useful reporting systems one can take advantage of is the technique of screening for disruption of interactions by ''reverse'' two-hybrid techniques [125-127]. These systems rely on the well-known genetic methodology of using counter-selectable markers (Table 2). Using interaction trap and other similar two-hybrid target discovery protocols puts one only a step away from having a workable screen. After validation of the interaction, cells are transformed with a selectable or counter-selectable marker and the screen is ready to
The main utility of the yeast two-hybrid technology has been in gene discovery. In the arena of drug screening, this usually means target discovery. But in a broader definition of therapeutics, including gene therapy or transgenic crop generation, protein interaction screens are a direct way of discovering the therapeutic gene.
Was this article helpful?