Protein–protein interactions and selection: yeast-based approaches that exploit guanine nucleotide-binding protein signaling

Authors


A. Kondo, Department of Chemical Science and Engineering, Graduate School of Engineering, Kobe University, 1-1 Rokkodaicho, Nada-ku, Kobe 657-8501, Japan
Fax: +81 78 803 6196
Tel: +81 78 803 6196
E-mail: akondo@kobe-u.ac.jp

Abstract

For elucidating protein–protein interactions, many methodologies have been developed during the past two decades. For investigation of interactions inside cells under physiological conditions, yeast is an attractive organism with which to quickly screen for hopeful candidates using versatile genetic technologies, and various types of approaches are now available. Among them, a variety of unique systems using the guanine nucleotide-binding protein (G-protein) signaling pathway in yeast have been established to investigate the interactions of proteins for biological study and pharmaceutical research. G-proteins involved in various cellular processes are mainly divided into two groups: small monomeric G-proteins, and heterotrimeric G-proteins. In this minireview, we summarize the basic principles and applications of yeast-based screening systems, using these two types of G-protein, which are typically used for elucidating biological protein interactions but are differentiated from traditional yeast two-hybrid systems.

Abbreviations
GAP

GTPase-activating proteins

GEF

guanine nucleotide exchange factor

GPCR

guanine nucleotide-binding protein-coupled receptor

G-protein

guanine nucleotide-binding protein

cyto

mutated yeast Gγ lacking membrane localization ability

MAPK

mitogen-activated protein kinase

M3R

M3 muscarinic acetylcholine receptor

mRas

mammalian Ras

RRS

Ras recruitment system

SRS

Sos recruitment system

Y2H

yeast two-hybrid

yRas

yeast Ras

Introduction

Protein–protein interactions have fundamental roles in a variety of biological functions, and are of central importance for virtually every process in a living cell. Hence, many methodologies for elucidating protein interactions have been developed during the past couple of decades. To investigate interactions inside cells under physiological conditions, especially, yeast would be a most typical organism, and various in vivo selection approaches are now available.

The budding yeast Saccharomyces cerevisiae is one of the simplest unicellular eukaryotes, and is often used as a eukaryotic model organism for cellular and molecular biology [1–5]. Yeast has several benefits, including the possession of eukaryotic secretory machinery, post-translational modifications, rapid cell growth, and well-established and versatile genetic techniques. Thus, it is also used to establish technologies with which to survey interactions of eukaryotic proteins. The yeast two-hybrid (Y2H) system, which was originally designed to detect protein–protein interactions in vivo by separation of a transcription factor into a DNA-binding domain and a transcription activation domain, is a typical representative of a yeast-based genetic approach [6], and numerous improved Y2H systems have been developed to overcome its potential problems [7–14]. The utility of Y2H systems has been demonstrated to varying degrees, involving analyses of comprehensive interactome networks [15–18], identification of novel interaction factors [19–22], investigations of homodimerization or heterodimerization [23–25], and the obtaining of conformational information [26–28]. Thus, yeast is definitely an attractive organism for analyzing the interactions of eukaryotic proteins.

Guanine nucleotide-binding proteins (G-proteins) are signaling molecules that are highly conserved among various eukaryotes, and that engage in a wide variety of cellular processes [3,29]. They switch from an inactive to an active state by exchanging a GDP molecule for GTP, and they return to the inactive state by hydrolyzing GTP to GDP. They are divided into two main groups: small monomeric G-proteins and heterotrimeric G-proteins [29]. Because eukaryotic yeast cells have both types of G-protein, but are not as complicated as higher eukaryotes, yeast has been used as the model organism for the study of G-protein machinery [30–32]. Much knowledge of G-protein signaling in yeast has been accumulated and used to study cellular processes, including protein interactions.

In this minireview series highlighting the methodologies for elucidating protein–protein interactions, the other two minireviews by K. Tomizaki et al. [33] and M. Umetsu et al. [34] deal with array based-technologies for detecting protein interactions in vitro, and constructive approaches to the generation of novel binding proteins on the basis of tertiary structural information, respectively. In this first minireview, we focus on and summarize the unique technologies used to exploit yeast G-protein signaling, which are commonly used for the exploration of biological protein interactions under physiological in vivo conditions but are distinguishable from conventional Y2H systems from a scientific and engineering perspective.

Ras signaling-based screening systems for protein–protein interactions

Small monomeric G-protein signaling in yeast

Small monomeric G-proteins, such as Ras and Ras-like proteins, are found mainly at the inner surface of the plasma membrane as monomers. They function as GTPases on their own, and are involved in controlling cell proliferation, differentiation, and apoptosis [29]. The Ras proteins are, in addition, necessary for the completion of mitosis and the regulation of filamentous growth [35]. In the yeast S. cerevisiae, growth and metabolism in response to nutrients, particularly glucose, is regulated to a large degree by the Ras–cAMP pathway [30,31,35]. Ras proteins activate adenylate cyclase, which synthesizes cAMP, and the increase in cytosolic cAMP levels activates the cAMP-dependent protein kinase, which has an essential role in the progression from the G1 phase to the S phase of the cell cycle.

Owing to their intrinsically slow GTPase and GTP–GDP exchange activities, Ras proteins are strictly controlled by two classes of regulatory proteins: GTPase-activating proteins (GAPs), and guanine nucleotide exchange factors (GEFs) [35]. RasGAPs, which act as negative regulators of Ras–cAMP signaling by accelerating hydrolysis of GTP to GDP on Ras proteins, can stimulate the GTPase activity of Ras proteins to terminate the signaling event. On the other hand, RasGEFs, which contain Cdc25p and Sdc25p in yeast, stimulate the exchange of GDP for GTP on Ras proteins. The stimulated RasGEFs activate the Ras–cAMP signaling pathway. Whereas Cdc25p is essential in most genetic backgrounds, Sdc25p is dispensable and is normally expressed only during nutrient depletion or in nonfermentative situations. Through its role in regulating cAMP levels, Cdc25p is involved in fermentative growth, nonfermentative growth, cell cycling, sporulation, and cell size regulation. Thus, the main positive regulator of yeast Ras proteins is Cdc25p.

Characteristic aspects of Ras signaling-based screening systems

Ras signaling-based yeast screening systems for the exploration of protein interaction partners allow for positive selection of interactions between soluble cytosolic proteins or between a soluble protein and a hydrophobic membrane protein through the restoration of Ras signaling [36–38]. These systems employ the cdc25 yeast strain, which is deficient in Ras signaling and regains it with the presence of interacting protein pairs. The machinery of intrinsic cell survival and proliferation of Ras signaling is utilized for the readout. Interactions of proteins of interest, including transcriptional activators or repressors that might induce transcription of a reporter or disable vital functions in yeast, can be investigated because of the restitution of Ras signaling on the plasma membrane but the absence of reconstitution of DNA-binding transcription factors in the nucleus. The restricted cell survival with Ras signaling-based selection is suitable for screening large libraries (Table 1), although the method has comparative difficulty in accurately assessing relative interaction strengths.

Table 1.   Protein–protein interaction pairs identified or applied in G-protein signaling-based systems.
Interaction pairReference
  1. a This system is to be used for monitoring receptor tyrosine kinase activity. b This system is to be used for selective isolation of affinity-enhanced variants.

Sos recruitment system
 c-Jun–JDP1 or c-Jun–JDP2 (Jun dimerization proteins)[36]
 c-Jun–Fra-2, c-Jun–FosB or c-Jun–c-Fos (Fos)[36]
 p110–p85[36]
 BRCA1 (breast cancer susceptibility gene 1)–CtIP (CtBP-interacting protein)[84]
 Sox9–PKA-Cα (protein kinase A catalytic subunit α)[85]
 VDAC1 (voltage-dependent anion-selective channel 1)–Tctex1 (t-complex testis expressed-1)[86]
 VDAC1–PBP74 (peptide-binding protein 74)[86]
 p5–p5[87]
 GABAA receptor γ2 subunit–GODZ (Golgi-specific DHHC zinc finger protein)[88]
 IRS-1 (insulin receptor substrate 1)–HDAC2 (histone deacetylase 2)[89]
 p73–PKA-Cβ (protein kinase A catalytic subunit β)[90]
 Truncated ERβ (estrogen receptor β)–truncated ERβ[91]
 HBO1 (histone acetyltransferase binding to ORC-1)–PR (progesterone receptor)[92]
 CMV 1a (cucumber mosaic virus 1a)–TIP1 or CMV 1a–TIP2 (tonoplast intrinsic proteins)[93]
 TRAF2 (tumor necrosis factor receptor associated factor 2)–Smurf2 (SMAD-specific E3 ubiquitin protein ligase 2)[94]
 EF3 (elongation factor 3)–Cch1 (high-affinity calcium channel)[95]
Ras recruitment system
 c-Jun–c-Fos[38]
 p110–p85[38]
 JDP2–C/EBPγ (CCAAT/enhancer-binding protein)[38]
 Pac65 (Pac2; p21-activated kinase 2)–Rac1 mutant[38]
 Pac65–Grb2 (growth factor receptor-binding protein 2)[38]
 Sos (son of sevenless)–Grb2 (growth factor receptor-bound protein 2)[38]
 Truncated EGFR (epidermal growth factor receptor) fused with M-Jun–truncated EGFR fused with M-Fosa[39]
 Glucocorticoid receptor NR3C1–ZKSCAN4 (zinc finger with KRAB and SCAN domains 4)[40]
 PacR (Pac2 regulatory domain)–Chp (Cdc42Hs homologous protein)[96]
 β-Catenin–CBP (CREB-binding protein)[97]
 JNK (c-Jun N-terminal kinase)–IKAP (IκB kinase complex-associated protein)[98]
 ErbB (EGFR)–Grb2[99]
 c-Myc–Krim-1A or c-Myc–Krim-1B (Krab box proteins interacting with Myc)[100]
 RalA (Ras-like protein A)–ZONAB (ZO-1-associated nucleic acid-binding protein)[101]
Yeast–mammal chimeric Gα system
 Snf1 (AMP-activated protein kinase)–Snf4 (regulatory subunit of Snf1 kinase complex)[78]
 Raf–Ras mutant[78]
Gγ interfering system (G-protein fusion system)
 Syntaxin 1a–nSec1 (neuronal Sec1)[79]
 FGFR3 (fibroblast-derived growth factor receptor 3)–SNT-1 (FGFR signaling adaptor)[79]
Gγ recruitment system
 ZZ domain or Z variants (Z domain: B domain mutant derived from protein A)–Fc part (of human IgG)[80]
Competitor-introduced Gγ recruitment systemb
 ZZ domain or Z variants–Fc part[102]

Sos recruitment system

The Sos recruitment system (SRS) was initially reported as a Ras signaling-based screening system, and it takes advantage of the fact that the human RasGEF protein, hSos, can substitute for the GEF of yeast endogenous Ras (yRas) protein, Cdc25p, to allow cell survival and proliferation (Fig. 1A) [36]. In the SRS, a yeast variant strain that has the temperature-sensitive cdc25-2 allele is required. The cdc25-2 strain cannot survive at a restrictive temperature (36 °C), owing to a lack of function of Cdc25p to activate Ras signaling, whereas it can grow at a lower temperature (25 °C). One protein should be membrane-associated or be attached to an inner membrane translocating signal involved in myristoylation and palmitoylation, and the other protein should be soluble and be fused to hSos to prevent false autoactivation by membrane localization of hSos. Only when the membrane-localized protein interacts with the hSos fusion protein will hSos be recruited to the plasma membrane and yeast Ras signaling be rescued. As a consequence, the temperature-sensitive mutant that expresses interacting protein pairs can grow at 36 °C.

Figure 1.

 Schematic illustration of Ras signaling-based screening systems. (A) The SRS system using the human RasGEF protein, hSos. (a) Noninteracting protein pairs are unable to activate the yeast Ras signaling pathway, and are also unable to drive cell growth. (b) Interacting protein pairs bring hSos to the plasma membrane, where it can exchange GDP for GTP of yeast endogenous Ras. The active form of GTP-bound yRas allows cell survival. (B) The RRS system using a constitutively active mutant of mammalian Ras lacking the lipid modification motif (mRas). (a) Noninteracting protein pairs are unable to activate the yeast Ras signaling pathway, and are also unable to drive cell growth. (b) Interacting protein pairs bring mRas to plasma membrane, where it can activate the yeast Ras signaling pathway. Ras signaling allows cell survival. X and Y represent test proteins for interaction analysis.

Using the SRS, a novel repressor that interacts with the c-Jun subunits of AP-1 and represses its activity was isolated [36] (Table 1). AP-1 is a transcription factor that binds to DNA through a leucine zipper motif. Thus, the ability of the SRS to identify transcriptional regulators has been reasonably well established, owing to the membrane-localized interaction, unlike conventional Y2H systems based on the reconstitution of DNA-binding transcription factors in the nucleus.

Ras recruitment system

The Ras recruitment system (RRS), using mammalian Ras (mRas), was later developed as an improved version of the SRS [38]. The RRS has the advantages of the SRS without some of its limitations. For example, the RRS permits more strict selection, owing to the stringent requirement for membrane localization of mRas, can eliminate the isolation of predictable Ras false positives, owing to the introduction of mRasGAP, and can more broadly detect interactions, owing to the relatively small size of Ras as compared with hSos [37,38]. The RRS is based on the absolute requirement that Ras be localized to the plasma membrane for its function (Fig. 1B). In the RRS, mRas lacking its CAAX motif for localization to the plasma membrane, but possessing a constitutively active mutation, is used as a substitute for hSos, and mRasGAP is additionally expressed. The membrane localization of mRas through protein–protein interactions in a cdc25-2 yeast strain results in the activation of its downstream effector, adenylyl cyclase, and restores its growth ability. In an initial report, the usefulness of the RRS was confirmed by practical screening of a cDNA library of 500 000 independent transformants [38] (Table 1). Later, the RRS was applied to detect the activity and inhibition of a dimerization-dependent receptor tyrosine kinase and to identify an interacting pair of human glucocorticoid receptors from a HeLa cell cDNA library [39,40] (Table 1).

Pheromone signaling-based screening systems

Heterotrimeric G-protein signaling in yeast

As peripheral membrane proteins, heterotrimeric G-proteins associate with the inner side of the plasma membrane. Heterotrimeric G-proteins consisting of three subunits, Gα, Gβ, and Gγ, exist in various subfamilies and are widely conserved among eukaryotic species. They transduce messages from ubiquitous receptors, which control important functions such as taste, smell, vision, heart rate, blood pressure, neurotransmission, and cell growth [29]. Yeast has only two types of heterotrimeric G-protein: pheromone signaling-related and nutrient signaling-related [30–32]. Nutrient signaling is profoundly and intricately linked to Ras signaling [30,31], whereas the pheromone signaling pathway is connected to mating processes [32].

The yeast pheromone signaling-related G-protein comprises three subunits, Gpa1p, Ste4p, and Ste18p, which structurally correspond to mammalian Gα, Gβ, and Gγ, respectively [32]. The heterotrimeric G-protein is divided into two key components from the perspective of structure and function. Gα (Gpa1p) is associated with the intracellular plasma membrane through dual lipid modifications of myristoylation and palmitoylation in the N-terminus [41], whereas the Gβγ dimer (the Ste4p–Ste18p complex) is also localized to the inner leaflet of the plasma membrane through dual lipid modifications of farnesylation and myristoylation in the C-terminus of Ste18p, and the formation of a complex between Ste4p and lipidated Ste18p [41,42]. They form part of the signaling cascade activated by G-protein-coupled receptors (GPCRs), and mediate cellular processes in mating in response to the presence of pheromone (Fig. 2A).

Figure 2.

 Yeast pheromone signaling pathway and its utilization for a GPCR biosensor. (A) Schematic illustration of the pheromone signaling pathway. (a) In the absence of α-factor, heterotrimeric G-protein is unable to activate the pheromone signaling pathway. (b) Binding of α-factor to Ste2p receptor activates the pheromone signaling pathway through heterotrimeric G-protein. Sequestered Ste4p–Ste18p complex from Gpa1p activates effectors and subsequent kinases that constitute the MAPK cascade, resulting in phosphorylation of Far1p and Ste12p. Phosphorylation of Far1p leads to cell cycle arrest. Phosphorylation of Ste12p induces global changes in transcription. Sst2p stimulates hydrolysis of GTP to GDP on Gpa1p, and helps to inactivate pheromone signaling. (B) Schematic illustration of typical genetic modifications enabling the pheromone signaling pathway to be used as a biosensor to represent activation of GPCRs. Intact or chimeric Gpa1p can transduce the signal from yeast endogenous Ste2p or heterologous GPCRs that are expressed on the yeast plasma membrane. Transcription machineries that are closely regulated by the phosphorylated transcription factor, Ste12p, are used to detect activation of pheromone signaling with various reporter genes. FAR1, SST2 and STE2 are often disrupted (shown in light gray) to prevent growth arrest, improve ligand sensitivity, and avoid competitive expression of yeast endogenous receptor.

The yeast haploid a-cell has a sole pheromone receptor, Ste2p, which is classified as a GPCR, and the tridecapeptide α-factor functions as a pheromone and binds to the Ste2p receptor on the cell surface [32]. The heterotrimeric G-proteins are closely associated with the intracellular domain of the Ste2p receptor, and the pheromone-bound receptor is conformationally changed and activates the G-protein [43]. Gpa1p is thereby changed from an inactive GDP-bound state to an active GTP-bound state and dissociates the Ste4p–Ste18p complex. Subsequently, the dissociated Ste4p–Ste18p complex binds to effectors through Ste4p, and then activates the mitogen-activated protein kinase (MAPK) cascade [44,45]. The Ste5 scaffold protein binds to the kinases of the MAPK cascade and brings them to the plasma membrane. The concentration of the bound kinases on the membrane possibly promotes amplification of the signal [46,47]. As a consequence, the activated pheromone signaling leads to the phosphorylation of Far1p and the transcription factor Ste12p. These phosphorylated proteins trigger cell cycle arrest in G1 [48–50] and global changes in transcription [51,52]. FUS1 gene expression is representative of the drastic changes in transcription in response to pheromone signaling [53,54]. As a principal negative regulator, the Gpa1-specific GAP Sst2p, a member of the regulator of G-protein signaling family, is also involved in the pathway [55,56].

Pheromone signaling-based screening systems – ligand–GPCR or GPCR–G-protein interactions

Background of pheromone signaling-based screening systems

GPCRs constitute the largest family of integral membrane proteins, and have a variety of biological functions. They are the most frequently addressed drug targets, and modulators of GPCRs form a key area for the pharmaceutical industry, representing nearly 30% of all Food and Drug Administration-approved drugs [57,58]. Yeast permits the functional expression of various heterologous GPCRs and other signaling molecules such as G-proteins. Yeast also facilitates versatile genetic techniques for screening and quantification. Therefore, it offers opportunities to establish fundamental technologies for drug discovery or basic medicinal study [59,60]. Yeast-based screening systems exploiting pheromone GPCR signaling enable the analysis of several interactions, including not only protein–protein but also ligand–receptor and receptor–protein interactions. These systems can recognize the on–off switching of a signal, such as the binding of an agonist/antagonist to a receptor, and critical mutations involved in ligand-dependent or constitutive activation/inactivation of signaling molecules. In addition, assays can be performed at the yeast optimum temperature of 30 °C, unlike with Ras signaling-based systems, which require the incubation of yeast cells at suboptimal temperatures (25 and 36 °C), and the monitoring or discrimination of the signaling changes through quantitative and survival readouts. Hence, they have been applied in various experiments, including target identification, ligand screening, and receptor mutagenesis.

Pheromone signaling as a biosensor for understanding GPCRs

GPCRs have a common tertiary structure, composed of seven hydrophobic integral membrane domains, and the mechanism of signaling that is mediated by heterotrimeric G-proteins is also conserved between yeast and mammalian cells. This has led to the construction of ingenious systems that provide for the mutual exchange of signals between heterologous GPCRs and yeast G-proteins in yeast without generating dysfunctions. With versatile screening techniques, yeast can be used as a sensor to detect the initiation of GPCR-associated signaling [59,60]. Briefly, in wild-type yeast a-cells, Ste2p receptor or mammalian receptors can activate the yeast pheromone signaling pathway via intracellular heterotrimeric G-proteins, including the native form or an engineered form of Gpa1p, in response to ligand binding. The activated pheromone signals cause cell cycle arrest and transcription activation, which are exploited as signaling readouts (Fig. 2A,B). These biosensing techniques have been established in yeast with engineered pheromone signaling, and numerous characteristics of pheromone signaling molecules have been successfully elucidated [43–45, 47–50, 53–55]. Moreover, pheromone signaling-related molecules, such as Ste2p receptor, G-proteins, and peptidic α-factor pheromone, have been extensively mapped with mutagenesis techniques, demonstrating their usefulness for screening huge libraries and for identification of important domains or amino acids [61–66].

Bioassay and transcriptional assay for signaling detection

The arrest of the cell cycle completely prevents cell growth during signaling. Monitoring of cell densities in liquid media with or without pheromones can distinguish signaling on the basis of delay of entry into the logarithmic growth phase. The agar diffusion bioassay (halo assay), in which cells are mixed with unsolidified fresh agar medium in which pheromone-spotted paper filter disks are placed, can also discriminate signaling by showing cleared-out areas around the disks, forming halos, owing to the robust inhibition of cell growth (the halos may look blacked out on a monochromatic figure) [55,62,63,66,67].

On the other hand, the use of transcriptional changes that are closely regulated by the signaling makes possible versatile procedures for detection. The FUS1 gene, which is engaged in drastic augmentation of the transcription level responding to the signal, is commonly taken as a reflector of signaling and is fused with various reporter genes associated with growth and photometry. Auxotrophic or drug-resistant reporter genes, such as HIS3 or hph, are generally used for selection, and are suitable for screening large-scale libraries [66–68]. Colorimetric, luminescent and fluorescent reporters, such as lacZ, luc, or GFP, are usually used for numerical conversion and are appropriate for relative and quantitative assessment of signaling levels [61–64,66–68].

Gene disruption for system modification

The arrest of the cell cycle caused through phosphorylation of Far1p allows for the examination of pheromone signaling [55,59,60,62,63,66,67]. However, this makes growth reporter genes for positive selection, such as HIS3, useless for the detection of signaling, owing to stagnation of cell growth [66,67], whereas the synchronization of the cell cycle in G1 arrest provides uniform levels of expression of reporter genes such as GFP for each cell [69]. For that reason, FAR1 is usually disrupted in positive selection screens using growth selection (Fig. 2B). Because the far1Δ strain never induces cell cycle arrest, it can be used in growth selection to screen for positive clones in response to pheromone signaling, which is represented by the expression of the HIS3 reporter gene on histidine-defective plates [66,67]. At the same time, it has been reported that the arrest of the cell cycle causes the drastic dropout of episomal plasmids, resulting in a serious problem when the library is screened and the target plasmids are collected, and hence the disruption of FAR1 could significantly improve plasmid retention rates [69]. Accordingly, disruption of FAR1 is required for positive growth screening.

The SST2-deficient strategy is widely used in utilizing pheromone signaling as a sensor, owing to hypersensitivity for ligand binding [59,60,63,67,69]. SST2 gene encodes the Gpa1-specific GAP that stimulates hydrolysis of GTP to GDP on Gpa1p and helps in the inactivation of pheromone signaling. Removal of Sst2p function causes a considerable decrease in GTPase activity for Gpa1p, and makes the conversion of GTP to GDP difficult, owing to a lack of competence of GTPase activity (Fig. 2B). The loss of SST2 could provide supersensitivity, even to a 250–10 000-fold lower concentration of α-factor [67]. However, a relatively high background signal of the sst2Δ strain, especially when grown in rich medium such as YPD, has been confirmed in the absence of α-factor pheromone by a transcription assay using the FUS1GFP reporter gene [69]. Although the SST2-deficient strategy is a powerful technology for experiments requiring high sensitivity, it does not necessarily produce the best signal-to-noise ratio. Accordingly, choosing the correct situation for using Sst2p is required for each experiment. In addition, STE2 is often disrupted, to avoid competitive expression of yeast endogenous receptor [59–64,66,69].

Expression of heterologous GPCRs

Many heterologous GPCRs containing adrenergic, muscarinic, serotonin, neurotensin, somatostatin, olfactory and many other receptors have been successfully expressed in yeast, and the feasibility of yeast-based GPCR screening systems has been demonstrated [59,60,68,70–75]. Yeast Gpa1p, which is equivalent to Gα, shares high homology, in part, with human Gαi classes, and a number of GPCRs of human and other species are able to interact with Gpa1p and activate pheromone signaling in yeast [73–75]. Many other human GPCRs can also function as yeast signaling modulators as a result of various genetic modifications, including one in which chimeric Gpa1p systems (so-called ‘transplants’) have only five amino acids in the C-terminus of Gpa1p substituted for those of human Gα subunits, including the Gαi/o, Gαs and Gαq families (Fig. 2B) [71]. Indeed, these transplants have allowed functional coupling of serotonin, muscarinic, purinergic and many other receptors to the yeast pheromone pathway [71–73,76].

The rat M3 muscarinic acetylcholine receptor has been used for rapid identification of functionally critical amino acids, with random mutagenesis of the entire sequence [72]. In this system, the CAN1 reporter gene coding for arginine–canavanine permease was integrated into the locus of a pheromone response gene in yeast cells whose endogenous CAN1 gene was deleted, and the recombinant strain expressed Can1p in response to ligand-dependent signaling. Owing to the cytotoxicity of canavanine caused by Can1p expression, recombinant strains with inactivating mutations in the receptor can survive on agar media containing canavanine and receptor-specific agonists. The recovered mutant M3 muscarinic acetylcholine receptors in this system also show substantial functional impairments in transfected mammalian cells, and the utility of the yeast-based procedure for GPCR mutagenesis has been proven.

Human formyl peptide receptor like-1, which was originally identified as an orphan GPCR, has been used to isolate agonists for GPCRs of unknown function [77]. Histidine prototrophic selection by the FUS1–HIS3 reporter gene was performed with secreted random tridecapeptides as a library and a mammalian/yeast hybrid Gα subunit which allows functional coupling with the receptor. As a result, surrogate agonists as peptidic candidates have been successfully screened, and the promoted activation of formyl peptide receptor like-1 expressed in human cells has been validated with synthetic versions of the peptides.

Pheromone signaling-based screening systems – protein–protein interactions

Yeast–mammal chimeric Gα system

Medici et al. [78] constructed an intelligent system for analysis of protein–protein interactions by managing heterotrimeric G-protein signaling in yeast (Fig. 3A). They initially found that a fusion protein between the yeast Ste2p receptor lacking the last 62 amino acids of the cytoplasmic tail and the full-length Gpa1p transduced the signal in response to the binding of α-factor in cells devoid of both endogenous STE2 and endogenous GPA1. Subsequently, a yeast–mammal chimeric Gα composed of the N-terminal 362 amino acids of Gpa1p and the C-terminal 128 amino acids of rat Gαs was prepared. The chimeric Gα is able to interact with the yeast Gβγ complex, but is not able to interact with the yeast Ste2p receptor, and it was fused to the truncated Ste2p receptor. Although a gpa1Δ yeast strain harboring the yeast–rat chimeric Gα does not respond to pheromone, a ste2Δ gpa1Δ yeast strain expressing the Ste2p–Gpa1p–Gαs fusion protein that is covalently linked to Ste2p and the chimeric Gα displayed a strong pheromone response in the presence of α-factor. These results suggest that the specific interaction of the receptor with the C-terminus of Gα is necessary to bring the two proteins into close proximity. This hypothesis was applied to the analysis of protein–protein interactions. It was demonstrated that the interaction of Gpa1p–Gαs fused to protein X and Ste2p receptor fused to protein Y permitted pheromone response signaling through the contact between Ste2p and Gpa1p–Gαs, using the interaction between Snf1 and Snf4, which form a kinase complex regulating transcriptional activation in glucose derepression, or between Raf and the constitutively active form of Ras (Table 1). In this system, a gpa1Δ haploid strain harboring the plasmid, which complements Gpa1p function to capture Ste4p/Ste18p subunits, or a GPA1/gpa1Δ diploid yeast strain was used to avoid lethality by spontaneous signaling from the liberated Ste4p/Ste18p subunits.

Figure 3.

 Schematic illustration of pheromone signaling-based screening systems for protein–protein interaction analysis. (A) The yeast–mammal chimeric Gα system uses chimeric Gpa1p, which is able to interact with the yeast Gβγ complex, but not with the yeast Ste2p receptor. Chimeric Gpa1p is fused to protein X, and yeast Ste2p receptor is fused to protein Y. (a) Noninteracting protein pairs are unable to activate the pheromone signaling pathway. (b) Interacting protein pairs bring Ste2p and chimeric Gpa1p into close proximity, and permit physical contact between the two, resulting in activation of pheromone signaling. (B) The Gγ interfering system can screen for negative mutants that do not interact. Ste18p genetically fused to the C-terminus of cytoplasmic protein X and integral membrane protein Y are coexpressed in a ste18Δ strain. (a) Noninteracting protein pairs are able to activate the pheromone signaling pathway. (b) Interacting protein pairs are unable to activate the pheromone signaling pathway, owing to the interruption of contacts between the Gβγ complex and its effector. (C) The Gγ recruitment system can completely eliminate background signals for noninteracting protein pairs. Mutated Ste18p lacking membrane localization fused to cytoplasmic protein X and membrane-associated protein Y are coexpressed in a ste18Δ strain. (a) Noninteracting protein pairs completely lack pheromone signaling, owing to the release of the Ste4p–Ste18p complex into the cytosol. (b) Interacting protein pairs restore signaling, owing to the recruitment of the Gβγ complex onto the plasma membrane.

Gγ interfering system

The Gγ interfering system (it was called a G-protein fusion system in the original literature) has been developed to monitor integral membrane protein–protein interactions and to screen for negative mutants with loss of the interaction capacity (Fig. 3B) [79]. The yeast Gγ-subunit Ste18p was genetically fused to the C-terminus of cytoplasmic protein X, and the protein X–Gγ fusion protein and integral membrane protein Y in its native form were coexpressed in a ste18Δ strain. The interaction between protein X–Gγ and protein Y inhibits pheromone signaling through the Gβγ complex, in spite of the presence of α-factor, whereas a lack of interaction between protein X and protein Y normally leads to signaling. This event might be attributed to the fact that restrictive localization or structural interruption by trapping of the Gβγ complex at the position of protein Y on the membrane disturbs the contact with its subsequent effector. In one example, interactions of attractive drug target candidates, syntaxin 1a and nSec1 or fibroblast-derived growth factor receptor 3 and SNT-1, were monitored, and nSec1 mutants that lost the ability to bind to syntaxin 1a were successfully identified by taking advantage of growth arrest induced through the protein–protein interaction [79] (Table 1).

Gγ recruitment system

The above-described systems for analysis of protein–protein interactions using pheromone signaling are proven techniques for selecting target proteins involved with membrane proteins. However, they might generate relatively high background signals, making them unfavorable for screening candidates by growth selection, because the machinery for distinguishing interactions does not always ensure complete inactivation of signaling in the presence of pheromone.

The Gγ recruitment system has recently been developed using the pheromone signaling pathway, and is a dependable system that completely eliminates background signals for noninteracting protein pairs in the presence of pheromone (Fig. 3C) [80]. This system can be used to investigate cytosolic–cytosolic or cytosolic–membrane protein interactions. A yeast strain with a mutated Gγ lacking membrane localization ability (Gγcyto) should be prepared by deletion of the dual lipid modification sites in the C-terminus of Ste18p, because yeast pheromone signaling strictly requires the localization of the Gβγ complex to the plasma membrane [41,42]. The release of Ste18p into the cytosol eliminates the signaling ability mediated by the Ste4p–Ste18p complex [41], and this technique therefore leads to absolute interruption of background signals. One test protein must be soluble and fused with Gγcyto to be expressed in the cytosol but not the membrane, whereas the other may be soluble but should have an added lipid modification site to allow association with the inner leaflet of plasma membrane, or it may be an intrinsically hydrophobic integral membrane protein or lipidated element of a membrane-associated protein. Consequently, when the cytosolic protein X–Gγcyto fusion protein and the membrane-associated protein Y are expressed in a ste18Δ haploid strain in the presence of α-factor pheromone, the interaction between protein X and protein Y restores signaling, owing to the recruitment of the Gβγ complex onto the plasma membrane, which can be monitored, but a lack of interaction between protein X and protein Y results in no background signaling.

In an original report, the ZZ domain derived from protein A of Staphylococcus aureus and the Fc portion of human IgG, which are both soluble proteins, were used as a model interaction pair (Table 1). The ZZ domain is a tandemly repeated Z domain that binds to human Fc protein and displays higher affinity than a Z domain monomer [81]. The interaction between the ZZ domain with an attached dual lipidation motif in its C-terminus and Fc fused to the C-terminus of Gγcyto was easily detected with a transcriptional assay using the pheromone response FIG1 promoter and a GFP reporter gene or a halo bioassay by growth arrest, whereas background signals from noninteracting pairs were never observed, owing to the loss of localization of the yeast Gβγ complex at the plasma membrane.

The wild-type and two variants of the Z domain that each possess a single mutation and exhibit different affinity constants were expressed as additional interaction pairs for the Fc fusion protein [82]. All variants with a wide range of affinity constants, from 8.0 × 103 to 6.8 × 108 m−1 [83], were clearly detectable, and moreover, the relatively faint interaction with an affinity constant of 8.0 × 103 m−1 was successfully detected because of the complete elimination of background signal for noninteracting pairs (Table 1). Surprisingly, a logarithmic proportional relationship between affinity constants and fluorescence intensities measured by the transcriptional assay was observed, suggesting that this approach may facilitate the rapid assessment of affinity constants.

Finally, the Gγ recruitment system has more recently been improved by the expression of a third cytosolic protein that competes with the candidate protein [102]. The competitor-introduced Gγ recruitment system could specifically isolate only affinity-enhanced variants from libraries containing a large majority of original proteins, clearly indicating the applicability of this new approach to directed evolution.

Concluding remarks

Yeast-based approaches with the G-protein signaling machineries presented here are remarkably useful for the detection and screening of interactions of proteins involved in various biological processes. These systems are essentially comparable to the Y2H systems that have been predominantly used to screen protein–protein interaction partners from large-scale libraries and to estimate the relative strengths of interactions, but are additionally able to detect activation or inactivation associated with the switching machinery of signaling molecules, such as major pharmaceutical targets of GPCRs. Yeast-based and signaling-mediated screening systems are obviously powerful and practical tools with which to quickly screen for possible candidates. In the future, we can be sure that they will be improved, with more powerful and user-friendly advanced modifications, and will be widely applied to various fields, such as protein engineering.

Acknowledgements

This work was supported in part by a Research Fellowship for Young Scientists from the Japan Society for the Promotion of Science and a Special Coordination Fund for Promoting Science and Technology, Creation of Innovation Centers for Advanced Interdisciplinary Research Areas (Innovative Bioproduction Kobe), from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

Ancillary