Introduction
Animals ranging from nematodes to humans sense their chemical environment through olfactory receptors (ORs)1, 2, 3. In the mammalian olfactory signaling pathway, upon activation by an odorant, the ORs of olfactory sensory neurons stimulate the G protein Golf (ref. 4) by catalyzing the exchange of guanine nucleotides in the 
subunit. The GTP-bound Golf then interacts with an olfactory epithelium-specific adenylyl cyclase (type III adenylyl cyclase, or ACIII)5, which leads to the synthesis of cyclic AMP (cAMP), which in turn stimulates a nucleotide-gated Ca2+ channel, thus resulting in the influx of Na+ and Ca2+ ions5, 6. The resultant action potential traverses through the axons to the glomeruli of olfactory bulb, where it is transmitted to the dendrites of the mitral neurons and then on to the higher olfactory cortical centers in which the signal is processed as an odor7, 8. The exquisite sensitivity of the olfactory system has been primarily ascribed to the combinatorial signaling by multiple ORs responding to a single ligand3, 8. Harnessing the combinatorial signaling potential of the olfactory signaling pathway, which can detect innumerable chemical agents with unparalleled sensitivity and selectivity, should be of immense value in the detection of environmental toxins and chemical warfare agents even at sublethal levels. Such a methodology should also prove extremely useful for screening a large number of candidate molecules of therapeutic value. However, owing to the difficulty of expressing ORs in heterologous cell systems, attempts to define specific receptors involved in combinatorial olfactory signaling and their ligands have been unsuccessful so far4. Of the different heterologous cell systems that can be used to express OR and the downstream signaling pathway, the presence of well-organized G protein–coupled receptors (GPCRs)9, 10, along with a fully defined genomic organization and robust growth characteristics, points to Saccharomyces cerevisiae as an ideal candidate. Therefore, we sought to engineer the mammalian olfactory signaling pathway into the budding yeast S. cerevisiae such that the yeast would be able to detect a defined chemical agent through the genetically integrated ORs and report it by emitting fluorescence through the expression of green fluorescent protein (GFP) (Fig. 1a).
Figure 1: Engineering of the WIF-1
strain.
![Figure 1 : Engineering of the WIF-1|[alpha]| strain.](/nchembio/journal/v3/n6/thumbs/nchembio882-F1.jpg)
(a) Schematic showing the construction of olfactory yeast strain WIF-1–RX. Mammalian olfactory signaling components, except the receptor, were cloned into the YPH501 strain of S. cerevisiae to form the WIF-1 strain. Transfection with a plasmid containing the GFP gene driven by the CREBP promoter formed the WIF-1 strain, in which GFP expression serves as a reporter for activation of the olfactory signaling system. The hypervariable domain of GPCRs of interest can be cloned into the RI7 OR scaffold and transfected into the system to generate the ligand-sensitive WIF-1–RX strain. (b) Expression of RI7, Golf and CREBP. Immunoblot analysis was carried out to monitor expression of engineered RI7, CREBP and Golf using lysate proteins from WIF-1 and parental control strains with appropriate antibodies. A nonspecific protein was used as a loading control (LC). (c) Localization of RI7 on WIF-1–RI7 cells. To monitor the localization of the RI7 on the cell surface, immunofluorescence analysis was carried out in the WIF-1 strain and WIF-1 strain expressing RI7 (WIF-1–RI7) using RI7 antibodies and Texas Red–labeled anti-rabbit IgG. Nonspecific antibody from preimmune serum was used as control. (d) Interaction between octylaldehyde and RI7 receptors. Spheroblasts from WIF-1–RI7 cells were incubated with 14C-octylaldehyde at 25 °C. Representative Scatchard plot derived from such analysis is presented. Results from the studies indicated the presence of binding sites for octylaldehyde with a Ka of 0.1 
104 M-1. Regression analysis indicated that there were 90 receptors per cell. Error bars are s.e.m.
To establish such a prototypic yeast strain, we cloned the components of the olfactory signaling pathway into the parental YPH501 strain of S. cerevisiae. Because signaling by the ORs specifically involves the subunit of Golf (refs. 4,11) and its effector ACIII, we cloned the complementary DNA inserts encoding rat Golf and ACIII into a bicistronic and bidirectional yeast expression vector pESC-Trp and transfected the vector into YPH501 cells to generate YPH501-Golf-ACIII transfectants. Efficient receptor and G-subunit interaction requires the presence of a functional heterotrimeric complex of the G protein involving specific , 
and 
subunits12. Although the endogenous subunit (also called Ste4/Ste18) of S. cerevisiae has been shown to interact with Golf (ref. 13), the mammalian subunit can couple to the homospecific Golf more efficiently and preclude the activation of endogenous Ste4/Ste18-dependent pathways that might contribute to increased background noise in signaling. Therefore, we cloned rat 2 and 5, which show preferential expression in olfactory neurons14, 15, into the yeast expression vector pESC-Ura and transfected into YPH501-Golf-ACIII cells. In S. cerevisiae, increased cAMP levels and the resultant activation of protein kinase A (PKA) downregulate the expression of a large number of endogenous genes16. This is in contrast to the mammalian cells, in which the cAMP-activated PKA stimulates the expression of many genes via the transcription factor cAMP response element binding protein (CREBP), which promotes the transcription of specific target genes by binding to the cAMP response elements (CREs) of their promoters17. To engineer such a cAMP-responsive gene-activation system in yeast, we cloned cDNA inserts encoding human CREBP and CRE-driven GFP into the expression vector pESC-His. We transfected this construct into YPH501-Golf 25-ACIII cells to generate the yeast strain YPH501-Golf 25-ACIII-CREBP-CREGFP, or WIF-1.
Although a few different GPCRs have been functionally expressed in S. cerevisiae18, 19, 20, attempts to express ORs in cell types other than olfactory neurons have been hampered by defective membrane targeting and lack of functional coupling with downstream signaling components. Recently it has been shown that the rat olfactory receptor I7 (RI7) that responds to octanal (8-CHO)3, 21, 22, 23 (Compound 1
) can be functionally expressed in many different cell types, including S. cerevisiae21, 24, 25. Therefore, to test the functional coupling of OR signaling in WIF-1 cells, we cloned a cDNA insert encoding RI7 into the expression vector pESC-Leu and transfected the resultant pESC-Leu-RI7 vector into WIF-1 strain cells. We performed immunoblot analyses of the WIF-1–RI7 cells to monitor the expression of the engineered olfactory signaling components and found that these components, namely RI7, Golf and CREBP, were successfully expressed in these cells (Fig. 1b). Further analysis using immunofluorescence microscopy indicated localization of RI7 on the cell surface of WIF-1–RI7 cells (Fig. 1c). However, it is possible that not all the receptors identified by immunofluorescence microscopy were functional. Therefore, to assess the binding ability of the expressed RI7 receptors, we carried out binding studies with WIF-1–RI7 spheroblasts using 14C-octanal (Compound 2
). Scatchard plot analysis indicated the presence of 90 receptors per cell with a Ka value of 0.1 104 M-1 (Fig. 1d). This is in agreement with the 10–100 
M threshold levels seen in RI7 for 8-CHO8, 23 and the view that ORs are generally low-affinity receptors8.
To characterize the functional integration of RI7 into the engineered GFP reporter, we monitored the GFP response of WIF-1–RI7 cells after exposing them to different concentrations of 8-CHO. The results revealed a dose-dependent increase in GFP expression as indicated by its fluorescence at 535 nm (Fig. 2a). The kinetics of GFP response to 8-CHO (25 M) indicated a time-dependent increase in GFP expression that we observed from 1 h onward; maximal levels of fluorescence were reached by 3 h (Fig. 2b). Exposure to the closely related aliphatic aldehydes 7-CHO (Compound 3
) and 6-CHO (Compound 4
) and the alcohol 8-OH (Compound 5
) indicated that WIF-1 cells are more responsive to 8-CHO at a concentration of 50 M (Fig. 2c). We also analyzed the specificity of response by fluorescence microscopy using WIF-1 cells exposed to 8-CHO, 7-CHO or 6-CHO. As shown, WIF-1–RI7 cells more robustly responded to 8-CHO (Fig. 2d). The WIF-1–RI7 cells showed a weaker response to 7-CHO, and the response was even weaker at higher concentrations, which is in agreement with the previous findings that RI7 responds to 7-CHO, albeit not as strongly as to 8-CHO23.
Figure 2: GFP responses of WIF-1–RI7 strain.
![Figure 2 : GFP responses of WIF-1|[alpha]||[ndash]|RI7 strain.](/nchembio/journal/v3/n6/thumbs/nchembio882-F2.jpg)
(a) WIF-1–RI7 response to different concentrations (10 M to 50 M) of octylaldehyde. GFP expression was measured in a microplate reader at 1-h intervals for 3 h with 485-nm excitation and 535-nm emission. Results indicated the expression of GFP increased with increasing octylaldehyde concentrations. (b) Kinetics of response. WIF-1–RI7 cells were exposed to 25 M octylaldehyde in a 48-well tissue culture plate. GFP fluorescence was monitored at 3-h intervals for 18 h using a Perkin-Elmer HTS 7000 Plus BioAssay reader with an excitation of 485 nm and emission at 535 nm. Fluorescence intensity was plotted as a function of time. (c) Specificity of WIF-1–RI7 response. Cells were incubated with 50 M hexanal (6-CHO), heptanal (7-CHO), octanal (8-CHO) or octanol (8-OH). GFP-mediated fluorescence was measured at 1-h intervals for 6 h in a microplate reader with 485-nm excitation and 535-nm emission. Results are expressed as a ratio of GFP expression over the control values at 3-h time points. Results are from three independent determinations (mean 
s.e.m.). (d) Specificity of WIF-1–RI7 response monitored by fluorescence microscopy. WIF-1–RI7 yeast cells were exposed to 25 M 6-CHO, 7-CHO, 8-CHO or 8-OH. Aliquots of cells (10 l) were mounted on microscope slides and visualized by fluorescence microscopy.
Next we investigated whether the RI7-based WIF-1 yeast strain can be used to functionally express other ORs. Sequence analyses of ORs (and also other GPCRs) have shown that the N termini of these receptors are involved in plasma membrane localization, whereas the C termini generally define the specificity for G protein interaction12, 22. The ligand-binding pockets of the ORs are defined by highly variable amino acids spanning transmembrane domains two (TMII) and seven (TMVII)1, 2, 3. Based on these observations, a chimeric OR that can activate phospholipase C via G15 has been functionally expressed in human embryonic kidney (HEK) 293 cells21, 24. Following this experimental paradigm, we constructed the receptor expression cassette pESC-Leu-RX, into which the hypervariable ligand-binding pocket of GPCRs of interest, including ORs, can be inserted in-frame. This receptor expression cassette contains an insert that encodes the N terminus (amino acids 1–61) and C terminus (amino acids 295–327) of the RI7 receptor flanking an intervening sequence containing multiple cloning sites. Using these multiple cloning sites, cDNA inserts encoding the hypervariable, ligand-binding pockets of GPCRs can be inserted 'in-frame' so that a chimeric receptor with defined N and C termini can be encoded by this expression cassette (Fig. 3a). Given that we retained the N and C termini of RI7, Golf interaction and subsequent signaling should not have been perturbed. Therefore, we reasoned that the shuttling of the ligand-binding pocket of a specific GPCR into the receptor expression cassette of pESC-Leu-RX and expressing it in WIF-1 cells would confer these cells with the ability to detect the ligand that is specific for the particular GPCR and report it via GFP.
Figure 3: Shuttling of ligand-binding domains of other GPCRs in the WIF-1 system.
![Figure 3 : Shuttling of ligand-binding domains of other GPCRs in the WIF-1|[alpha]| system.](/nchembio/journal/v3/n6/thumbs/nchembio882-F3.jpg)
(a) Schematic illustration of receptor expression strategy. The receptor expression cassette of the pESC-LeuRX vector was constructed to contain an insert that encodes the N and C termini of the RI7 receptor flanking an intervening sequence containing multiple cloning sites. The ligand-binding pocket of different ORs can be shuttled into this modular RI7 receptor–based template as indicated. (b) Response of 2AR-expressing WIF-1–2AR strain to isoproterenol. WIF-1–2AR yeast cells were exposed to 1 M or 10 M isoproterenol. An aliquot of cells (10 l) was mounted on microscope slides and visualized by fluorescence microscopy. (c) Response of WIF-1–vanillin receptor (WIF-1–VanR) and WIF-1–citronellal receptor (WIF-1–CitR) strains. WIF-1–VanR cells were exposed to 10 M vanillin, isoproterenol or 8-CHO. WIF-1–CitR cells were exposed to 10 M citronellal, isoproterenol or 8-CHO. GFP-mediated fluorescence was determined at 3 h using a Perkin-Elmer HTS 7000 Plus BioAssay reader with an excitation of 485 nm and emission at 535 nm. Results from three independent determinations (mean s.e.m.) are expressed as ratio of GFP expression over the control at the 3-h time point.
To validate such signal coupling, we inserted the ligand-binding domain spanning TMII to TMVII (amino acids 71–324)26 of the hamster lung 2-adrenergic receptor (2AR) into the pESC-Leu-RX cassette and transfected the resultant pESC-Leu-RX-2AR vector into WIF-1 yeast cells. When we exposed these WIF-1–2AR transfectants to 1 M or 10 M isoproterenol (Compound 6
), GFP was expressed by 3 h (Fig. 3b), which indicates the functional integration of the 2AR ligand-binding pocket to the RI7 receptor scaffold and validates its signaling through the heterologously expressed Golf, 25, ACIII, CREBP, and finally the CRE-driven GFP. To examine further whether the ligand-binding pockets of other ORs can be functionally integrated into the pESC-Leu-RX receptor cassette, we shuttled the cDNA inserts encoding the ligand-binding domains (amino acids 107–296) of vanillin (ORL829; GenBank accession number AB061229)27 and citronellal (ORL451; GenBank accession number AB102523)21 receptors into the receptor expression cassette of pESC-Leu-RX. After verifying the in-frame integration of the ligand-binding domains to the receptor expression cassette of pESC-Leu-RX by DNA sequencing, we transfected these vectors into WIF-1 cells to obtain WIF-1–vanillin and WIF-1–citronellal receptor strains. We monitored the responses of these cells to vanillin (Compound 7
) and citronellal (Compound 8
), respectively. Both these strains responded to their respective ligands in a dose-dependent manner from 1–25 M, and we could see an optimal response at 10 M by 3 h, as indicated by the expression of GFP (Fig. 3c). These responses were found to be specific, as equivalent concentrations of isoproterenol or 8-CHO failed to elicit any GFP response (Fig. 3c).
These results indicated that by expressing a library of ligand-binding pockets of 'orphan' receptors using the receptor scaffold described here, the identity of the receptor specific for a particular ligand could potentially be established. Therefore, we sought to use this strategy to identify an OR for a specific odorant. Considering the critical need to develop a biosensor for the detection of explosives, their residues and derivatives, we focused on identifying an OR capable of detecting 2,4-dinitrotoluene (DNT) (Compound 9
), a mimic for the explosive trinitrotoluene (TNT). To accomplish this, we cloned a library of cDNA inserts encoding the ligand-binding domains of rat ORs (derived from rat olfactory epithelium) into the pESC-Leu-RX cassette and transfected them into WIF-1 cells to obtain WIF-1–OR transfectants. By exposing these WIF-1–OR cells to DNT (50 M) and scoring the cells that emitted green fluorescence, we identified several DNT-responsive clones. After enriching these DNT-responsive individual colonies by serial dilution, we monitored their response to DNT (25 M) by fluorescence microscopy (Fig. 4a). After confirming the presence of 
600-base-pair OR inserts in these DNT-responsive yeast cells by colony PCR methods (Fig. 4b), we rescued the pESC-Leu-RX vectors from these cells for sequence analysis28. After reconfirming the responsiveness, the putative DNT-responsive OR insert was PCR-amplified and sequenced. Sequence analysis indicated that the putative OR insert shuttled into the pESC-Leu-RX vector encodes the hypervariable, ligand-binding pocket of a new OR (Fig. 4c). Further analysis indicated that this protein shows extensive homology to the ligand-binding region spanning TMII through TMV of mouse ORs Olfr2 and MOR226-1 (see Supplementary Fig. 1 online). Owing to its high similarity with the amino acid sequence of the mouse OR MOR226-1, we have named this receptor Olfr226 (GenBank accession number EF061115). Importantly, the identity of an odorant that can stimulate either Olfr2 or MOR226-1 has yet to be identified. In this context, it is worth noting that the strategy presented here has identified a new rat OR that can detect DNT. Although studies using cell-based and/or olfactory sensory neuron–based assays are necessary to corroborate our findings, our results demonstrate that the yeast-based expression system presented here is capable of identifying novel receptor-ligand interactions. Further studies should define whether this system can be used to detect TNT and similar noxious agents in the environment.
Figure 4: Screening of WIF-1–OR strains for DNT-responsive colonies.
![Figure 4 : Screening of WIF-1|[alpha]||[ndash]|OR strains for DNT-responsive colonies.](/nchembio/journal/v3/n6/thumbs/nchembio882-F4.jpg)
(a) Response of WIF-1–OR strain to DNT. Ligand-binding regions spanning TMII to TMVII of ORs were PCR-cloned into the pESC-LeuRX receptor template as described in the Methods. These engineered vectors (pESC-Leu-OR) were transfected into the WIF-1 yeast strain. The clones that responded to 50 M DNT were scored, isolated and enriched. The cells were exposed to 25 M DNT for 3 h, and an aliquot of these cells (10 l) was mounted on microscope slides and visualized by fluorescence microscopy. (b) Identification of DNT-responsive OR inserts. The presence of 600-base-pair OR inserts in these strains (WIF-1–OR) was verified by colony PCR methods. M, DNA markers; C, control (empty vector). (c) Sequence of the DNT-responsive OR insert. These vectors were rescued from the DNT-responsive WIF-1–OR clones and sequenced (GenBank accession number EF061115). The sequence of the 157 amino acids spanning TMII–TMV that contribute to DNT responsiveness is presented. Amino acids encoding transmembrane domains are highlighted. Further analysis has indicated that this protein shows extensive homology to the mouse ORs Olfr2 and MOR226-1 (see Supplementary Fig. 1).
The findings reported here are significant for several reasons. First, they demonstrate for the first time that the reconstruction of the complete cAMP system with heterologous genes and the cassette-based receptor-scaffold strategy can be used for the functional expression of both olfactory and nonolfactory GPCRs. Second, using the engineered reporter, this system can be used to screen libraries of compounds to identify agonists or antagonists for GPCRs of interest. With further genetic fine-tuning of the engineered pathway, the WIF-1 strains can be adapted for high-throughput screening. For example, with the engineering of alternate reporter systems such as those involving Ca2+ and aequorin, real-time luminescence-based assay systems can be developed. In addition, by substituting the N terminus of RI7 with that of Ste2, the yeast mating pheromone GPCR, one can increase the expression of the receptor. Mutation of the serine and threonine residues (Thr309, Thr320 and Ser324) of the C-terminal tail of RI7—which are likely to be involved in receptor desensitization and internalization—may serve to increase the half-life of the expressed receptor. At present, we are pursuing these approaches to adapt WIF-1 strains for a high-throughput screening platform.
Considering the importance of 'de-orphanizing' the hundreds of ORs and other GPCRs, our findings have an added significance. For instance, by shuttling the hypervariable domain of orphan GPCRs, including those of the OR family, one can use this system to identify the putative ligands. Likewise, the hypervariable domain of a known receptor can be shuttled into the yeast system presented here to screen for novel therapeutic agonists of interest. Now that we have validated this approach, with further refinement the engineered olfactory signaling pathway should be amenable for high-throughput screening to identify different receptors for a specific ligand or different ligands for a specific receptor.
Methods
Reagents and fine chemicals.
All the chemicals used in these studies were of analytical grade and were from Fisher. The primers used were from Sigma-Genosys, and the odorants octanal, heptanal, hexanal, octyl alcohol, vanillin, citronellal and DNT were from Sigma-Aldrich, Inc. The stock solutions of the odorants in DMSO were made just before the experiment and diluted at least 100-fold in yeast culture medium for exposure to yeast cells.
Yeast strain and vectors.
Wild-type yeast strain YPH501 (Stratagene) was maintained in YPAD agar plates (containing 2% Difco Bacto peptone, 2% D-glucose, 1% yeast extract, 0.0075% adenine hemisulfate salt and 2% agar). The bidirectional pESC vectors were from Stratagene.
Construction of WIF-1
strain of yeast and immunoblot analysis.
See Supplementary Methods online.
RI7 localization by fluorescence microscopy.
WIF-1–RI7 cells were permeabilized using 20 units of lyticase (Sigma) for 30 min at 30 °C. Cells were collected by centrifugation and incubated with RI7 antibodies (ImmunoDetect, Inc. # 5/20-1) at 1:200 dilution for 1 h. This was followed by further incubation with secondary antibody (at 1:500 dilution) labeled with Texas Red fluorophore (Molecular Probes # T-2767) for 1 h. 10 l of cells were mounted and observed by laser confocal microscopy. Cells stained with preimmune serum were used as control.
Receptor binding assay.
The binding assay was carried out using 14C-octylaldehyde (American Radiolabeled Chemicals Inc.) to quantify the number of RI7 receptors expressed in yeast cells. Spheroplasts were prepared from yeast cultures. Cells were treated with lyticase and spheroplasts were collected on a sucrose-Ficoll gradient29. 100-l aliquots of cells (1 107 cells ml-1) were suspended in phosphate-buffered saline (PBS) and incubated with varying concentrations of 14C-octylaldehyde (50 mCi mmol-1), ranging from 1 M to 1 mM, at 30 °C for 15 min. The reaction was terminated by the addition of chilled PBS. Nonspecific binding was determined with 100-fold molar excess of unlabeled ligand. Similar binding analysis was carried out with the parental cells not expressing RI7 receptors. The cell pellets were washed twice with PBS, and the radiolabel bound to the pellets was quantified by scintillation counting. Specific binding was calculated by subtracting the specific binding of control cells (WIF-1) from the specific binding obtained with RI7-expressing WIF-1–RI7 cells. Scatchard plot analysis was carried out using GraphPad Prism software (http://www.graphpad.com). The slope and intercept were obtained by regression analysis, and the x intercept was used to calculate the number receptors expressed per yeast cell.
Monitoring GFP response in WIF-1 constructs.
WIF-1–RI7 or other receptor-variant yeast strains were inoculated in synthetic dextrose minimal medium (Qbiogene Inc.) at 30 °C for 8 h to reach mid-log phase. The absorbance was read at 600 nm and the cell number was calculated using the correlation that 0.25 optical density at 600 nm equals approximately 3 106 cells ml-1 (ref. 30). Cells were harvested by centrifugation and suspended in synthetic galactose minimal medium (3 108 cells per 100 ml) to induce protein expression at 30 °C for 14 h. Cells were collected by centrifugation and resuspended in synthetic galactose minimal medium. Cells were aliquoted in each well at a density of 7 107 cells per 200 l and stimulated with the compounds of interest at concentrations ranging from 10 M to 50 M. The expression of GFP was measured at 1-h intervals of time for different lengths in a microplate reader (Perkin-Elmer HTS 7000 Plus BioAssay reader) with an excitation of 485 nm and emission at 535 nm. Results were expressed as ratio of GFP expression over the control. All response studies were carried out in triplicate in order to ensure reproducibility.
Analysis of GFP-mediated fluorescence by confocal fluorescence microscopy.
GFP expression as a response to different compounds was visualized and analyzed by fluorescence microscopy on an Olympus Fluoview confocal microscope with a 60, 1.4 numerical aperture plan apochromat objective. After exposing the cells to the respective agents, 10-l aliquots of yeast cells were mounted on microscope slides for analysis. All prepared slides were kept in the dark until they were visualized using 468-nm excitation wavelength and 535-nm emission.
Accession codes.
GenBank: The Olfr226 sequences reported in this paper have been deposited under accession number EF061115. Sequences cited from previous studies include the ligand-binding domains of the vanillin (accession number AB061229)27 and citronellal (accession number AB102523)21 receptors.
Note: Supplementary information and chemical compound information is available on the Nature Chemical Biology website.
