Functional Expression of a Mammalian Odorant Receptor

See allHide authors and affiliations

Science  09 Jan 1998:
Vol. 279, Issue 5348, pp. 237-242
DOI: 10.1126/science.279.5348.237


Candidate mammalian odorant receptors were first cloned some 6 years ago. The physiological function of these receptors in initiating transduction in olfactory receptor neurons remains to be established. Here, a recombinant adenovirus was used to drive expression of a particular receptor gene in an increased number of sensory neurons in the rat olfactory epithelium. Electrophysiological recording showed that increased expression of a single gene led to greater sensitivity to a small subset of odorants.

Olfactory transduction begins with the binding of an odorant ligand to a protein receptor on the olfactory neuron cell surface, initiating a cascade of enzymatic reactions that results in the production of a second messenger and the eventual depolarization of the cell membrane (1). This relatively straightforward and common signaling motif is complicated by the existence of several thousand odorants, mostly low-molecular-weight organic molecules, and nearly a thousand different putative receptors (2, 3). The receptors are believed to be members of the superfamily of G protein–coupled receptors (GPCRs) that recognize diverse ligands, including the biogenic amine neurotransmitters. Although the putative odorant receptors constitute the largest subfamily of GPCRs, in some ways they remain the most enigmatic, because no particular mammalian receptor has been definitively paired with any ligand. Functional expression of cloned odorant receptors would allow the characterization of the chemical receptive fields that provide the basis for coding and organization in the olfactory system.

A functional expression system for odorant receptors requires both that the receptors are properly targeted to the plasma membrane, and that they couple efficiently with a second messenger system that produces a measurable response to ligand stimulation. On the simple assumption that olfactory neurons themselves would be the most capable cells for expressing, targeting, and coupling odorant receptors, we have endeavored to use the rat nasal epithelium as an expression system, driving the expression of a particular receptor by including it in a recombinant adenovirus and infecting rat nasal epithelia in vivo. Here, we relied on the large number of putative odorant receptors, and their approximately equal expression among the 6 million neurons of the rat olfactory epithelium, to identify the average increase in response in an epithelium in which one of these receptors is overexpressed (4). This can be measured extracellularly as a transepithelial potential resulting from the summed activity of many olfactory neurons, a measurement known as the EOG or electro-olfactogram (5). The amplitude of this voltage is determined by both the size of the response in individual cells and the number of cells responding.

Adenovirus vectors have been developed as a tool for efficient gene transfer in mammalian cells and have shown promise in a variety of experimental and clinical applications (6). We generated an adenovirus vector, AdexCAG-I7-IRES-GFP (Ad-I7) (Fig.1A) (7), that contained an expression unit for a particular odorant receptor, rat I7 (2). This adenovirus vector is missing the E1 and E3 early genes, rendering it replication-incompetent and thereby preventing a lytic infection in treated olfactory epithelia (8). To aid in EOG electrode placement, we included the gene for the physiological marker green fluorescent protein (GFP) in the expression cassette, using an internal ribosomal entry site (IRES) insert to produce a bicistronic transcript that would result in the expression of odorant receptor and GFP as separate proteins in the same cells (9).

Figure 1

(A) Construct of the recombinant adenovirus AdexCAG-I7-IRES-GFP (Ad-I7). The replication-defective adenovirus expression vector Adex consists of the human adenovirus type 5 (Ad5) genome lacking the E1a, E1b, and E3 regions. Ad-I7 has a bicistronic expression unit including the CAG promoter (CAG), odorant receptor I7 (OR-I7) coding sequence, internal ribosomal entry site (IRES), the S65T mutant of green fluorescent protein (GFP), and rabbit β-globin polyadenylation signal (GpA). The CAG promoter drives transcription of the bicistronic transcript I7-IRES-GFP, producing I7 odorant receptor and GFP as separate proteins. (B) Side view of the medial surface of the rat nasal turbinates. The turbinates are labeled with Roman numerals. Dorsal is up, anterior to the left. This animal was infected with Ad-I7 virus. (C) Fluorescent micrograph of the same tissue as in (B) showing the heterogeneous expression of GFP, which marked the location and the degree of virus infection. (D) Closer view of the Ad-I7 virus–infected endoturbinate II′ in (C). In regions of high fluorescence, we estimate that infection rates were near 10% of neurons. (E) Northern blot hybridization of total RNA from uninfected olfactory epithelia (Ctrl OE), Ad-I7 virus–infected olfactory epithelia (I7 OE), and rat brain with an I7 probe. A band of about 2.8 kb was detected in infected OE. This is near the expected size with the insert of 2.35 kb (I7-IRES-GFP) and 0.47 kb of untranslated sequence from the CAG promoter, and the β-globin polyadenylation signal sequence. Lanes were loaded with 20 μg of total RNA from each tissue. Examination of 28S and 18S ribosomal RNA confirmed the integrity of the RNA. (F) Cryosection (15 μm) of Ad-I7 virus–infected olfactory epithelium reacted with an antibody to GFP, showing stained olfactory neurons with characteristic morphology [including soma (arrowhead) and single dendrite (arrow)] and position within the olfactory epithelium. ML, mucous layer; OE, olfactory epithelium; BL, basal lamina. Scale bars, 3 mm [(B) and (C)], 1 mm (D), 50 μm (F).

Animals were killed 3 to 8 days after infection (10) and the nasal cavity was opened, exposing the medial surface of the nasal turbinates (Fig. 1B). Under fluorescent illumination, the GFP clearly marked the pattern of viral infection and protein expression (Fig. 1, C and D). Expression was heterogeneous; in some regions of the epithelia as many as 20% of the cells were infected, whereas in others there was virtually no sign of infection. Overall, about 1 to 2% of the sensory neurons were infected and expressed the GFP gene product. There was significant variability between animals, with the highest infection rates often found in the second and third turbinates, usually near the edges (Fig. 1D). The reasons for these patterns of infectivity are not known, but the most likely explanation is simply access of the virus to tissue within the rather complex structures of the nasal cavity.

Expression of the bicistronic mRNA for the I7 receptor and GFP was verified by Northern blot of the infected epithelia (11). Using a probe that covered the entire coding sequence of the I7 gene, we detected a single band of about 2.8 kb in infected but not in uninfected epithelia, where native I7 expression is presumably below the level of detection (Fig. 1E). Although this does not provide a quantitative measure of the extent of mRNA expression in single cells, it does provide clear evidence that expression of the I7 receptor transcript is much higher in infected versus uninfected tissue. The high-gain signal amplification in olfactory neurons assures that even the activation of a few receptors by ligand will produce at least some sensory current in individual cells (12). In the absence of antibodies specific for the I7 odorant receptor, we used GFP antibodies (13) to further verify that more than 80% of the infected cells were olfactory neurons (Fig. 1F). Thus, areas of high GFP fluorescence signal the positive infection of sensory neurons.

One difficulty in determining the ligand specificity of odorant receptors is the enormous stimulus repertoire to be tested. We developed a panel of 74 odorants, including aromatic and short-chain aliphatic compounds with various functional groups (14). Odorants were prepared by standard methods (15) and applied to the epithelium in the vapor phase by injecting a pressurized pulse of odor vapor into a continuous stream of humidified clean air. All odors were tested at a solution concentration of 10 3 to 10 2 M. Concentrations of stimulus at the olfactory epithelium could not be known, but the system reliably delivered the same amount of stimulus on each trial, as there was little variability between responses to successive pulses of the same odorant. The absolute odorant concentration at the tissue was not critical because it was the relative change in responsiveness between treated and untreated animals that was measured.

EOG recordings were obtained with a glass capillary electrode placed on the surface of the epithelium and connected to a differential amplifier (16). The EOG response is a negative-going spike of potential that is typically 0.5 to 15 mV in amplitude and 1.5 to 5 s in duration. To account for between-animal variability, we used a standard odorant, amyl acetate, to which all other odorant responses were normalized (17). EOGs were recorded from the region of olfactory epithelium that showed high numbers of fluorescent (infected) cells in Ad-I7 virus–infected turbinates; EOGs were also recorded in a similar location from uninfected animals. All infected animals were able to respond to all of the 74 odorants in the test panel, although responses to all odors in virally infected animals were on average 30% smaller in amplitude. Nonetheless, the amplitudes of the responses relative to the amyl acetate standard were unchanged. This was the only nonspecific effect attributable to the virus that we observed, and the same nonspecific reduction in response amplitude was also seen in animals infected with viruses containing only thelacZ or GFP genes.

Responses to eight representative odorants from the panel of 74 are shown for an uninfected control animal and an animal infected with the Ad-I7 virus (Fig. 2). For seven of the eight odorants shown (plus the amyl acetate standard), as for 62 of the 65 other odorants in the panel, there was no appreciable change in responsiveness between the infected and uninfected animal. However, for one odorant—octyl aldehyde (octanal), an eight-carbon, straight-chain, saturated aliphatic aldehyde—the response was substantially greater, both in amplitude and time course, in the infected animal. The large increase in amplitude and time course indicated that more individual olfactory neurons were responsive to octyl aldehyde in the infected versus uninfected tissue.

Figure 2

Representative EOG recordings from an uninfected animal (A) and an Ad-I7 virus–infected animal (B). All EOGs are responses to odorants at a solution concentration of 10−3 M and are normalized to the reference odorant, amyl acetate, which is given the value of 1. (C) Subtraction of (A) from (B), which brings out the much-increased response to octyl aldehyde in the Ad-I7 virus–infected animal. The slight positive direction of some of the subtracted traces is primarily the result of differences in the time course of the responses in the different animals.

The average response amplitudes for 14 of the odorants selected from the panel of 74, at a solution concentration of 10 3 M, were compared for both uninfected and Ad-I7–infected animals (Fig. 3A). Octyl aldehyde responses of infected animals were, on average, 1.7 times those of uninfected animals, whereas all other odorants were near control levels. The responses to octyl aldehyde were also compared with a control in which animals were infected with a virus containing only the GFP gene. Expression of GFP in these animals was comparable to that seen in animals infected with Ad-I7, but the response to octyl aldehyde was not different from uninfected animals. Thus, neither viral infection nor GFP was sufficient to generate the increase in responsivity to octyl aldehyde. Finally, the response increase was location-sensitive. If the EOG electrode was moved to an area with no GFP expression, the increased response to octanal was reduced to control levels.

Figure 3

(A) Comparison of average EOG amplitudes in Ad-I7 virus–infected and uninfected animals to 14 odorants at a solution concentration of 10−3 M (n = number of recordings, N = number of animals; error bars represent SD). All responses were normalized to the reference odorant, amyl acetate. Responses in uninfected animals are given the value of 1, and responses in infected animals are scaled accordingly. As a further control, the response to octyl aldehyde was compared with that in animals infected with an adenovirus carrying only GFP. Statistical significance of the data was evaluated with thet test (**P < 0.001). (B) I7 virus– infected animals have increased odorant responses to heptaldehyde (C7), octyl aldehyde (C8), nonyl aldehyde (C9), and decyl aldehyde (C10), but not to hexaldehyde (C6) or undecylic aldehyde (C11). The bars are the result of a subtraction that shows the relative increase of the average responses to these odorants in infected versus uninfected animals (n = number of recordings, N = number of animals). Statistical significance of the data was evaluated with the t test (**P < 0.001, *P < 0.01). (C) Comparison of responses (EOG amplitude) in an infected and uninfected animal to increasing concentrations of amyl acetate (AA, triangles) and octyl aldehyde (OA, squares). Heavy lines and solid symbols are from the infected animal; light lines and open symbols are from the uninfected control animal. The x axis is the logarithm of the molar concentration of odorant solutions. (D) The chemical structure of octyl aldehyde, CH3(CH2)6CHO, variously described as having a light citrus, soapy, or fatty odor quality.

We wished to determine if other, related odorants might also be recognized by the I7 receptor. First, varying the carbon chain length, we tested saturated aldehydes from C3 to C12. In addition to octyl aldehyde (C8) there were significantly increased responses to heptyl (C7), nonyl (C9), and decyl (C10) aldehydes, but no increases were detected for aldehydes with less than 7 or more than 10 carbons (Fig. 3B). Other C8 aliphatic compounds with different functional groups also failed to elicit responses larger than in normal animals, and at least one C8 unsaturated aliphatic aldehyde,trans-2-octenal, also failed to elicit a response. A group of five aromatic aldehydes also failed to give responses. Thus, the response profile of the I7 receptor, at least within the scope of the 74-odorant panel screened here, is relatively specific for C7 to C10 saturated aliphatic aldehydes.

At the comparatively high concentrations used in these experiments, the response to octyl aldehyde in infected animals was on average 1.7 times the response in control animals. However, these high concentrations mask the full effect of the virally induced odorant receptor expression because of nonlinearities associated with high concentration responses. As can be seen in the dose-response relations (Fig. 3C), comparison of the octyl aldehyde response to the amyl acetate response at various concentrations revealed up to a sevenfold difference in response magnitude in the infected versus the uninfected epithelium. For example, at an odorant concentration of 5 × 10 5 M in solution, the response ratio of octanal to amyl acetate in the infected animal was 4.0:0.6 mV—a ratio of about 7. At the same concentration in the normal animal, the responses were nearly equal and the ratio was 1. Similar maximum increases, with response ratios ranging from 4 to 7, were measured in four infected animals.

Because the infected cells expressed GFP, it was possible to identify individual positive neurons after dissociation of the epithelium, allowing responses to be tested in single cells (Fig.4). In whole-cell patch clamp recordings of single infected neurons, we recorded responses to octanal in each of seven GFP-positive cells (18). In uninfected cells we recorded a response to octanal in only 2 of 28 cells. The responses displayed latencies (150 to 250 ms) and amplitudes (90 to 500 pA) within the normal range (12), and both the kinetics and amplitude were concentration-dependent. Responses to maintained odor stimuli displayed a characteristic sag in the response, owing to adaptation (Fig. 4C). By these measures, the responses to octanal in virally transduced cells followed the pattern of normal odor responses in untreated cells, indicating that the endogenous second messenger system was most likely responsible for generating the response in the infected cells.

Figure 4

(A) Freshly dissociated rat olfactory neurons (arrows) can be easily identified by their morphology. (B) In the same field under fluorescence illumination, an olfactory neuron infected by Ad-I7 virus can be identified by expression of GFP. Scale bar, 20 μm. (C) Sample whole-cell recordings from a GFP-positive cell. Top traces are responses to 5 × 10−4 M (larger) and 5 × 10−5 M (smaller) octanal; lower trace shows the response to a 6-s step of octanal at 10−4 M. Holding potential was –60 mV.

Thus, a member of the multigene family first identified as encoding putative odorant receptors (2) does indeed code for a protein that is capable of specific odor binding leading to a physiological response. On the basis of the controls, we believe that the increased octanal sensitivity in infected animals was not likely to have resulted from viral infection alone. Viral infection with eitherlacZ (19) or GFP viruses had no effect; the octanal response was selective for regions of the epithelium that had been infected; and the only nonspecific effect observed was a general reduction (by about 30%) in responsivity to all odors in infected epithelia.

Identifying the molecular receptive field of an olfactory sensory neuron is a critical first step in understanding how olfactory perception is achieved by higher brain centers. Within the limited set of odors tested, the I7 receptor is nonetheless striking in its ability to discriminate the C6 hexanal, to which it shows no response, from the C7 heptanal, which gives a strong response. Interestingly, to humans, hexanal has a grassy odor, whereas the C7 to C10 aldehydes share a primarily fatty odor and the higher aldehydes have a stronger fruity or citrus quality. The chemical receptive field for this single odor receptor is similar to the response patterns recently measured in single olfactory neurons and in mitral cells of the olfactory bulb (20). One difference was that for mitral cells the carbon chain length appeared to be somewhat more critical than functional group in determining stimulus efficacy, whereas our results indicate that both are critical determinants. In an earlier report of odor receptor expression, the rat OR5 receptor, expressed by the baculovirus system in insect sf9 cells, showed a surprisingly wide-ranging sensitivity to odors from at least five separate chemical classes, all at low concentrations (21). This result is somewhat at odds with the data from mitral cell recordings and with our data, but it may be related to the expression system or the method of applying odor stimuli. Among nonmammalian odor receptors, the nematode receptor (ODR10) displayed a narrow ligand specificity (22), whereas three fish receptors responsive to a commercial fish food mixture have not been paired with specific ligands (23).

Genes of the odorant receptor subfamily have a hypervariable region corresponding to the second through fifth transmembrane domains (2), the presumed ligand-binding site in GPCRs (24). However, in some cases odorant receptors of the same subfamily differ from each other by only a few residues in this region (25). These receptors can now be tested to measure the effects of naturally occurring sequence variations on ligand sensitivity. Such a program, coupled with introduced mutations, could lead to a detailed, experimentally testable understanding of the relation between gene sequence, protein structure, and ligand-binding specificity in membrane-bound receptors.


  1. View this table:

Stay Connected to Science

Navigate This Article