Most explanations of the brain systems involved in flavor start with the sense of taste. However, we have made the case that, paradoxically, smell is more important for taste than taste is, so we start with smell and give most of our attention to it. This is the first step toward a scientifically based neurogastronomy.
The sense of smell begins with the action of smell molecules on the receptor molecules in our nose. Here lies a second paradox. Smell molecules in our food, as described in the previous chapter, have been the subject of study for many years. Companies producing the ingredients of processed foods employ armies of organic chemists, who characterize the stimulating properties of thousands of different chemicals and try to relate them to the perceptions they arouse. But it takes two to tango: these smell molecules, and the receptor molecules with which they interact. Until 1991 nothing was known about the receptor molecules, so the critical molecular basis of the smell response and the subsequent perception was missing.
In 1991 the receptor molecules were discovered, making it possible to start the tango with both partners together. However, because of a host of difficult issues with the receptors, the scientific field was very slow to take up the problem of how the two partners interact. Subsequently, when molecular gastronomy came onto the scene in the late 1990s, information about the receptors was barely starting, and it is still slow in coming because of the many problems that still have to be solved. Thus the paradox: you might think that molecular gastronomy is molecular because it concerns both the smell molecules and the receptor molecules, but in fact so far it is mainly concerned with the perceptions of the smell molecules.
For neurogastronomy, the receptor molecules are a critical part of the smell and flavor system in the brain, so our interest begins with the interaction of the smell molecules and the receptor molecules in the nose. From this perspective, we are describing a subfield that we might call molecular neurogastronomy. As the new kid on the block, it lacks the richness of lore and knowledge about foods and their flavors that gastronomy and molecular gastronomy embrace. However, all this knowledge will merge increasingly to delight our culinary as well as scientific appetites.
Everything that follows in building up the sensations of smell and flavor starts with this critical and fascinating meeting between the molecules that are in our food and those that are in our receptor cells. How does it work?
The Concept of a Lock and Key
Imagine that a molecule is like the key to your house. Run your finger along the jagged edge that fits exactly the complementary edge inside the lock. When you insert the key into the lock the two come together, you can turn the key, release the lock, and open the door. This “lock-and-key” concept has been used by biologists for more than a century as the model for how two molecules interact with each other. When the key turns the lock, the molecule changes its internal structure. This change results in a microkick delivered to a neighboring molecule, which starts a cascade of microkicks to other molecules that results in the cell doing what it is supposed to do.
An odor molecule is made up of different types of atoms that give it an irregular structure. It functions as the key. What does the lock look like, and how does it work? This is one of the big questions in the modern science of smell. Solving this problem has given us a remarkable insight: how the information contained in the odor molecule is represented by an image in the brain.
Challenges
Studying how a smell molecule activates a smell receptor is therefore part of the larger enterprise of understanding how the information carried in a sensory stimulus is transformed into a representation in the nervous system. This process is best understood in vision, where individual photons activate rhodopsin molecules in the photoreceptor cells in our retinas. It is also well studied in hearing, where sound waves are converted first into vibrations within our inner ears, which activate the hair cells in our cochleas. In each case, the optimal stimulus is known and can be controlled very precisely.
Sensory stimulation of smell receptors is much more difficult. We do not have the advantage of being able to “see” or “hear” the stimuli we are delivering. Monitoring a smell molecule in the air has to be done indirectly by instrumentation. The receptor cells are hidden inside the nasal cavity and are hard to get at in order to record their responses. The cells are easily fatigued by repeated stimuli (just as we rapidly adapt to an odorous environment), so testing has to proceed slowly. We usually do not know beforehand which odors will activate a given cell; finding this out among the thousands of possible odors takes a lot of time. Even when we identify an odor, we have to figure out what other odors are within the receptive range of this particular cell. All this work is carried out with orthonasal smell, which can be controlled by puffs of smell, but the basic mechanisms will apply to retronasal smell as well.
Features of Smell Molecules
The keys, in our analogy, are the smell molecules, many of them described in the previous chapter. They range from small molecules to larger musks and pheromones and even fragments of proteins given off by the body that are borne on particles in the air. This panoply of smell molecules is particularly relevant to the smells we obtain by sniffing in. However, our problem is simplified because retronasal smell molecules are mainly the smaller molecules that volatilize (evaporate) from the liquids and foods released from within our mouths.
What part of these small molecules is the part that stimulates the receptors? If we know that answer, we can begin to understand how that information gets represented in the brain. It has been known by organic chemists and shown by a host of physiological studies that changes in a single molecular feature, a single part of the molecule such as a single atom, can change smell perception. The fundamental bits of knowledge contained in smell molecules and represented in the brain are therefore likely to be individual features of a smell molecule. These are of several types.
I will build on chapter 4 to describe them. These features are what link molecular gastronomy to neurogastronomy.
Most obviously, odor molecules may vary by their length; thus, straight-chain (called aliphatic) molecules may vary from having one to a dozen or more carbon atoms in their backbone. A second difference resides in a terminal functional group. Many odor molecules have a terminal group that defines whether it is an alkane, acid, or aldehyde, each of which is associated with characteristic smells. As we saw, these are found in many kinds of foods. With these first two features, we can define a homologous chemical series as one consisting of molecules with the same terminal functional group (hence homologous) but with chains consisting of different numbers of carbon atoms (hence series).
A third difference is whether a functional group is contained within the carbon atom chain, such as an oxygen atom in a ketone. A fourth difference is in a side group attached to the side of a carbon chain, such as a phenol ring. A fifth difference is in the chirality of the molecule—that is, whether it is right-handed or left-handed. A sixth difference is in the geometrical shape of the molecule, such as the ringlike conformation of a terpene. A final difference is in the overall size.
The ability to detect these differences of single atoms within smell molecules makes the olfactory system one of the most sensitive molecular detector systems in the entire body. These features are at a much finer level than in the immune system, for example, where an antibody interacts with an antigen site consisting of dozens of individual molecules.
The Race for the Smell Receptors
What kind of receptor molecule could detect such fine differences between molecules, and in addition, between thousands of different molecules? This was one of the great mysteries of science. The earliest ideas were by industrial chemists working with organic molecules. They posited that particular molecular features interact with unknown receptors. These ideas merged after a time with the larger field of study of what are called structure-activity relations in molecule-molecule interactions. These studies evolved into Quantitative Structure-Activity Relations (QSAR), a standard approach by pharmaceutical companies for developing new drugs.
These studies revealed a daunting complexity of odor molecules and their relation to perception. On the one hand, a series of molecules with similar molecular features—such as alcohols, esters, and aldehydes—might give a series of similar perceptions. On the other hand, similar molecules could give quite distinct perceptions. This was baffling for the classical QSAR approach.
On the basis of an analysis of molecular shapes, John Amoore, an Oxford biochemist, suggested that the receptors are tuned to the shapes of the odor molecules; this is the so-called stereochemical theory of odor receptors. But the nature of the receptors, the specific kinds of proteins in the cell membrane, remained unknown.
The first breakthrough came with biochemical research by a former student of mine, Doron Lancet, in Israel. In 1985 he galvanized the field by showing that odors stimulate an enzyme called adenylate cyclase, which produces a well-known cell signaling molecule called cyclic AMP. It was already known that cyclic AMP occurs in a signaling pathway that starts with a receptor that gives a microkick to a so-called G-protein, which forms a large class of G-protein coupled receptors (GPCRs). Lancet drew on his earlier training as an immunologist to predict that a large number of different receptors—as many as 100 to 10,000—would be needed to encode all the different types of odor molecules, a prediction that turned out to be surprisingly close. He also drew on predictions by another former student, John Kauer, and earlier work by Kjeld DØving in Norway, that there would be a “combinatorial relation” between receptors and odor molecules; that is, a given receptor could interact with many odor molecules, and a given odor molecule could interact with many receptors.
Suddenly a number of laboratories in molecular biology realized that discovering the identity of these receptors was one of the hottest topics in biology. The race was on. One by one, the intermediate “microkicking” molecules in the odor-signaling pathway in the receptor cells were cloned and sequenced and identified by their pharmacological properties: the G-protein, the adenylate cyclase, and a protein that forms a channel activated by cyclic AMP that lets in charged particles to give the electrical response. The first recordings of odor responses of isolated receptor cells were made by Stuart Firestein, then a graduate student in Berkeley. The electrical responses in the cilia themselves were recorded. It was an exciting time. The whole sensory cascade was coming into view.
The whole cascade, that is, except for the receptors. Repeated attempts to identify the receptors met with no success. It became obvious that this was probably going to come from a major molecular biology laboratory with the needed expertise and resources. And that is what happened.
A Beautiful Experiment
Linda Buck was a postdoctoral fellow in the laboratory of a leading molecular biologist, Richard Axel, at Columbia University. She had been there for several years, working on several projects on endocrine receptors and on antibodies in the immune system. She began to read about smell and became fascinated by the field and by the problem of finding the receptors.
A few years previously, a technique called the polymerase chain reaction (PCR) had been invented, which made it possible to expose any given tissue in the body to a tiny part of a gene suspected to be present. The tiny part acted as a “probe” to recognize that whole gene and any similar genes if they were there, and to amplify them by repeated probing so that you had enough to recognize them and do something with them. You use this basic “probe” technique to fish out a text from the world literature. For example, if you enter “We hold these truths” into Google, you will recover the entire Declaration of Independence (you can try doing this with other phrases: it is remarkably effective). So PCR was just the right tool for going after the olfactory genes in the tissue in the nose, but nobody had yet made it work for that purpose.
On the basis of the prediction that the olfactory receptors would be a large subfamily of the G-protein coupled receptors, Buck designed her probes to fish out any possible new members of that family in the olfactory epithelium of a rat. She used snipping enzymes to cut her collected sequences in such a way that it produced a set of stripes on a biochemical gel showing that indeed there was a new and very large family of genes belonging to the GPCR family localized in the olfactory epithelium.
As she tells the story, at first she couldn’t believe the results and put them away overnight. However, the next morning she realized that they were real, and she showed them to Axel, who was also elated. Reporting in 1991, they dubbed them the long-sought olfactory receptor genes that carry the genetic code the cell uses to make olfactory receptor proteins. They deduced that the family could contain as many as 1,000 members, in the range that had been predicted.
Discovering the olfactory receptors not only solved the problem of the smell receptors, but opened the door to a new era of research in the olfactory pathway. For these advances, the Nobel Committee in 2004 recognized Buck and Axel for their discovery, one of the greatest in the history of biology.
The Concept of an Odor-binding Pocket
How do these receptors fit with the lock-and-key model? Several years earlier, we and others had started to think about this problem. This led to the hypothesis that the interaction takes place not in a narrow lock that responds to only one key, but in a larger space called a binding pocket, analogous to that of many other receptors for signaling molecules between cells. We further hypothesized that a given receptor cell might carry just one type of receptor, which would require that the receptor does not have a narrow affinity for a given molecule, as with the neurotransmitter receptors, but rather a broad spectrum of affinities reflecting the known broad odor responses of olfactory receptor cells. The nice thing about this hypothesis is that it makes the olfactory receptor cells similar to photoreceptor cells in the retina, each of which carries only one of the three types of color receptors, but with a broad responsiveness to different wavelengths of light.
The “one cell–one receptor” concept was experimentally established by work in Richard Axel’s laboratory and has received direct support from single-cell studies in the rat by Linda Buck’s laboratory, then at Harvard University, and in the insect by studies by John Carlson at Yale University.
The problem is thus to account for a binding pocket that is analogous to that of other transmitter receptors and photoreceptors but can interact differently with different odor molecules. Thus, in contrast with the traditional lock-and-key metaphor for enzymatic and receptor specificity, the olfactory receptor is hypothesized to function by a broad affinity mechanism. This is an example of a new concept of receptor activation involving broad receptor-signal molecule interactions.
Interactions of Smell Molecules and Smell Receptors
The standard way to study receptor-signal molecule interactions in the drug industry is to use genetic tools to “express” the receptor (the technical jargon for make it appear) in a carrier cell, and then test it against different signal molecules and different potential drugs that block or enhance it. By genetic engineering, each amino acid in the chain that makes up the receptor can be changed at will, enabling the scientist to determine which amino acids are essential to the function of the receptor and which interact specifically with different sites (determinants) on the signal molecule.
Unfortunately, olfactory receptors have been very difficult to express experimentally in carrier cells. The breakthrough came in 1998 from my former student Stuart Firestein, his graduate student Haixing Zhao, and their colleagues at Yale and Columbia. They played a molecular trick by attaching a particular receptor gene to a virus, which they then poured over the sheet of receptor cells in the nose of a rat. This enabled the virus to infect all the cells so that they would all “express” that particular receptor (known as the I7 receptor). They could then record from any cell and find out which odor molecules could excite it. Among some 200 odor molecules they tried, the aliphatic (straight-chain) aldehyde composed of a chain of eight carbon atoms (C8, octanal) was preferred, with decreasing responsiveness to flanking members in the series with chain lengths from C6 to C10.
This implied a binding pocket selective for at least two features: terminal functional group and chain length. To test this hypothesis, Michael Singer, a Yale undergraduate and then graduate student in our laboratory, carried out a molecular modeling study. A computational modeling analysis in 2000 of the I7 model receptor showed that, with blind automated docking, octanal interacts within a binding pocket as hypothesized. These studies thus identify four types of stimulating features on odor molecules: functional group, chain length, molecule size, and molecule shape.
A critical test of the computational results is the extent to which they reproduce the binding preferences of I7. There was close agreement: the model showed the same relative preference for octanal over the flanking members of the aldehyde series that the experiments showed. A parallel study from Firestein’s laboratory demonstrated how the odor molecules that interact with I7 are tightly constrained at their functional head but can fit more loosely at their tail ends. It is presumed that this accounts for the graded affinities of these different molecules for the I7 receptor.
The combined result of these experimental and computational studies thus supports the hypothesis that odor determinants interact within a binding pocket in the olfactory receptors that is similar to that of other GPCRs, but with a set of critical sites that varies with different receptors and can show graded interactions with the determinants of different odor molecules.
Combinatorial Interactions Between Smell Molecules and Smell Receptors
Independent experimental testing of the hypothesis was provided in 1999 by the experiments of Bettina Malnic and colleagues in Linda Buck’s laboratory. They first recorded physiologically the odor spectra (the range of different odor molecules that give responses) of isolated olfactory sensory neurons. They then sucked out the contents of the nucleus, which contains the genes, with a fine pipette and recovered the gene of the receptor by the PCR method. They could thus show the differing strengths of interactions between known receptors and different odor molecules. These and other experimental and theoretical studies support the hypothesis that the critical sites on the odor molecules (what we are calling determinants) are transduced (transformed) within a receptor binding pocket. The multiple combinations of amino acid residues and the multiple combinations of odor determinants within the binding pocket provide a fertile substrate for the combinatorial interactions that have long been posited to take place in encoding odor molecules, as described earlier.
These combinations are made even more subtle and complex by interactions at the receptors. The analogy here is with the drug industry. Pharmaceutical companies invest millions of dollars trying to find molecules that act as drugs to antagonize the response of a receptor to a neurotransmitter or to enhance it. We predicted that similar interactions could occur at olfactory receptors, except between different odor molecules competing for the same binding pocket. These indeed have been found, in which a response is seen to molecule A alone but not to molecule B alone; however, when they are presented together, the response to A is reduced because of the “silent” antagonistic effect of B on A. These interactions at the receptor binding pocket indicate that the complexity of a smell perception begins to be shaped at the first step in the smell response.
Understanding these interactions is obviously necessary for relating the molecular properties of the smell molecules to the perception of smell and flavor.
The Broad Molecular Receptive Ranges of Smell Receptors
Both experimental and computational studies reveal that the receptors have relatively broad response spectra that overlap with one another. The broad spectra have been termed molecular receptive ranges (MRRs) by a former student, Kensaku Mori, and Yoshihoro Yoshihara in Japan. This is in analogy with the spatial receptive fields (RFs) that are seen in cells responding to a visual scene in the retina.
The broad overlapping MRRs have been confusing to some, implying a lack of specificity in the responses. The way that specificity can be achieved despite broad overlapping responses is easily explained by analogy with the color system in vision. The cone receptors in the retina contain three types of receptors. They are called red, green, and blue, terms reflecting the peaks of their sensitivities to different wavelengths of light. However, on either side of the peaks, all the receptors have decreasing sensitivities that overlap with one another across most of the wavelength spectrum. This means that at any wavelength there is a unique combination of responsiveness to the three receptors. This unique combination is what gives rise to the unique perception we call color. A given combination stays the same regardless of the intensity of illumination, which enables us to discriminate color despite changes in light intensity.
We and others have suggested that a similar principle applies to smell, except that here one is dealing with the overlapping responses of hundreds of receptors, not just three. Nonetheless, it is the overlaps that enable us to identify unique combinations of receptor responses giving rise to a specific odor perception, no matter how weak or intense it may be. In this way, the brain achieves specificity of an ensemble of neuron responses despite lack of specificity of individual neuron responses.
The Multidimensional Nature of the Odor World
In contrast to color perception, in which wavelength varies along one dimension, odor molecules differ in all the ways we have indicated. Thus we say that “odor space” is multidimensional. This is because of the multiple combinations of the odor features that are possible; that is, a given molecule may belong to more than one dimension, such as a given functional group, a given carbon length, a given degree of saturation, and a given three-dimensional shape. Identifying molecular features is thus a necessary step in being able to characterize more precisely the odor world. As I will explain, this multidimensional nature of odor space poses special problems for its neural representation in the brain.
A Database for Olfactory Receptors
The mammalian genome is believed to contain some 30,000 genes. Three percent of them (approximately 1,000) are olfactory receptors, constituting the largest family in the genome. Just keeping track of all these receptors, comparing them and classifying them has presented a large challenge. For this purpose, my laboratory has created the Olfactory Receptor Database (ORDB). There are now more than 14,000 receptor genes and proteins in the database, representing the human, mouse, rat, dog, chimpanzee, invertebrate, fruit fly (Drosophila melanogaster), mosquito, bee, and roundworm (Cenorhabditis elegans) categories. (ORDB is available on the Web for inspection and download at the SenseLab Web site.)