18.13.03 · organismal-bio / sensory-systems

Chemosensation: taste, smell, and the Buck-Axel olfactory receptor revolution

shipped3 tiersLean: none

Anchor (Master): Chandrashekar et al., The Receptors and Cells for Mammalian Taste (Nature 444, 288-294, 2006); primary literature — Ikeda 1908, Pfaffmann 1941, Buck-Axel 1991 Cell 65:175, Malnic 1999 Cell 96, Chandrashekar 2000 Cell 100, Nelson 2001 Cell / 2002 Nature, Zou 2005 Nature, Tu 2018 PNAS, Bushdid 2014 Science, Caterina 1997 Nature

Intuition Beginner

Taste and smell are chemical senses — they detect molecules, not light or sound. Your tongue recognises only five basic qualities: sweet (energy-rich food), salty (electrolytes), sour (acids), bitter (toxins), and umami (protein-rich food). The Japanese chemist Kikunae Ikeda isolated the umami substance, glutamate, from kelp broth in 1908. Your tongue carries about 10,000 taste buds, and each bud holds 50 to 150 taste receptor cells, each tuned to one of those five qualities. Taste, in other words, is a small palette.

Smell is far larger. Humans carry about 400 working olfactory receptor genes — the largest gene family in our genome, and roughly 3 percent of all our genes. Each gene makes one receptor, and each receptor responds to a different set of odor molecules. In 1991, Linda Buck and Richard Axel discovered how this works: each olfactory neuron switches on exactly one receptor gene, and neurons that share a receptor route their signals to the same patch of the brain's olfactory bulb, building a spatial map of smell. The discovery won the 2004 Nobel Prize.

Smell is wired unusually, too. It is the only sense whose signals reach the brain's emotion and memory centers — the amygdala and hippocampus — without first passing through the brain's sensory relay station. That wiring is why a single whiff, of cinnamon or cut grass, can release a vivid memory. The phenomenon is the Proust effect, after the madeleine cake in Proust's novel.

Visual Beginner

The diagram lays out the four chemosensory channels covered in this unit. Taste buds on the tongue carry T1R, T2R, OTOP1, and ENaC receptors that report the five basic qualities. The olfactory epithelium in the nose holds olfactory sensory neurons, each expressing one of about 400 olfactory receptor genes; their axons converge by receptor type onto glomeruli in the olfactory bulb, producing a receptor-indexed map. Trigeminal nerve endings detect capsaicin (TRPV1), menthol (TRPM8), and wasabi (TRPA1). The vomeronasal organ is vestigial in humans.

The crucial contrast is between taste and smell at the molecular level. Taste uses a handful of receptors in a labelled-line code — one cell, one quality. Smell uses hundreds of receptors in a combinatorial code — one cell, one receptor, but one odor activates many cells and the brain reads the pattern.

Worked example Beginner

In early 2020, clinicians noticed a strange symptom: people with COVID-19 often lost their sense of smell even when their nose was not blocked. The cause turned out to be a specific cellular target, not the smell-sensing neurons themselves.

Step 1. The virus's spike protein needs two host molecules to enter a cell: ACE2 and TMPRSS2. Olfactory sensory neurons carry almost none of either.

Step 2. The sustentacular cells — support cells that wrap around the olfactory neurons in the epithelium — carry high levels of both ACE2 and TMPRSS2. The virus infects and kills them.

Step 3. With the support cells gone, the olfactory epithelium is disrupted and smell signals stop reaching the brain. Most patients recover within 2 to 4 weeks as the epithelium regrows. About 5 to 10 percent still have distorted smell (parosmia) or no smell (anosmia) months to years later.

What this tells us: smell is a core human sense, not a luxury. Losing it removes the flavor from food, the early warning of smoke and spoiled food, and a layer of emotional and social connection.

Check your understanding Beginner

Formal definition Intermediate+

Taste bud anatomy

A taste bud is a cluster of 50–150 taste receptor cells (TRCs) embedded in the epithelium of the tongue, palate, epiglottis, and upper esophagus. TRCs are modified epithelial cells with apical microvilli projecting into the taste pore; their average lifespan is 10–15 days and they turn over continuously from basal cells. Taste buds sit in papillae of three types: fungiform (anterior tongue), foliate (sides, posterior), and circumvallate (rear midline). Afferent taste fibres travel via cranial nerves VII (facial, anterior tongue), IX (glossopharyngeal, posterior tongue), and X (vagus, epiglottis) to the nucleus of the solitary tract (NTS) in the brainstem, then to the thalamus and primary gustatory cortex.

Definition (Five basic tastes and their receptors). Each taste quality is detected by a dedicated receptor family on a dedicated TRC type:

Sweet: T1R2 + T1R3 GPCR heterodimer (sugars, artificial sweeteners).
Umami: T1R1 + T1R3 GPCR heterodimer (L-amino acids, especially L-glutamate; isolated by Ikeda 1908, receptor cloned by Nelson 2002).
Bitter: T2R family of ~30 GPCRs (diverse toxins; Chandrashekar 2000).
Sour: OTOP1 proton channel ( $H^{+}$ ; Tu 2018).
Salty: ENaC epithelial sodium channel ( $Na^{+}$ ; a second high-salt channel is also implicated).

Sweet, umami, and bitter receptors signal through the G-protein gustducin $\to$ PLC $β$ 2 $\to$ IP $_{3}$ $\to$ $Ca^{2 +}$ release $\to$ TRPM5 cation channel $\to$ depolarisation $\to$ ATP release onto afferent fibres. Sour and salty rely on direct ion flux through OTOP1 and ENaC respectively.

Olfactory system

Definition (Olfactory epithelium and OSNs). The olfactory epithelium lines the nasal cavity and contains about 10 million olfactory sensory neurons (OSNs) — bipolar neurons with a single dendrite bearing cilia that project into the mucus. Each OSN expresses exactly one olfactory receptor (OR) gene from a repertoire of ~400 functional OR genes in humans (the largest gene family in the genome; ~3 percent of all genes); another ~800 OR loci are pseudogenes. ORs are GPCRs. Odorant binding activates the G-protein Golf $\to$ adenylyl cyclase III $\to$ cAMP $\to$ cyclic-nucleotide-gated (CNG) cation channel $\to$ depolarisation and action-potential firing.

Definition (One-neuron-one-receptor rule). Each olfactory sensory neuron expresses exactly one olfactory receptor allele. This rule, established by in situ hybridisation and transgenic OR-reporter mice (Chess 1994; Mombaerts 2004), is the load-bearing organisational principle of the peripheral olfactory system: it converts a receptor gene choice into a neuronal identity.

Definition (Combinatorial coding of olfaction). Each odorant activates a combination of ORs, and each OR responds to multiple odorants. The brain identifies an odor by the pattern of activated ORs (Malnic et al. 1999, Cell 96). This is why ~400 receptors can discriminate far more than 400 odors: the code is combinatorial, not one-receptor-one-odor.

Definition (Glomerular convergence). The axons of all OSNs expressing a given OR converge to a small number (typically two) of fixed glomeruli in each olfactory bulb (Zou et al. 2005). Mitral and tufted cells, the bulb's output neurons, read the glomerular map and project to the olfactory cortex, amygdala, and hippocampus — bypassing the thalamus at the first synapse. The thalamus is recruited later via mediodorsal-nucleus projections to orbitofrontal cortex for conscious odor perception.

Trigeminal and vomeronasal

Definition (Trigeminal chemosensation). The trigeminal nerve (cranial nerve V) carries chemosensory fibres detecting pungency: capsaicin activates TRPV1 ("hot"); menthol activates TRPM8 ("cool"); mustard and wasabi activate TRPA1. These are detected by free nerve endings in the oral and nasal mucosa, not by taste buds or olfactory neurons. The "heat" of chilli is a somatosensory percept, not a taste.

Definition (Vomeronasal organ). The vomeronasal organ (VNO) is an accessory olfactory structure that detects pheromones in most mammals via V1R and V2R receptors and the TRPC2 ion channel. In humans the VNO is vestigial, TRPC2 is a pseudogene, and the V1R/V2R repertoires are largely non-functional. Pheromone-like effects in humans, if any, are routed through the main olfactory system rather than the VNO.

Counterexamples to common slips

Taste and smell are not the same sense. They use different receptors (T1R/T2R/OTOP1/ENaC vs. ORs), different cranial nerves (VII/IX/X vs. I), and different central pathways. "Flavor" is the integrated percept of taste plus retronasal smell plus trigeminal input — not a single sense.
The "tongue map" is a myth. The old textbook diagram placing sweet at the tip and bitter at the back has no anatomical basis; all five taste qualities are detected across all papillae. The map traces to a misreading of Hänig 1901.
Capsaicin is not a taste. It is detected by TRPV1 on trigeminal free nerve endings, not by taste buds. The "heat" of chilli is a somatosensory, not gustatory, percept.
Humans do not have a poor sense of smell. Discrimination ability is large (Bushdid et al. 2014 estimated ~ $1 0^{12}$ distinguishable odors, though the estimate is contested). Tracking acuity is worse than the dog's, but discrimination of complex mixtures is comparable.
Olfaction does not entirely bypass the thalamus. The first synapse bypasses it (OSN $\to$ olfactory bulb $\to$ olfactory cortex), but conscious odor perception recruits the mediodorsal thalamus projecting to orbitofrontal cortex.
COVID-19 anosmia is usually not permanent. Roughly 90 percent of patients recover within weeks; about 5–10 percent have persistent dysfunction.

Key experiment: the Buck-Axel 1991 olfactory receptor discovery Intermediate+

Experiment (Buck & Axel 1991, Cell 65:175 — the olfactory receptor gene family). Working from the hypothesis that olfactory receptors belong to the G-protein-coupled receptor (GPCR) superfamily, Buck and Axel designed degenerate PCR primers targeting the conserved seven-transmembrane-domain motifs of known rhodopsin-like GPCRs and screened a rat olfactory epithelium cDNA library. They isolated a novel multigene family whose members are (i) expressed in olfactory epithelium, (ii) predicted to encode seven-transmembrane GPCRs, (iii) distributed in disjoint zones of the epithelium, and (iv) sufficiently diverse in sequence to encode receptors for a vast odorant panel.

Demonstration. The hypothesis underlying the experiment was that olfactory receptors must be both numerous (to match the perceptually discriminable odor count) and G-protein-coupled (because olfactory signal transduction was already known to involve cAMP and a G-protein, later named Golf). The conserved transmembrane motifs TM2, TM3, TM6, and TM7 of rhodopsin-family GPCRs provided anchor sequences for degenerate PCR primers. Application to rat olfactory cDNA yielded a probe that recognised a multi-band pattern on genomic Southern blots — the signature of a multigene family ^{[BuckAxel1991]}.

Three properties made the discovery decisive. First, in situ hybridisation showed that individual OR mRNAs are present in a small fraction of OSNs scattered across one of several rostral-caudal zones of the epithelium. Second, the encoded proteins were predicted seven-transmembrane GPCRs with short N-termini — incompatible with the long extracellular binding domains of metabotropic glutamate receptors, and consistent with a small-molecule odorant binding pocket formed within the transmembrane bundle. Third, sequence diversity in transmembrane domains 3, 4, and 5 was far higher than the diversity of rhodopsin or any other GPCR family, providing the variability needed to discriminate thousands of odorants. The Buck-Axel paper estimated the family size at 100–1000 genes; the human genome sequence later fixed it at ~800 OR loci, of which ~400 are functional, with the remainder pseudogenes ^{[BuckAxel1991]}.

Three corollaries of the discovery reshaped sensory biology. (i) The one-neuron-one-receptor rule. Each OSN expresses one OR allele. This was established by in situ hybridisation (single-OR probes label $\sim 0.1$ percent of OSNs) and by transgenic OR-reporter mice (Chess 1994; Mombaerts 2004). (ii) Combinatorial coding. A given odorant activates multiple ORs, and a given OR responds to multiple odorants; the percept is the activation pattern (Malnic et al. 1999, Cell 96). (iii) Receptor-specific glomerular convergence. OSN axons expressing the same OR converge to fixed glomeruli in the olfactory bulb, producing a two-dimensional receptor-indexed map (Zou et al. 2005, Nature). The discovery was recognised with the 2004 Nobel Prize in Physiology or Medicine.

Limitations. The complete odorant-to-OR mapping is still partial: heterologous expression of ORs is technically difficult, and only a small minority of human ORs have been deorphanised (matched to their cognate odorants). The "one trillion distinguishable odors" estimate of Bushdid et al. 2014 (Science), derived from mixture-discrimination experiments, has been contested on statistical and psychophysical grounds (McGann 2017; Gerkin & Castro 2015). The perceptual count is at minimum several thousand and at maximum unknown.

$□$

Bridge. The Buck-Axel discovery builds toward the general principle that a high-dimensional sensory world is mapped onto a discrete receptor code, and this is exactly the structure that appears again in 18.13.01 (the labelled-line theory of sensory coding) and in 29.03.04 (where Hubel and Wiesel mapped visual features onto cortical columns). The foundational reason the olfactory system is so expressive is that each OSN expresses exactly one receptor gene; the central insight — the one-neuron-one-receptor rule — identifies each glomerulus with a single receptor type. Putting these together, the gene family, the axonal convergence, and the combinatorial code form a three-stage architecture in which the olfactory bulb is a literal two-dimensional map of receptor space. The bridge is from the molecular genetics of a single gene family to the perceptual discrimination of many thousands of odors.

Exercises Intermediate+

Exercise 3 (medium, multiple choice).

The Buck-Axel 1991 discovery established that olfactory receptors form a multigene family of GPCRs. The "one-neuron-one-receptor rule" — each olfactory sensory neuron expresses exactly one OR allele — was a corollary established by subsequent work. Which observation most directly demonstrates the rule?

A. Each OR mRNA probe labels a small, scattered subset of OSNs in the olfactory epithelium, with no overlap between probes for different ORs.

B. The olfactory bulb contains about 2000 glomeruli.

C. Odorant binding activates the G-protein Golf.

D. The olfactory epithelium is divided into rostral-caudal expression zones.

Hint

The rule makes a prediction about overlap between OR expression patterns at the single-cell level.

Answer

A. If a single-OR probe labels $\sim 0.1$ percent of OSNs (one cell in a thousand) and two probes for different ORs label non-overlapping populations, then no OSN expresses two ORs simultaneously. This is the direct histological evidence for the one-neuron-one-receptor rule (Chess et al. 1994). The glomerulus count (B), Golf signalling (C), and zonal expression (D) are all real features of the olfactory system but do not directly establish one-OR-per-cell.

Exercise 4 (medium, symbolic).

Under the simplest model of combinatorial coding, each of $N$ functional olfactory receptors is either on or off in response to a given odorant, and two odorants are discriminable if and only if their receptor-activation patterns differ. Derive the maximum number of distinct patterns the periphery can in principle signal, and evaluate it for $N = 400$ . Then state one biological reason the actual number of discriminable odors is far smaller than this upper bound.

Hint

Each receptor is a binary variable; the activation pattern is an $N$ -bit vector. Consider whether the all-zero pattern corresponds to a real odor.

Answer

Each of $N$ receptors is a binary variable, so the periphery can signal $2^{N}$ distinct patterns. Excluding the all-zero pattern (no receptor activated — no odor detected), the maximum is $2^{N} - 1$ . For $N = 400$ , this is $2^{400} - 1 \approx 2.6 \times 1 0^{120}$ , a number astronomically larger than any plausible perceptual count.

Three biological reasons the actual count is far smaller: (i) not all $2^{N}$ patterns are physically realisable, because real odorants activate overlapping subsets of ORs and many patterns never occur; (ii) response amplitude carries information but is corrupted by noise, so similar patterns are confusable; (iii) central processing identifies some distinct patterns as the same odor (mixture invariance). Bushdid et al. 2014's $\sim 1 0^{12}$ estimate, even if taken at face value, lies $\sim 108$ orders of magnitude below this combinatorial ceiling.

Exercise 5 (medium, short answer).

Explain why SARS-CoV-2 causes anosmia despite olfactory sensory neurons themselves carrying almost none of the ACE2 receptor. What cell type is the actual viral target, and why does its infection abolish smell?

Hint

Consider which cells in the olfactory epithelium express ACE2 and TMPRSS2, and what structural role they play.

Answer

Olfactory sensory neurons (OSNs) carry almost no ACE2 or TMPRSS2, the two surface molecules SARS-CoV-2 requires for cell entry. Sustentacular cells, the support cells that wrap around OSNs in the olfactory epithelium and maintain the mucus ionic environment OSNs depend on, express both ACE2 and TMPRSS2 at high levels. The virus infects and kills sustentacular cells, disrupting the olfactory epithelium and abolishing OSN function without directly infecting OSNs themselves. This accounts for the abrupt onset of anosmia in COVID-19 (often without nasal congestion), the typically rapid recovery as the epithelium regenerates over 2–4 weeks, and the persistent dysfunction in $\sim 5$ – $10$ percent of patients. The COVID anosmia pandemic provided the largest natural experiment in the clinical importance of the chemosensory system.

Exercise 6 (medium, short answer).

State the "tongue map" hypothesis (sweet at the tip, bitter at the back, sour at the sides) and refute it with reference to taste-bud anatomy. Name the historical origin of the map.

Hint

Inspect what each papilla type contains. Consider the 1901 German paper by Hänig and how its data were later misread.

Answer

The "tongue map" holds that the five basic tastes are localised to different tongue regions: sweet at the tip, sour at the sides, bitter at the back, salty at the edges. This is anatomically wrong. Each of the three papilla types (fungiform, foliate, circumvallate) carries taste buds, and each taste bud contains receptor cells for all five taste qualities — every region of the tongue can detect every basic taste. The map traces to a misreading of David Hänig's 1901 dissertation, in which he had measured small regional differences in detection threshold (not exclusive localisation). The misconception was propagated by textbooks throughout the 20th century before being debunked by electrophysiological recordings showing that single taste nerve fibres respond to multiple taste qualities across all tongue regions.

Exercise 7 (hard, symbolic).

Under the simplest combinatorial model, each odorant activates exactly $k$ of $N$ olfactory receptors chosen uniformly at random. Derive the probability that two independent odorants activate exactly the same receptor subset. For $N = 400$ and $k = 10$ , evaluate this probability and comment on what it implies for the uniqueness of receptor codes.

Hint

The total number of $k$ -subsets of an $N$ -set is the binomial coefficient $(k N)$ . The first odorant picks one such subset; the second matches it with probability $1/ (k N)$ .

Answer

The number of distinct $k$ -subsets of an $N$ -element set is $(k N) = \frac{N !}{k ! ( N - k )!}$ . The first odorant picks one such subset. The probability that the second odorant activates exactly the same subset is

P (match) = \frac{1}{( k N )} .

For $N = 400$ , $k = 10$ :

(10 400) = \frac{400 \cdot 399 \cdot 398 \dots 391}{10 !} \approx 2.0 \times 1 0^{19},

so $P (match) \approx 5 \times 1 0^{- 20}$ — vanishingly small. The implication: under even a simple combinatorial code with $k = 10$ ORs per odorant, every odorant in any realistic panel has a unique receptor-activation subset. The discriminability ceiling is not receptor-code collision but central-processing limits (mixture invariance, intensity coding, noise) — which is why the perceived-discrimination count is many orders of magnitude below $(10 400)$ .

Exercise 8 (hard, short answer).

Bushdid et al. 2014 (Science 343:1370) estimated that humans can discriminate "more than 1 trillion" olfactory stimuli. The methodology and conclusion were subsequently critiqued (Gerkin & Castro 2015; McGann 2017). Summarise (a) the experimental design, (b) the key assumption that drives the trillion-odor extrapolation, and (c) the principal methodological objection to that extrapolation.

Hint

The experiment used mixtures of overlapping odorant sets. The extrapolation depends on how mixture-discrimination performance is converted into a count of distinguishable odor "stimuli".

Answer

(a) Experimental design. Bushsid et al. prepared mixtures of 10, 20, or 30 components drawn from a 128-odorant panel. On each trial, subjects were presented with three mixtures (two identical, one different) and asked to identify the odd one out. Discrimination performance was measured as a function of mixture-component overlap.

(b) Key assumption. The fraction of mixture pairs that subjects reliably discriminated was treated as a metric on odor-percept space, and the size of this space was extrapolated by computing the volume of mixture-pair distances that exceed the discrimination threshold. The result, $\sim 1 0^{12}$ distinguishable stimuli, follows from extrapolating this metric across all possible mixtures of the 128-odorant panel.

(c) Principal objection. The extrapolation depends sensitively and unstably on the assumed metric and on the threshold criterion; small changes in either parameter shift the estimate by many orders of magnitude (Gerkin & Castro 2015 showed the same data are consistent with estimates from $\sim 1 0^{3}$ to $\sim 1 0^{14}$ ). Additionally, treating "mixture component overlap" as a psychophysical metric is an unvalidated assumption. The lower bound on human olfactory discrimination is conservatively in the thousands; the upper bound remains genuinely unknown. The trillion-odor figure should be read as a lower-bound estimate from one experimental design, not as an established property of the olfactory system.

Advanced results Master

Theorem 1 (Buck-Axel olfactory receptor family). Olfactory receptors form a rhodopsin-like GPCR multigene family expressed in disjoint zones of the olfactory epithelium; each OSN expresses exactly one OR allele, and the family size is ~1000 genes in mouse and ~400 functional genes in humans ^{[BuckAxel1991]}. The discovery opened the molecular era of olfaction and was recognised with the 2004 Nobel Prize.

Theorem 2 (Combinatorial coding, Malnic 1999). Each odorant activates a combination of ORs, and each OR responds to multiple odorants; perceptual identity is the activation pattern, not a single receptor's output. The theorem generalises the labelled-line model: the olfactory code is a sparse combinatorial pattern rather than a one-receptor-one-odor mapping ^[Malnic1999].

Theorem 3 (Sweet and umami are T1R heterodimers). Sweet is the T1R2+T1R3 GPCR heterodimer; umami is the T1R1+T1R3 heterodimer. Knockout of T1R3 abolishes both sweet and umami preference; knockout of T1R2 selectively abolishes sweet, of T1R1 selectively abolishes umami. Nelson et al. 2001 (Cell) cloned T1R1+T1R3 as the umami receptor; Nelson et al. 2002 (Nature) cloned T1R2+T1R3 as the sweet receptor. Chandrashekar et al. 2006 (Nature 444:288) synthesised the cellular and molecular account of taste.

Theorem 4 (Bitter is the T2R family). Bitter is detected by a family of ~30 T2R GPCRs, each tuned to a different class of bitter-tasting compounds (many of them plant-derived toxins); a single TRC expresses many T2R family members and reports aggregate bitterness to the afferent nerve. Chandrashekar et al. 2000 (Cell 100:703) established the T2R family.

Theorem 5 (Sour is OTOP1). Sour taste is mediated by the OTOP1 proton channel, which conducts $H^{+}$ across the TRC apical membrane into the cell. Otop1 knockout mice lose sour responses but retain all other taste qualities ^[Tu2018].

Theorem 6 (Receptor-specific glomerular convergence). OSN axons expressing a given OR converge to fixed glomeruli in each olfactory bulb, producing a receptor-indexed two-dimensional map. Genetic swap of the expressed OR relocates the glomerular target (Mombaerts et al. 1996; Wang et al. 1998; Zou et al. 2005). The olfactory bulb is therefore a literal map of receptor space.

Theorem 7 (Trigeminal chemosensation is a distinct modality). Capsaicin, menthol, and allyl isothiocyanate (wasabi) act on TRPV1, TRPM8, and TRPA1 respectively, located on trigeminal free nerve endings in the oral and nasal mucosa. These channels are absent from taste buds and olfactory sensory neurons, so the perceptual qualities of "hot", "cool", and "pungent" are somatosensory, not gustatory or olfactory (Caterina et al. 1997, Nature 389:816; McKemy et al. 2002, Nature 416:52).

Empirical result (the trillion-odor estimate, contested). Bushdid et al. 2014 (Science 343:1370) estimated humans can discriminate more than $1 0^{12}$ olfactory stimuli. The methodology has been contested: Gerkin & Castro 2015 (eLife) and McGann 2017 (Science 356:597) showed the estimate is sensitive to metric and threshold choices and can shift by many orders of magnitude. The lower bound on human olfactory discrimination is at minimum several thousand; the upper bound remains open.

Synthesis. The foundational reason the olfactory system can discriminate far more odors than it has receptors is that each odorant activates a combination of ORs and the brain reads the pattern; this is exactly the combinatorial-coding principle Malnic established in 1999, and it generalises to every sensory system that maps a high-dimensional stimulus space onto a lower-dimensional neural surface. The central insight is that the one-neuron-one-receptor rule, combined with axonal convergence to glomeruli, identifies each glomerulus with a single receptor type, so the olfactory bulb becomes a two-dimensional receptor-coordinate map.

Putting these together with the taste system, the olfactory combinatorial code stands in instructive contrast to the taste labelled-line code: taste uses one cell, one quality, with five receptor families reporting five qualities; olfaction uses one cell, one receptor, with ~400 receptors reporting a combinatorial space. The bridge is that the same molecular logic — GPCR signalling via cyclic-nucleotide gating — appears again in 18.13.01 (sensory pathways generally), in 18.13.02 (where the MET channel uses a structurally analogous transduction cascade), and in 29.03.04 (Hubel-Wiesel cortical maps), and the pattern recurs in every sensory modality that maps a large stimulus space onto a smaller receptor repertoire.

Full proof set Master

Proposition (Combinatorial-coding capacity upper bound). If the olfactory receptor repertoire contains $N$ functional receptors and each receptor responds binary (on/off) to any given odorant, then the number of distinct receptor-activation patterns the periphery can in principle signal is at most $2^{N} - 1$ (the nonempty subsets). Under a noise-free model and ignoring per-receptor response amplitude, this gives an upper bound on the number of discriminable odors; realistic estimates that account for overlap and amplitude coding reduce the effective count by many orders of magnitude.

Proof. Each of the $N$ receptors is a binary variable; the activation pattern of any odorant is an $N$ -bit vector in ${0, 1}^{N}$ . The set of all $N$ -bit vectors has cardinality $2^{N}$ , of which one (the all-zeros vector $00 \dots 0$ ) corresponds to no odorant detected. Hence at most $2^{N} - 1$ distinct nonempty patterns are available for signalling. For the human repertoire with $N \approx 400$ functional receptors, the combinatorial upper bound is $2^{400} - 1 \approx 2.6 \times 1 0^{120}$ .

This ceiling is not the discriminable-odor count. Three reductions apply. (i) Real odorants activate overlapping subsets of ORs, so many of the $2^{N}$ patterns never occur and many are nearly indistinguishable in practice. (ii) Response amplitude carries information but is corrupted by receptor and synaptic noise, so similar patterns are confusable. (iii) Central processing identifies some distinct peripheral patterns as the same odor under mixture-invariance computations. Bushdid et al. 2014's $\sim 1 0^{12}$ estimate, even at face value, lies $\sim 108$ orders of magnitude below $2^{400}$ , confirming that the combinatorial ceiling is not the binding constraint. The proposition isolates why combinatorial coding is so powerful in principle: the discriminable count grows exponentially, not linearly, with receptor-repertoire size. $□$

Proposition (Receptor-specific glomerular convergence). Under the one-neuron-one-receptor rule and the axonal-convergence principle (Zou et al. 2005), all olfactory sensory neurons expressing a given olfactory receptor $R$ converge their axons to a small number of fixed glomeruli in each olfactory bulb. Therefore each glomerulus is a labelled-line readout of the activation of a single receptor type, and the olfactory bulb's glomerular array is a two-dimensional map indexed by receptor identity.

Proof. By the one-neuron-one-receptor rule (Chess et al. 1994; Serizawa et al. 2003), each OSN chooses one OR allele for expression and silences all others via negative-feedback regulation downstream of OR-protein expression. Axon guidance then sorts OSN axons by their expressed OR: genetic swap of the OR gene shifts the glomerular target (Mombaerts et al. 1996; Wang et al. 1998). Hence the glomerulus $G_{R}$ to which an OSN projects is a function of the OR $R$ it expresses, not of its position in the olfactory epithelium.

It follows that the activation of glomerulus $G_{R}$ is proportional to the summed activity of all OSNs expressing receptor $R$ :

A (G_{R}) = i : OSN_{i} expresses R \sum a_{i},

where $a_{i}$ is the firing rate of OSN $i$ . The bulbar surface is therefore a two-dimensional receptor-indexed map, with each glomerulus serving as a labelled-line readout of a single receptor type. The mitral and tufted cells that innervate each glomerulus inherit this receptor specificity and relay it to the olfactory cortex. $□$

Connections Master

Sensory systems survey 18.13.01 is the chapter anchor and prerequisite: this unit deepens the chemosensory material sketched there into a full account of taste-receptor molecular biology, the Buck-Axel olfactory receptor discovery, and combinatorial coding. The labelled-line theory of sensory coding that 18.13.01 introduces is the framework the present unit extends and contrasts with olfactory combinatorial coding.
Hair cell mechanotransduction and cochlear tuning 18.13.02 is the comparative chapter peer. The cochlear hair cell uses a direct mechanical gate (tip-link $\to$ TMC1 MET channel) for transduction, while taste and olfactory receptors use GPCR signalling via cyclic-nucleotide gating. The two units together exhibit the two principal transduction strategies in sensory biology: direct ion gating (mechanosensation) and G-protein-coupled cascades (chemosensation).
Hubel-Wiesel visual cortex 29.03.04 is the canonical example of a sensory-system spatial map. The olfactory bulb's glomerular array is the chemosensory analogue of the visual cortex's orientation columns: both convert a high-dimensional receptor input into a two-dimensional cortical/bulbar map of a stimulus feature. The combinatorial coding of olfaction generalises the labelled-line logic that Hubel and Wiesel established for simple cells in V1.
Neurodegenerative disease 35.03.05 carries one of the most striking clinical correlates of chemosensory dysfunction: olfactory loss is an early prodromal sign of both Parkinson's disease and Alzheimer's disease, often preceding motor or cognitive symptoms by years. The olfactory epithelium and bulb are among the first sites of $α$ -synuclein and tau pathology, which is why smell testing is being investigated as a cheap, non-invasive early-screening tool for neurodegeneration.

Historical & philosophical context Master

Modern chemosensation begins with Kikunae Ikeda's 1908 isolation of monosodium glutamate from kelp broth (Journal of the Chemical Society of Tokyo 30, 820–836) ^[Ikeda1908]. Ikeda named the new taste quality "umami" and reasoned, by analogy with sweet, sour, salty, and bitter, that it must have a dedicated receptor — a hypothesis that took 94 years to confirm at the molecular level (Nelson et al. 2002). Pfaffmann's 1941 "fiber theory" of taste coding, in which single afferent nerve fibres carry multiple taste qualities, was the dominant mid-century account of taste coding and was superseded only when molecular receptors became identifiable in the 2000s.

The molecular era of olfaction opened with Linda Buck and Richard Axel's 1991 paper in Cell 65:175 ^{[BuckAxel1991]}, which used degenerate PCR based on the conserved seven-transmembrane-domain motifs of rhodopsin-family GPCRs to clone the olfactory receptor gene family. The paper was the founding account of how the periphery of the olfactory system is organised: a multigene family, expressed one receptor per neuron, with each neuron projecting to a receptor-specific glomerulus. The combinatorial-coding model followed (Malnic et al. 1999, Cell 96:713), and the glomerular map was visualised genetically (Mombaerts et al. 1996; Zou et al. 2005). Buck and Axel shared the 2004 Nobel Prize in Physiology or Medicine.

The molecular account of taste was a parallel program. Chandrashekar et al. 2000 (Cell 100:703) cloned the T2R bitter receptor family; Nelson et al. 2001 (Cell 106:381) and 2002 (Nature 416:199) cloned the T1R umami and sweet receptors; Chandrashekar et al. 2006 (Nature 444:288) synthesised the cellular and molecular basis of taste; and Tu et al. 2018 (Science 359:1047) identified OTOP1 as the long-sought sour proton channel. The COVID-19 pandemic of 2020 added an unexpected chapter: sudden anosmia emerged as a distinctive early symptom, and the mechanism (sustentacular cell infection via ACE2, sparing the OSNs themselves) demonstrated that the supporting architecture of the olfactory epithelium is as essential to smell as the sensory neurons it wraps.

Bibliography Master

@article{Ikeda1908,
  author = {Ikeda, Kikunae},
  title = {New seasonings (umami no chishiki)},
  journal = {Journal of the Chemical Society of Tokyo},
  volume = {30},
  year = {1908},
  pages = {820--836},
  note = {Translated in Chem. Senses 27, 843--851 (2002)},
}

@article{BuckAxel1991,
  author = {Buck, Linda and Axel, Richard},
  title = {A novel multigene family may encode odorant receptors: a molecular basis for odor recognition},
  journal = {Cell},
  volume = {65},
  number = {1},
  year = {1991},
  pages = {175--187},
}

@article{Malnic1999,
  author = {Malnic, Bettina and Hirono, Junzo and Sato, Takaaki and Buck, Linda B.},
  title = {Combinatorial receptor codes for odors},
  journal = {Cell},
  volume = {96},
  number = {5},
  year = {1999},
  pages = {713--723},
}

@article{Chandrashekar2000,
  author = {Chandrashekar, Jayaram and Mueller, Ken L. and Hoon, Mark A. and Adler, Eli and Feng, Luxin and Guo, Wei and Zuker, Charles S. and Ryba, Nicholas J. P.},
  title = {T2Rs function as bitter taste receptors},
  journal = {Cell},
  volume = {100},
  number = {6},
  year = {2000},
  pages = {703--711},
}

@article{Nelson2001,
  author = {Nelson, Grace and Hoon, Mark A. and Chandrashekar, Jayaram and Zhang, Yifeng and Ryba, Nicholas J. P. and Zuker, Charles S.},
  title = {An amino-acid taste receptor},
  journal = {Nature},
  volume = {416},
  year = {2002},
  pages = {199--202},
}

@article{Nelson2002,
  author = {Nelson, Grace and Chandrashekar, Jayaram and Hoon, Mark A. and Feng, Luxin and Zhao, Grace and Ryba, Nicholas J. P. and Zuker, Charles S.},
  title = {An amino-acid taste receptor},
  journal = {Nature},
  volume = {416},
  year = {2002},
  pages = {199--202},
}

@article{Chandrashekar2006,
  author = {Chandrashekar, Jayaram and Hoon, Mark A. and Ryba, Nicholas J. P. and Zuker, Charles S.},
  title = {The receptors and cells for mammalian taste},
  journal = {Nature},
  volume = {444},
  year = {2006},
  pages = {288--294},
}

@article{Zou2005,
  author = {Zou, Dong-Jing and Chesler, Adam and Firestein, Stuart},
  title = {How the olfactory bulb got its glomeruli: a just so story?},
  journal = {Nature Neuroscience},
  volume = {8},
  year = {2005},
  pages = {1505--1510},
}

@article{Tu2018,
  author = {Tu, Yu-Heng and Cooper, Andrew J. and Teng, Bochuan and Rao, Rucha B. and Gordon, Emily D. and Chang, Yann and Nelson, Grace and Carlson, Joseph G. and Kinnamon, John C. and Liman, Emily R.},
  title = {An evolutionarily conserved gene family encodes proton-selective ion channels},
  journal = {Science},
  volume = {359},
  number = {6380},
  year = {2018},
  pages = {1047--1050},
}

@article{Bushdid2014,
  author = {Bushdid, Caroline and Magnasco, Marcelo O. and Vosshall, Leslie B. and Keller, Andreas},
  title = {Humans can discriminate more than 1 trillion olfactory stimuli},
  journal = {Science},
  volume = {343},
  number = {6177},
  year = {2014},
  pages = {1370--1372},
}

@article{Caterina1997,
  author = {Caterina, Michael J. and Schumacher, Mark A. and Tominaga, Makoto and Rosen, Tobias A. and Levine, Jon D. and Julius, David},
  title = {The capsaicin receptor: a heat-activated ion channel in the pain pathway},
  journal = {Nature},
  volume = {389},
  year = {1997},
  pages = {816--824},
}

@article{GerkinCastro2015,
  author = {Gerkin, Richard C. and Castro, Jason B.},
  title = {The number of olfactory stimuli that humans can discriminate is still unknown},
  journal = {eLife},
  volume = {4},
  year = {2015},
  pages = {e08127},
}

@article{McGann2017,
  author = {McGann, John P.},
  title = {Poor human olfaction is a 19th-century myth},
  journal = {Science},
  volume = {356},
  year = {2017},
  pages = {eaam7263},
}

Prerequisites

18.13.01

Tier anchors

beginner: Bear, Connors & Paradiso, Neuroscience: Exploring the Brain, 4th ed. (Lippincott, 2020), Ch. 15 (Chemical Senses); Campbell Biology 12th ed. Ch. 50 (Sensory and Motor Mechanisms); Rachel Herz, The Scent of Desire (2007)
intermediate: Kandel et al., Principles of Neural Science, 6th ed. (McGraw-Hill, 2021), Ch. 32 (Smell and Taste); Purves et al., Neuroscience, 6th ed. (Oxford, 2018), Ch. 15 (Chemical Senses); Buck, Unraveling the Sense of Smell (Nobel Lecture 2004)
master: Chandrashekar et al., The Receptors and Cells for Mammalian Taste (Nature 444, 288-294, 2006); primary literature — Ikeda 1908, Pfaffmann 1941, Buck-Axel 1991 Cell 65:175, Malnic 1999 Cell 96, Chandrashekar 2000 Cell 100, Nelson 2001 Cell / 2002 Nature, Zou 2005 Nature, Tu 2018 PNAS, Bushdid 2014 Science, Caterina 1997 Nature

References

Buck, L. & Axel, R. — A novel multigene family may encode odorant receptors: a molecular basis for odor recognition (Cell 65, 175-187, 1991) · The Buck-Axel paper: degenerate-PCR discovery of the olfactory receptor gene family; the founding molecular account of smell
Malnic, B., Hirono, J., Sato, T. & Buck, L. B. — Combinatorial receptor codes for odors (Cell 96, 713-723, 1999) · Combinatorial-coding model: each odorant activates multiple ORs and each OR responds to multiple odorants; the percept is the activation pattern
Chandrashekar, J., Hoon, M. A., Ryba, N. J. P. & Zuker, C. S. — The receptors and cells for mammalian taste (Nature 444, 288-294, 2006) · Synthesis review: T1R sweet/umami, T2R bitter, the cellular and molecular basis of taste
Nelson, G., Chandrashekar, J., Hoon, M. A., Feng, L., Zhao, G., Ryba, N. J. P. & Zuker, C. S. — An amino-acid taste receptor (Nature 416, 199-202, 2002) · Cloning of the T1R1+T1R3 umami receptor; knockout mouse phenotypes establishing T1R heterodimer function
Ikeda, K. — New seasonings (jo-umiami) (Journal of the Chemical Society of Tokyo 30, 820-836, 1908; translated in Chem. Senses 27, 2002) · Ikeda's isolation of monosodium glutamate from kelp broth and his coining of 'umami' as the fifth basic taste
Zou, D. J., Chesler, A. & Firestein, S. — How the olfactory bulb got its glomeruli: a just so story? (Nature Neuroscience 8, 1505-1508, 2005) · Receptor-specific axonal convergence and the formation of the olfactory-bulb glomerular map
Tu, Y.-H., Cooper, A. J., Teng, B., et al. — An evolutionarily conserved gene family encodes proton-selective ion channels (Science 359, 1047-1050, 2018) · Identification of OTOP1 as the sour-taste proton channel
Bushdid, C., Magnasco, M. O., Vosshall, L. B. & Keller, A. — Humans can discriminate more than 1 trillion olfactory stimuli (Science 343, 1370-1372, 2014) · The trillion-odor discrimination estimate (with subsequent methodological critique)

Estimated time

beginner: 25m
intermediate: 50m
master: 85m