Module 6: Chemoreception — Olfaction, Gustation & Pheromones

The olfactory receptor (OR) family, discovered by Linda Buck and Richard Axel in 1991 (Nobel Prize in Physiology or Medicine, 2004), is the largest gene family in the mammalian genome. Each OR is a seven-transmembrane G-protein-coupled receptor (GPCR) expressed in the cilia of olfactory sensory neurons (OSNs) in the main olfactory epithelium (MOE).

Niimura (2014) catalogued OR repertoire sizes across vertebrates and uncovered extreme ecological variation. Humans possess approximately 800 OR genes of which only 396 are functional (∼56% pseudogenes), whereas dogs retain 811 functional genes and the African elephant genome encodes roughly 2000 ORs—the largest known repertoire of any mammal. At the other extreme, cetaceans such as the bottlenose dolphin carry only about 50 OR genes: echolocation and an aquatic lifestyle have made airborne olfaction nearly useless, and the genes have collapsed into pseudogenes. The monotreme platypus retains roughly 130 functional ORs, reflecting its semiaquatic foraging niche.

OR Signaling Cascade

Upon odorant binding, an OR activates the olfactory-specific G-protein \(G_{\text{olf}}\), which stimulates adenylyl cyclase III (ACIII) to produce cAMP. The second messenger opens a cyclic-nucleotide-gated (CNG) cation channel, admitting Ca\(^{2+}\) and Na\(^+\) into the ciliary lumen. Ca\(^{2+}\)-activated Cl\(^{-}\) channels (ANO2/TMEM16B) then produce a large amplifying efflux of chloride, giving the OSN an unusually high input resistance and remarkable single-molecule sensitivity:

Shape vs. Vibration Controversy

Mainstream biochemistry holds that OR ligand specificity is determined by the three-dimensional shape of the odorant molecule binding within a hydrophobic pocket of the transmembrane helices—the classical lock-and-key or \(induced-fit\)paradigm. The minority vibrational theory proposed by Turin (1996) argues that receptors detect the inelastic electron-tunneling spectrum of the bound ligand (\(\nu \sim 10^{13}\) Hz). Experimental tests using deuterated odorants have produced conflicting results (Franco 2011; Keller & Vosshall 2004), and the shape model remains the dominant framework.

Each olfactory sensory neuron expresses, with extraordinary precision, exactly one OR allele via the stochastic choice and feedback mechanism described by Chess et al. (1994). Axons from OSNs expressing the same OR converge onto one or two glomeruli in the olfactory bulb (Mombaerts et al. 1996), producing a clean spatial map of receptor activation. A single odorant typically binds a subset of ORs with varying affinities, and a given OR binds many odorants; the brain reads the combinatorial pattern as a high-dimensional sparse code.

Malnic, Hirono, Sato, and Buck (1999) applied calcium imaging to individual OSNs and demonstrated that distinct odorants activate distinct combinations of ORs, while one OR can respond to multiple related molecules. The theoretical discrimination capacity of such a code grows combinatorially:

The controversial estimate by Bushdid et al. (2014)that humans can discriminate at least 1 trillion olfactory stimuli was sharply critiqued by Meister (2015), who argued that the statistical extrapolation confounds receptor combinatorial capacity with perceptual discriminability. The realistic range lies between\(10^4\) and \(10^{12}\) depending on definition, but the general principle—that combinatorial codes with even a modest OR repertoire produce enormous discrimination capacity—is not disputed.

Hill-Function Receptor Response

A useful quantitative model treats each (OR, odorant) pair as a ligand-binding equilibrium with Hill coefficient \(n\) and dissociation constant \(K_D\):

The affinity distribution \(\log K_{D,ij}\) is approximately Gaussian (Si et al. 2019 — fly ORs), so most odorants have weak but nonzero cross-talk with many receptors. This drives the characteristic sparse, distributed code observed in optical imaging of the mammalian olfactory bulb (Friedrich & Korsching 1997; Wachowiak & Cohen 2001).

Insect olfactory receptors are evolutionarily unrelated to mammalian ORs: they are heteromeric ligand-gated ion channels, not GPCRs. Each tuning receptor (OR_x) assembles with the obligate co-receptor Orco (odorant receptor coreceptor, Larsson et al. 2004). Binding of the odorant opens the OR⋅Orco channel complex directly, giving insects millisecond-latency responses ideally suited to tracking turbulent pheromone filaments.

Benton, Vannice, Gomez-Diaz and Vosshall (2009) identified a second olfactory family in Drosophila: the ionotropic receptors (IRs), evolutionarily derived from ionotropic glutamate receptors. IRs are expressed in coeloconic sensilla and tune to acids, amines, and humidity (Enjin et al. 2016). A third, smaller family of gustatory receptors (GRs) detects CO\(_2\) via Gr21a/Gr63a.

Bombyx mori: One Molecule, One Behavior

Kaissling (1971) showed that a male silkmoth Bombyx mori initiates wing-flutter behavior after detecting as few as 1–10 molecules of the female pheromone bombykol (10E,12Z-hexadecadien-1-ol) per second striking its feathered antennae. The receptor BmOR-1 was identified by Sakurai et al. (2004). Sensitivity follows from the combination of an enormous antennal surface area covered in sensilla trichodea and the amplifying OR⋅Orco ion-channel cascade:

Moth antennae function as efficient molecular sieves with capture probability approaching unity per molecular hit on a sensillum pore. Species-specific pheromone blends (Heliothis, Spodoptera) encode mate choice in ratios with precision better than 1%: downwind males can distinguish 80:20 from 85:15 binary mixtures using contrast encoding in projection neurons of the macroglomerular complex (Hansson 1995).

cVA and the Pheromone Circuit

In Drosophila melanogaster, the male pheromone cVA (cis-vaccenyl acetate) is detected by Or67d expressed in T1 neurons of at1 trichoid sensilla (Kurtovic, Widmer & Dickson 2007). cVA activates the female-specific fruitless circuit in DA1 glomerulus and modulates male–male aggression; same input, opposite behavior via dimorphic downstream wiring. The cVA circuit is a canonical example of how a labeled-line pheromone channel coexists with the combinatorial general-odor code.

The vomeronasal organ (VNO), also called Jacobson’s organ, is a paired chemosensory structure in the nasal septum of many tetrapods, specialized for detecting non-volatile pheromones and kairomones. Mouse VNO sensory neurons express one of two distinct GPCR families: V1Rs (small, lipophilic ligands; Dulac & Axel 1995) and V2Rs (larger peptides and major urinary proteins; Matsunami & Buck 1997). Both classes converge on the TRPC2 ion channel, which is pheromone-specific: TRPC2\(^{-/-}\) male mice display indiscriminate mounting behavior toward both sexes (Stowers et al. 2002).

In humans the VNO is anatomically vestigial: a small epithelial pocket lacking functional sensory neurons or a distinct accessory olfactory bulb. Human TRPC2 is a pseudogene. Subjective reports of “pheromonal” effects in our species are almost certainly mediated through the main olfactory epithelium rather than the VNO.

Snake Tongue-Flicking

Squamate reptiles (snakes and lizards) sample volatile and nonvolatile compounds by flicking a bifid tongue and transferring molecules to a pair of ducts leading directly to the VNO (Halpern 1992). The bifurcated tongue tips deliver bilaterally separated samples, allowing tropotaxis—the reptilian brain compares left-tip and right-tip concentrations in a single flick to estimate the gradient direction (Schwenk 1994). This enables garter snakes to follow a prey trail and pit vipers to track wounded prey.

The classical dogma that birds have poor olfaction was overturned by a century of careful work. Bang (1960) demonstrated that the turkey vulture (Cathartes aura) tracks the decomposition gas ethanethiol (CH\(_3\)CH\(_2\)SH) to carcasses beneath dense forest canopy where vision is useless. Procellariiform seabirds (albatrosses, petrels, shearwaters) orient over thousands of kilometers of ocean toward patches of dimethyl sulfide (DMS) and trimethylamine emitted by phytoplankton and decomposing krill (Nevitt 2008). The kiwi (Apteryx) probes soil with nostril-equipped bill tips and has the largest olfactory bulb relative to brain of any bird (Martin et al. 2007). Chickens carry roughly 550 functional OR genes, comparable to many mammals.

Fish olfactory epithelium responds primarily to amino acids (food), bile acids (conspecifics), and sex steroids (reproduction). Hasler (1951) and Dittman et al. (1996) demonstrated that salmon (Oncorhynchus) imprint on the distinctive amino-acid signature of their natal stream during the smolt stage and use this olfactory memory to return across thousands of kilometers of ocean years later. Catfish (Ictalurus) carry taste buds over their entire body surface, effectively swimming inside a chemosensory receptor carpet.

Shark Olfactory Acuity

Atema, Brodnanski, and Johnson (1980) and subsequent work showed that the bonnethead shark (Sphyrna tiburo) tracks an amino-acid odor plume by asymmetric sampling of laterally-separated nostrils: the shark turns toward the nostril that registered first, not strongest, using interchirp timing differences as short as\(\Delta t \sim 100\) ms. Modern behavioral thresholds are remarkably low: bonnetheads respond to amino acids at concentrations of \(10^{-9}\) mol/L.

Mammalian taste is mediated by two GPCR families plus ion-channel-based sour and salty modalities. TAS1R receptors form heterodimers: T1R2+T1R3 is the sweet receptor, T1R1+T1R3 detects umami (L-glutamate, 5′-ribonucleotides). The TAS2R family (∼25 receptors in humans, >30 in mice) detects bitter compounds—typically plant alkaloids and other potential toxins—via a low-affinity, broad-specificity set of GPCRs.

Jiang et al. (2012) showed that all Felidae(domestic cat, lion, tiger, cheetah) carry a pseudogenized TAS1R2 and therefore cannot taste sweet: a natural consequence of their obligate hypercarnivorous diet. Similar sweet blindness appears independently in sea lions, dolphins, and spotted hyenas. Conversely, giant pandas, whose diet is almost exclusively bamboo, have lost the umami receptor T1R1. Gene loss tracks dietary specialization with striking fidelity.

Insect gustation relies on GRs expressed in taste bristles on the labellum, legs, and wings. Honeybees (Apis mellifera) taste sugars through receptors in their front tarsi: a foraging bee that steps on a nectar droplet extends its proboscis within 200 ms. Flies and butterflies show similar tarsal gustation.

Star-Nosed Mole Underwater Sniffing

Catania (2006) discovered that the star-nosed mole (Condylura cristata) produces air bubbles in water and rapidly re-inhales them, a technique now known as underwater sniffing. Each bubble briefly contacts an object and picks up volatiles; the mole re-inhales within 100–200 ms, transferring the odorant-laden air to the main olfactory epithelium. This is the first documented case of a mammal using true olfaction underwater. Similar bubble-sniffing has since been observed in water shrews.

Polar Bear Long-Range Olfaction

Polar bears (Ursus maritimus) detect seal breathing holes from up to 32 km away and track odor trails across open sea ice. Their large ethmoid turbinates greatly increase epithelial surface area, and behavioral experiments show they locate buried food at concentrations \(\sim 10^{-12}\) of the detection level of a human observer.

Platypus Bill: Chemistry Meets Electricity

The platypus bill integrates >40,000 mechanoreceptors, >40,000 electroreceptors, and distributed chemoreceptors in a multimodal array. Platypuses hunt with eyes, ears, and nostrils closed: prey detection in murky water relies entirely on bill chemoreception, mechanoreception, and electrical signals from invertebrate muscle contractions.

Odor plumes in natural environments are not smooth Gaussian clouds: turbulence shreds them into intermittent filaments whose spatial statistics are approximately log-normal. The mean concentration downwind of a continuous point source in a neutral atmospheric boundary layer is modelled by the Gaussian plume equation:

The intermittency factor \(\gamma\)is the fraction of time that concentration exceeds a detection threshold at a fixed downwind point. In turbulent flow \(\gamma\) can be as low as 0.1, so the animal detects brief filament hits separated by odor-free intervals. The cast-and-surge strategy exploits exactly this statistical structure: upon detecting a filament, surge upwind; upon losing it, cast crosswind to reacquire.

Optimal Search

Vergassola, Villermaux, and Shraiman (2007) formulated the “infotaxis” algorithm for optimal plume navigation: the agent chooses moves that maximize expected information gain about the source location. Infotaxis reproduces the observed cast-and-surge behavior of flying insects and provides a Bayesian framework for chemical search. Later work by Reddy et al. (2022) used deep reinforcement learning to train virtual moths, finding strategies qualitatively indistinguishable from biology.

At equilibrium, the occupancy of a receptor by a single odorant ligand follows the Langmuir isotherm, extended with Hill cooperativity \(n\):

For olfactory receptors \(K_D\) typically ranges from \(10^{-10}\)M (narrowly-tuned pheromone receptors such as BmOR-1) to \(10^{-4}\) M (broadly-tuned general odorant receptors). The six orders of magnitude in affinity space, combined with the log-distributed concentrations that animals encounter in turbulent plumes, motivate the observed log-normal distribution of \(\log K_D\) across the OR repertoire.

Noise limits push receptor detection toward the Berg–Purcell bound on concentration sensing: the minimum measurable relative concentration\(\delta c / c\) for a receptor of linear dimension \(a\) sampling for time \(T\) is

Moth sensilla with their enormous collective pore area exploit this scaling: aggregating over ∼17,000 sensilla per antenna approaches \(a_{\text{eff}} \sim\) 1 mm, driving detection thresholds to single-molecule resolution.

Chemical signals can be honest (costly to produce, correlated with quality) or deceptive. A celebrated example of deception is the orchid Ophrys sphegodes, whose flower emits a pheromone mimicking a receptive female solitary bee; males pseudocopulate with the flower and transfer pollen. Zahavi’s handicap principle predicts that honest signals evolve when the marginal cost of production increases faster for low-quality senders.

Conversely, predator–prey chemical arms races drive rapid receptor evolution. The garter snake Thamnophis sirtalis feeds on tetrodotoxin-laden rough-skinned newts (Taricha granulosa); population-level matching of toxin concentration to resistance illustrates coevolutionary dynamics visible in receptor sequences (Brodie et al. 2002).

Kairomones are chemical cues inadvertently released by one organism and exploited by another. Mosquitoes home to vertebrate hosts by integrating CO\(_2\) (Gr21a/Gr63a), lactic acid, octenol, and skin microbiome volatiles—a multimodal kairomone plume. Blood-feeding insects therefore anchor at least three modules of our course: olfaction (module 6), thermoreception (module 7), and multimodal integration (module 8).

The glomerular map in the olfactory bulb is remarkably stereotyped: the same OR class projects to the same glomerulus in every mouse of a given strain, producing a “receptor-space” representation of odor identity. Mitral and tufted cells relay this representation to piriform cortex, where Poo & Isaacson (2009) found a distributed, non-topographic encoding: any given odor activates a spatially scattered subset of neurons across piriform, with little of the glomerular topography preserved.

This shuffling of the combinatorial code through dense random-like projections is a canonical example of expansion-recoding of sensory inputs, strikingly similar to the Kenyon-cell expansion in the insect mushroom body (Caron et al. 2013). Both circuits implement a high-dimensional sparse recoding that facilitates associative learning: linear discriminability increases, while mutual information about the specific receptor activations is reduced.

Olfactory Learning and Memory

Classical Pavlovian conditioning in Drosophila pairs an odor (conditioned stimulus) with electric shock (unconditioned stimulus). Mushroom body output neurons (MBONs) receive plasticity signals from dopaminergic neurons, modifying the sparse Kenyon-cell\(\to\) MBON weights (Aso et al. 2014). In mammals, the same principles apply to piriform–orbitofrontal projections and reward-guided odor learning.

Dimensionality of Odor Space

Koulakov et al. (2011) analyzed the perceptual olfactory space using data from the Dravnieks atlas and showed that olfactory percepts span an intrinsically high-dimensional manifold (\(d \gtrsim 20\)), in contrast with the three-dimensional color space of trichromatic vision. The high dimensionality is the natural consequence of combinatorial coding: any low-dimensional embedding of the receptor-activation patterns would collapse information.