Prescott & Liberles (2022) — Internal senses of the vagus nerve
The wiki has been carrying the transduction stage of interoception secondhand. interoceptive-sensors and neuropod-cells were both built from Berntson & Khalsa’s one-paragraph inventory, and both pages say so in their notes. This is the primary review those pages were reaching through — 21 pages, ~250 references, from the Liberles lab that produced much of the underlying work (the baroreceptor PIEZO papers, the airway water-sensing NP19 neurons, the gut satiety and thirst circuits, the vagal airway-projection atlas). Where Berntson & Khalsa give a sentence, this gives the receptor, the Cre line, the knockout phenotype and the reflex.
It is not an interoception-theory paper. It engages none of this wiki’s construct debates, proposes no taxonomy, and measures nothing. Its value is entirely at the periphery — the layer every predictive-coding page on this wiki treats as classical and complete, and which this review shows is neither.
The organizing facts
The vagus is the 10th and longest cranial nerve, ~80% of its fibres sensory, providing the principal sensory line from almost every internal organ. Afferent cell bodies cluster in two ganglia with different developmental origins — the nodose (epibranchial placode) and jugular (neural crest) — which in humans are distinct structures but in mice fuse into a supraganglion. Each vagal sensory neuron is pseudounipolar, extending one axon that projects in both directions (to the organ and to the brain). Nodose afferents target the nucleus of the solitary tract (NTS); jugular afferents instead target the paratrigeminal nucleus and spinal trigeminal tract, aligning visceral and somatosensory pain pathways.
Numbers scale with body size: ~2,300 vagal sensory neurons per side in mouse, 20,000–30,000 in larger mammals, up to ~100,000 in humans — still a tiny fraction of the nervous system overseeing a vast array of vital functions.
The dark matter. Single-cell RNA-seq reveals dozens of transcriptomically distinct vagal sensory types, ~10–160 neurons each in mouse, most with unknown receptors, terminal morphology and function. The authors call these the “dark matter of the vagus nerve,” and the phrase is the review’s honest through-line: cellular diversity vastly exceeds the catalogue of known vagal reflexes, so most of what the vagus senses has not been assigned to a cell.
Three systems, three sets of sensors
Respiratory. The Hering-Breuer inspiratory reflex (lung inflation → apnea, described 1868) is carried by vagal slowly-adapting stretch receptors, and its molecular sensor is now PIEZO2 (Nonomura et al. 2017) — optogenetic PIEZO2 activation reproduces the reflex; global knockout is lethal at birth from respiratory distress. Airway protective reflexes (cough, laryngeal closure, apnea) run through separate labelled lines: a small cluster of P2RY1 (NP19) water-sensing neurons in the SLN, TRPV1/TRPA1 chemonociceptors, the itch receptor MRGPRC11, and NPY1R/NPY2R subsets with opposing effects on breathing. This is the receptor-level detail behind the respiratory-interoception page’s claim that the breath is an exceptional channel.
Cardiovascular. The baroreceptor reflex — arterial pressure sensing at the aortic arch and carotid sinus — depends on both PIEZO1 and PIEZO2 in vagal and glossopharyngeal neurons (Zeng et al. 2018; Min et al. 2019); dual knockout abolishes the reflex and produces labile hypertension, while removing either channel alone leaves it largely intact (a dual requirement, unusual for a mechanosensory reflex). The carotid body is the classical hypoxia chemosensor: glucose-like glomus cells sense low O2 through mitochondrial electron-transport-chain signalling and HIF2alpha, releasing ATP onto afferent nerves — a second-order chemosensation, the afferent neuron reading out from a sentinel cell rather than transducing directly. The Bainbridge and Bezold-Jarisch reflexes are named as still-mysterious cardiac afferent circuits.
Digestive / gut-brain axis. The richest and least understood territory. Enteroendocrine cells are wired sensory transducers synapsing onto vagal afferents; gut hormones (CCK, GLP1, PYY, ghrelin) and mechanical distension are read by distinct vagal populations. GLP1R- and OXTR-expressing vagal neurons transduce stomach/intestine distension and drive satiety (IGLEs — intraganglionic laminar endings — are the mechanosensory terminals). Area postrema neurons (a circumventricular organ exposed to blood, expressing GLP1R and GFRAL) mediate nausea and conditioned food aversion. The review makes the translational point explicit: GLP1-receptor agonists (the obesity/diabetes drugs) and bariatric surgery both act on this gut-brain sensory machinery.
What it changes for the wiki
It upgrades two pages from secondhand to primary. interoceptive-sensors and neuropod-cells were the wiki’s only pages on transduction, both built from Berntson & Khalsa. Everything they assert about PIEZO baroreceptors, gut TAS2Rs, glomus-cell hypoxia sensing and neuropod transduction has a fuller, first-lab statement here. The revisions attach the primary receptor and knockout evidence to claims the wiki had been holding on a single reviewing sentence.
It hardens the ‘the periphery is recent and unsettled’ warning. interoceptive-sensors argues that the theoretical pages (predictive coding, active inference, EPIC) are built on an afferent anatomy tacitly treated as complete, when much of it is post-2015. This review is the strongest single statement of that: dozens of vagal sensory types, most functionally uncharacterized, from the lab doing the characterizing. Any account that treats “the interoceptive signal” as one quantity a prior is weighed against is underdescribing a periphery whose cell census is not even finished. See feedforward-vs-predictive-interoception.
It supplies the afferent anatomy the polyvagal debate argues over the efferent version of. Grossman’s refutation of polyvagal-theory leans on vagal target-organ specificity — cardiac, pulmonary and gastric vagal activity do not covary — and on the anatomy of the two vagal motor nuclei (nucleus ambiguus vs DMV). This review is about the sensory side, but it independently documents the same organ-by-organ heterogeneity on the afferent limb: different Cre lines, receptors and reflexes for lung vs heart vs gut, with no unified “vagal” signal. It is background, not adjudication — the review never mentions Polyvagal Theory — but it is consistent with the anatomical case against treating “the vagus” as one functional unit. See respiratory-sinus-arrhythmia, is-polyvagal-theory-valid.
The thesis the wiki should keep. The review’s closing claim is that vagal interoception lags the external senses by decades of mechanistic understanding, and that closing the gap is a molecular-genetics project. That is a useful counterweight to a literature (this wiki included) that theorizes interoceptive inference elaborately while the sensors being inferred over are, in the authors’ own frame, mostly dark matter.