Nucleus tractus solitarius

In brief

What it is, and where

The nucleus tractus solitarius — the solitary nucleus, or in English usage the nucleus of the solitary tract — is a thin column of gray matter buried in the back of the medulla oblongata, the lowest part of the brainstem, just above where the brain becomes the spinal cord. It is paired, left and right, flanking the midline under the floor of the fourth ventricle and extending the length of the medulla ^1,3. Its name describes its relationship to a bundle of fibers rather than anything about the cells: it is the nucleus that wraps around the solitary tract ^1,3.

The solitary tract is the incoming cable. Sensory fibers from the seventh, ninth and tenth cranial nerves — the facial, the glossopharyngeal and the vagus — enter the medulla, turn downward, and run together in a single compact bundle before peeling off to make their connections in the surrounding gray matter. 'Tract' here means a bundle of nerve fibers traveling together; the fibers are the axons of sensory cells whose bodies sit in ganglia outside the brain, in the neck. Everything those nerves hear from the tongue, the throat, the larynx, the lungs, the heart, the great arteries, the stomach and the intestine arrives along this one tract ^2,3.

The nucleus does not sit alone. Immediately above and behind it, at the very back of the medulla, is the area postrema, a small region with no proper blood-brain barrier, which can sample circulating poisons and hormones directly. Immediately below it is the dorsal motor nucleus of the vagus, the cell bodies whose axons run back out to the gut and heart. Anatomists group the three together as the dorsal vagal complex: sensing, chemical detection and outgoing command in a space a few millimeters across ^4,5. The practical consequence is that no lesion, drug or stimulating electrode reaches one of the three without touching the others, which is a persistent problem in reading the evidence that follows.

It is also not only neurons. A 2026 study of mouse brain slices found that when glutamate — the brain's main excitatory signal — was applied to astrocytes in the part of the nucleus beneath the area postrema, only some of them raised their internal calcium, a smaller group lowered it, and blocking nerve impulses with tetrodotoxin halved the proportion that rose ³². Astrocytes are the star-shaped supporting cells of the brain, long assumed to be passive; in this nucleus they are a heterogeneous population responding in several directions at once. The modern framing for what the whole structure does is interoception: the continuous, mostly unconscious monitoring of the state of the body's own organs, as distinct from the senses that look outward ³³.

The map: subnuclei, and a body laid out in the brainstem

The nucleus is divided into subnuclei, and the division is not arbitrary. The method that settled the point arrived in the late 1970s, when horseradish peroxidase made it possible to mark a nerve and then find where its fibers end inside the brain. Applying it in 32 cats to the cut cervical vagus and to the nodose ganglion, Kalia and Mesulam reported in 1980 that vagal sensory fibers reach most of the subnuclei of the solitary nucleus on both sides, more densely on the injected side, and also end in the area postrema and among the neurons of the dorsal motor nucleus of the vagus — that last an anatomical basis for a reflex with only one synapse between the body and the outgoing command ⁴⁰. In 1989 a tracing study in 172 rats injected horseradish peroxidase and related tracers — substances taken up by nerve endings and carried back to the brain, where staining reveals them — into the ninth and tenth cranial nerves and into the wall of the upper digestive tract at different levels. The soft palate and pharynx labeled the interstitial and intermediate subnuclei; the esophagus labeled subnucleus centralis; the wall of the stomach labeled subnucleus gelatinosus ². Different organs, different addresses. A later study by the same group found that tracer from any part of the colon except the rectum reached only the commissural and medial subnuclei, while tracer from the rectum never reached the brainstem at all, because the rectum reports to the spinal cord instead ⁴¹.

Reviewing this literature in 2001, Hornby set out the functional version of the same map: swallowing in subnucleus centralis, gastric sensation in subnucleus gelatinosus, laryngeal and pharyngeal sensation in the intermediate and interstitial subnuclei, baroreceptor function in the medial subnucleus, respiration in the ventrolateral subnucleus ³. These assignments are drawn from rat and cat anatomy and from lesion work, and they should be read as a good working map rather than as a settled parts list.

Until recently that map was inferred from where fibers ended, not from watching the nucleus work. In 2022 a two-photon calcium imaging preparation made it possible to watch hundreds of solitary nucleus neurons in a living mouse while the gut and upper airway were stimulated. Individual neurons were tuned to particular organs; the neurons were arranged topographically according to position in the body; some mechanical and some chemical signals from the same organ converged on the same cells; and when brainstem inhibition was pharmacologically blocked, the tuning broadened and the map blurred ^21,34. That last result matters: the sharpness of the map is actively maintained by inhibition inside the nucleus, not simply inherited from how the incoming axons are sorted. A 2025 review of this work describes the nucleus as a place where organ signals are mapped, computed locally, modulated by descending input from higher brain regions, and then distributed to specific targets ³³.

The front end of the nucleus is a different structure from the back end. The rostral, or head-end, portion is the first relay for taste; the caudal, or tail-end, portion is the visceral part. They are connected: viral tracing in mice showed glutamatergic, GABAergic and catecholaminergic projections all running from the visceral end forward into the taste end, densest for the glutamatergic and sparsest for the catecholaminergic ⁶. This is one anatomical reason why feeling sick changes how food tastes. The taste end is also where a recent correction landed. Work from Zuker's laboratory had identified the brainstem neurons that carry sweet and bitter from the tongue toward the cortex in mice, the somatostatin-expressing population of the rostral solitary nucleus among them, and had reported that those cells, and not the calbindin or dynorphin cells, respond specifically to bitter and are needed to suppress licking of quinine ⁴². A 2026 preprint, not yet peer reviewed, found instead that most of them responded best to bitter but many responded to other tastes, that some non-somatostatin cells responded to bitter, and that roughly equal numbers of these cells carried excitatory and inhibitory markers; suppressing them chemogenetically increased licking of both quinine and sucrose, which fits two populations rather than one ⁷. A 2026 methods paper came at the same junction from the other end: with an improved surgical route into the mouse petrosal, nodose and jugular ganglia, it confirmed that the gustatory fibers of that complex terminate in the rostral solitary nucleus, while the same ganglia also send axons to the area postrema, the nucleus ambiguus, the lateral reticular nucleus, the spinal trigeminal tract and the cervical spinal cord ⁴³.

The modern anatomy is being redone by gene expression rather than by tracer. A 2025 atlas combining single-nucleus RNA sequencing from mouse and rat cataloged more than 180,000 cells of the dorsal vagal complex and 123 cell identities at five levels of resolution, and reported that mouse and rat differ — among other things in a leptin-receptor neuron population found in the rat area postrema ⁴. A 2026 tracing study then showed that molecular identity predicts where a neuron sends its axon: Lmx1b-expressing excitatory neurons project widely to both brainstem and forebrain, inhibitory neurons project mostly within the brainstem, and the catecholaminergic, cholecystokinin, neuropeptide-FF, prodynorphin and Hsd11b2 populations each have their own distinct pattern of targets ³⁵. Human confirmation is thin. The subnuclei can be recognized in human brainstem sections — one careful neuropathological study of a patient with spinocerebellar ataxia type 3 identified neuron loss and astrogliosis in the gelatinous, medial, parvocellular and pigmented solitary nuclei by name ⁴⁴ — but nobody has imaged a map in a living human brain.

The whole history, including the reversals

The history of this nucleus is mostly the history of a mistake being corrected twice. The first version was set out by Wang and Borison between 1950 and 1953. Stimulating and destroying parts of the medulla in dogs and cats, they described a vomiting center there: a discrete place that, when stimulated, produced vomiting ⁴⁵, fed by a separate chemoreceptor trigger zone in the area postrema that sampled the blood ⁴⁶. Their 1953 review in Pharmacological Reviews made both of those doctrine, and the scheme governed the field for forty years ⁴⁷. The PubMed records for all three carry no abstract, and they are cited here for the framing they are universally credited with rather than for any number. The second version, which held through the 1980s and was argued most clearly by Carpenter in 1990, moved the machinery into the solitary nucleus itself: 'it is likely that the sequence of muscle excitation and inhibition is controlled by a central pattern generator located in the nucleus tractus solitarius,' he wrote, with humoral signals arriving by way of the area postrema and visceral signals by way of the vagus, converging there ²⁶. A central pattern generator is a set of neurons that, once triggered, produces a fixed sequence of muscle actions on its own, without step-by-step instruction from above; walking, breathing, swallowing and vomiting each have one. The companion review in the same issue described, in detail, the sequence of events in the gut that the generator must drive ⁴⁸.

Two years later Fukuda and Koga took that hypothesis apart. Recording in decerebrate dogs — animals in which the brain above the brainstem has been disconnected, so that reflexes can be studied without anesthesia and without any influence from the forebrain — they found 78 non-respiratory neurons in the Bötzinger complex — a group in the upper medulla well in front of the solitary nucleus — that responded to vagal stimulation at a mean latency of 387 milliseconds, and half of which burst in time with each retch. Cooling the medial solitary nucleus to block transmission abolished the retching and the bursts in all eleven such neurons tested, while stimulating the opposite vagus still produced both ⁸. The reading is precise: the solitary nucleus is the necessary gate, but the rhythm is generated elsewhere. By 2001 Hornby could write that 'there is not an isolated vomiting center, but rather groups of loosely organized neurons throughout the medulla that may be activated in sequence by a CPG' ³ — a distributed generator, which is still the current view. A 2023 review puts it as the generator 'may be in the Bötzinger nucleus,' and the hedge is the author's own ⁵.

The second correction concerned the chemistry of the first synapse. Because drugs blocking the NK1 receptor — the receptor for the peptide substance P — stop vomiting from many causes, and because the emetic vagal fibers synapse in the medial solitary nucleus, it was natural to suppose substance P was the transmitter there. Fukuda's group tested that directly. The NK1 blocker GR205171, at 0.05 to 0.7 milligrams per kilogram intravenously, abolished retching in dogs within five minutes, while leaving the firing of medial solitary nucleus neurons and the evoked potentials there entirely unchanged ¹¹. Substance P was not the transmitter carrying the signal in. A companion study found that the non-NMDA glutamate blocker NBQX abolished both retching and salivation, while the NMDA blocker MK-801 changed nothing ⁹ — so the first synapse is a fast glutamate synapse. Where the NK1 antagonist does act was then localized: GR205171 at 25 to 50 micrograms per kilogram abolished the slow build-up and the retching bursts of the pattern generator neurons themselves while sparing the fast initial response, placing the drug's action at the generator or on the path to it ⁴⁹.

The newest reversal is about what those drugs actually achieve. A 2023 assessment by Andrews, Golding and Sanger concluded that NK1 antagonists block vomiting across many triggers but do much less, and possibly nothing, for the sensation of nausea — even though imaging shows very high receptor occupancy at the doses used in patients ²⁵. Calling a drug an antiemetic, they argue, gives a false impression of what it does to the feeling. That gap, between stopping the reflex and relieving the symptom, is the live problem in the field.

Three doors into one nucleus: how the vomiting reflex is triggered

Vomiting can be set off from at least three directions, and all three converge on this nucleus. The first and best understood is the gut. Cells scattered through the lining of the intestine, called enterochromaffin cells, make and release serotonin; by one accounting they make about 80 percent of all the serotonin in the body ²⁸. Cytotoxic chemotherapy damages the gut lining and causes these cells to dump their serotonin, which raises serotonin concentrations in the intestinal lining and in the area postrema, and the released serotonin acts on 5-HT3 receptors sitting on the nearby vagal sensory endings ^28,53. The 5-HT3 receptor is an ion channel: serotonin binding opens a pore, the fiber depolarizes, and the signal travels up the vagus into the medial solitary nucleus. Cutting the abdominal vagus suppresses cisplatin-induced vomiting in animals, which is the evidence that this is the main route ²⁸. Blocking the receptor is the basis of the setron drugs: in a review of trials, oral granisetron at 2 milligrams once daily or 1 milligram twice daily had been studied in more than 4,500 patients on emetogenic chemotherapy, with total control of vomiting in 44 to 60 percent and complete response in 70 to 94 percent ²⁴. That route carries the vomiting reflex, but it may not carry the malaise. In 2026, mice bred without the 5-HT3 receptor ate as little and lost as much weight after cisplatin as normal mice, moved as little, showed as much pica — the eating of non-food material, which in a rodent is read as a sign of feeling ill — and learned the same aversion to a flavor paired with the drug. In both groups the number of cells staining for c-Fos, a protein made when a neuron has recently been active, rose in the area postrema and in the solitary nucleus; among them were cells carrying the brainstem receptor for a stress hormone called growth differentiation factor 15, and the blood level of that hormone stayed raised for at least a day after the drug ⁵². In mice, then, the sickness can be produced without the serotonin receptor the setrons block. Mice cannot vomit, so this speaks to the malaise and not to the reflex.

Where those receptors sit in a human brain has been measured, and by two methods that are worth keeping apart. In postmortem human brainstem, autoradiography showed the greatest accumulation of 5-HT3 binding in the area postrema and the area subpostrema. Dissecting those regions out and measuring binding in tissue homogenates then put the area postrema and subpostrema at 112.1 femtomoles per milligram of protein, a pooled dorsal vagal complex sample — the solitary nucleus excluding the subpostrema, together with the dorsal motor nucleus of the vagus and the hypoglossal nucleus — at 22.9, and the olivary nucleus at 3.9, with none detectable in the reticular formation beneath ¹⁴. The 22.9 figure is therefore not a solitary nucleus figure. A second study using a different ligand put the solitary nucleus, area postrema and spinal trigeminal nucleus at 50 to 200 femtomoles per milligram tissue equivalent, against 4 to 17 in forebrain regions — and its authors added an explicit warning that human and animal 5-HT3 receptor pharmacology differ enough that predicting a clinical response from animal data is unsafe ¹⁵. These are binding measurements. They say where the receptor is, not what happens when it is occupied in a living person.

The second door is the bloodstream, by way of the area postrema. Because that small region lacks a normal blood-brain barrier, circulating substances reach its neurons directly, and its output goes next door into the solitary nucleus ⁵⁴. Single-nucleus sequencing of the mouse area postrema found only a few neuron types; genetically switching on particular excitatory types produced nausea-related behavior, and one type carried the aversion caused by several different poisons ¹⁸. One of the receptors there has a well-characterized ligand. GFRAL — glial cell line-derived neurotrophic factor family receptor alpha-like — is the receptor for growth differentiation factor 15, or GDF15, the stress hormone named in the first paragraph of this section: a signal released by cells under stress or injury. It is found in the mouse brainstem only in the area postrema and the solitary nucleus, and mice lacking the gene for it resist the anorexia and weight loss caused by chemotherapy ⁵⁰. That paper came almost entirely from the staff of NGM Biopharmaceuticals, with co-authors from Merck Research Labs, which is worth naming because GDF15 blockade is a commercial drug target. Nine years on, that target has an active development pipeline. A 2026 review describes it as GDF15 acting on GFRAL together with a co-receptor called RET, named for the gene it was first found in, and reports it as an active mediator rather than merely a marker in cancer anorexia and cachexia, in chemotherapy-induced nausea and vomiting, in obesity and in hyperemesis gravidarum, with agonists, antagonists and methods of binding and clearing circulating GDF15 all in development or in trials ⁵⁵. Three of its four authors are the chemists and physiologists who built the brain-penetrance-reduced GLP-1 conjugates described later on this page ^56,57, which is the reason to name the review rather than lean on it.

The third door is the inner ear. Tracing in rabbits found axons running from the caudal medial and inferior vestibular nuclei to the solitary nucleus on both sides, with fewer reaching the dorsal motor nucleus ²⁹; later work in rats established that those projecting cells are glutamatergic, carry vesicular glutamate transporter 2 along with the modulator imidazoleacetic acid ribotide, and are themselves wrapped in inhibitory terminals ³⁰. This is the anatomical substrate for motion sickness and for the nausea of vestibular disease. It is an anatomy, not a mechanism: no one has shown in a person that activity in this specific pathway is what produces the feeling. The attempt has been made. In 2026, 17 people were given 21 minutes of low-frequency sinusoidal galvanic vestibular stimulation — a weak alternating current delivered across the skull behind the ears, plus or minus 2 milliamps at 0.08 hertz, which produces a sensation of being swayed from side to side — while lying in an MRI scanner. Only four of the 17 reported incipient or overt nausea, and the signal changes that tracked the long stimulation were in the nucleus accumbens and the locus coeruleus rather than in the solitary nucleus. The authors suggest that lying supine with the head fixed is itself a poor way to make a person feel sick, which is the difficulty of this whole question in miniature ⁵⁸.

The clinical reviews describe the same convergence — vagal, vestibular and chemoreceptor trigger zone input arriving at the solitary nucleus, which relays onward to the pattern generator ^3,59,60. What those same reviews also say is that acute and chronic nausea appear to be different problems, that the chronic kind resembles neuropathic pain more than it resembles poisoning, that conventional antiemetics work poorly for it, and that the neuromodulator drugs used instead have not been properly studied in chronic functional nausea ^59,60. One of them describes the understanding of central nausea pathways, in 2016, as 'still very basic.'

The pattern generator, and what replaced the vomiting center

Vomiting is not a single muscle contraction. It is a long, fixed, stereotyped sequence: the proximal stomach relaxes, a retrograde giant contraction sweeps the contents of the small intestine back toward the mouth, the striated muscle of the esophagus contracts lengthwise and pulls the relaxed stomach up into the chest, the diaphragm and abdominal muscles make several rhythmic efforts against a closed throat, and only then does the hiatal part of the diaphragm relax and the contents come out ^5,48. Something has to run that program in order, and something has to run it the same way every time. That something is called a central pattern generator.

For a decade the generator was thought to live in the solitary nucleus ²⁶. The experiment that moved it was done in decerebrate dogs. Recording from the Bötzinger complex — a group of neurons in the upper medulla in front of and below the solitary nucleus, named for its role in breathing — Fukuda and Koga found 78 non-respiratory neurons that responded to pulse-train stimulation of the vagus at a mean latency of 387 milliseconds. Half of them burst in time with each retch, a third burst in the gaps between retches, and two-thirds of the first group ramped up their firing in the period before retching began, which corresponds in time to the prodrome a person feels. When the medial solitary nucleus was cooled to block transmission, stimulating the vagus on that side produced neither retching nor any of those firing patterns in all eleven such neurons tested; stimulating the opposite vagus still produced both ⁸. The nucleus is the door the signal must pass through. It is not where the rhythm is made.

Almost all the respiratory neurons in the same region changed their firing during retching, which is the second important finding: the generator is not a separate machine but a reconfiguration of neurons that otherwise run breathing ⁸. That is what the distributed-generator view means in practice, and it is why Hornby's 2001 formulation — loosely organized groups of neurons throughout the medulla, activated in sequence — replaced the idea of a center ³. A 2023 review of the motor sequence still says only that the generator 'may be' in the Bötzinger nucleus, which is an accurate statement of how much is settled ⁵.

The pharmacology that followed localized the antiemetics. At the first synapse, inside the solitary nucleus, the transmitter is glutamate acting on non-NMDA receptors: NBQX abolished both retching and salivation in dogs, MK-801 did nothing ⁹. The incoming emetic fibers are capsaicin-sensitive C fibers — thin, slow, unmyelinated — as shown by applying capsaicin at 33 millimolar or resiniferatoxin at 160 micromolar to the floor of the fourth ventricle, which first excited the neurons and sometimes produced retching and then abolished both responses, while direct electrical stimulation of the nucleus still produced retching ¹⁰. Substance P, despite the effectiveness of drugs that block its receptor, is not the transmitter carrying the emetic signal into the nucleus: blocking NK1 abolished retching within five minutes while leaving solitary nucleus firing and evoked potentials unchanged ¹¹. The NK1 antagonist works one step further on, at the generator itself or on the path from nucleus to generator, where it abolished the slow build-up and the retching bursts but spared the fast initial response ⁴⁹.

That whole body of work was done on the act of vomiting. The feeling of nausea is a separate question, and the separation is now the central difficulty. The 2023 assessment of NK1 antagonists found that they are broadly effective against vomiting from many different triggers but much less effective, and for some stimuli apparently not effective at all, against the self-reported sensation of nausea, even though imaging shows very high receptor occupancy in the relevant brain regions at the doses patients receive ²⁵. The authors point out that the pattern generator neurons and the solitary nucleus neurons are multifunctional — they do other things besides vomiting, and the drugs do not disturb those other reflexes — which suggests the antiemetic action depends on the state of the pathway rather than on simply switching a cell off. Nausea, they argue, requires projection from the solitary nucleus up to the forebrain, and we do not understand that part. A review of postoperative nausea and vomiting reaches the same conclusion from the clinical side: there is still a lack of fundamental knowledge about both the hindbrain generator and the forebrain pathways, and that gap is what has limited new drugs ⁶¹.

The GLP-1 neurons inside the nucleus, and what the mouse work does and does not show

Glucagon-like peptide-1, or GLP-1, is best known as a gut hormone and as the target of the semaglutide class of drugs. It is also made inside the brain, by a small group of neurons in the caudal solitary nucleus that express the preproglucagon gene — preproglucagon being the large precursor protein that is cut up to make GLP-1 and its relatives. These are usually called PPG neurons. For years the natural assumption was that they form the relay in a single gut-to-brain satiety circuit: gut releases GLP-1, vagus carries the message, these neurons pass it on. That assumption has not survived.

A 2021 study tested it directly in mice. The PPG neurons of the solitary nucleus did encode satiation, consistent with signaling stomach and intestinal distension. But the vagal input they received came predominantly from oxytocin-receptor-expressing vagal neurons rather than from the GLP-1-receptor-expressing ones, so they are not the second step in the peripheral GLP-1 pathway. More decisively, they were not necessary for a GLP-1 receptor agonist to suppress eating, and activating them at the same time as giving semaglutide suppressed eating more than semaglutide alone ¹². The central and peripheral GLP-1 systems, the authors concluded, are independent circuits that happen to use the same peptide. The same group had already shown a parallel dissociation for heart rate: exendin-4 raised heart rate in mice through beta-adrenoceptors and did so just as well after the PPG neurons were destroyed, without ever activating them — and yet switching those neurons on artificially also raised heart rate ⁶². A cell can be capable of an effect without being the route a drug takes to produce it, and conflating the two is the commonest error in reading this literature.

The most recent work extends the picture. A 2026 study found that the GLP-1 neurons of the solitary nucleus and of the neighboring intermediate reticular nucleus together set meal size; that their total number runs inversely to weight gain; that the two groups receive different direct inputs but send their axons to overlapping targets; that ablating both delays the onset of satiation enough to cause weight gain in mice eating ordinary chow; and that daily chemogenetic activation of the solitary nucleus group drove sustained weight loss in obese mice 'without notable adverse effects' ¹³. A separate 2026 study of the receptor for glucose-dependent insulinotropic polypeptide, or GIP — a second gut hormone, and the other half of the dual-agonist obesity drugs — found that the extra weight loss produced by adding a GIP receptor blocker to liraglutide depended on hypothalamic receptors and was, in its authors' words, not mediated by solitary nucleus preproglucagon neurons ⁶³. These cells are also recruited by psychological stress: activating them moderately increased anxiety-like behavior in mice in a sex-dependent way, driven by input from the paraventricular hypothalamus, though the paper notes their role in anxiety-like behavior is controversial ⁶⁴.

Separately, the nucleus has a direct line to the brain's reward machinery. Neurons projecting from the solitary nucleus to the ventral tegmental area in mice were largely catecholaminergic, often GLP-1-containing, mostly connected to the vagus only indirectly through intervening neurons, and mostly inhibited rather than excited by the satiety hormone cholecystokinin ⁶⁵. Most of this nucleus is not a simple relay; the majority of the cells studied were higher-order.

Now the limits, which matter more here than the findings. Every result in the preceding three paragraphs is from mice. Mice and rats cannot vomit, so none of this work can measure the side effect that actually limits GLP-1 drugs in people. Aversion and malaise behavior in a rodent are proxies, and 'without notable adverse effects' in a species with no emetic reflex is weak reassurance about nausea in a patient ¹³. The field knows this: a 2024 methods paper described a ligand-conjugated saporin — a toxin tethered to a molecule that binds the GLP-1 receptor — capable of killing GLP-1-receptor cells in human tissue, in mice, and in the musk shrew, which can vomit, precisely in order to make the comparison possible ²⁰. That tool exists; the comparison has not yet been published. Until it is, the human relevance of the brainstem GLP-1 circuitry rests on anatomy and inference.

What none of that work can say is what these cells do to nausea. That half of the question has been approached, but from the drug side rather than the cell side, and in an animal that vomits. In the musk shrew, systemic exendin-4, a GLP-1 receptor agonist, reliably produced emesis, while an equimolar dose of the same peptide tethered to vitamin B12, which enters the brain far less, produced almost none; injected directly into the brain the two were equally emetic, which places the vomiting at central receptors rather than peripheral ones ⁵⁷. A companion paper built the same kind of conjugate on dicyanocobinamide, a precursor of vitamin B12, and reported the same profound reduction in emetic episodes in shrews with glucose control preserved ⁵⁶. In 2026 a biased agonist — a drug that engages only part of what a receptor can do, in this case favoring the cyclic AMP arm of the response over the recruitment of beta-arrestin, the protein that draws the receptor inward and starts a second kind of response — produced fewer emetic episodes than exendin-4 in 30 musk shrews and little or no pica in 54 rats, while improving glucose tolerance in mice, rats and shrews alike; c-Fos rose in the nausea- and emesis-related brain regions of all three species, and several of the authors worked at Lilly Research Laboratories, which develops drugs of this class ⁶⁶. So central GLP-1 receptor activation does cause vomiting in a species capable of it. What has not been done is the experiment this section is actually about: no published study activates or removes the preproglucagon neurons of the solitary nucleus in an emetic species. The rodent work can report only avoidance, reduced intake and pica ^13,18; 'without notable adverse effects' in a mouse cannot detect emesis at all ¹³; and the ligand-conjugated saporin built to strip GLP-1-receptor cells from the musk shrew as well as the mouse was published as a tool in 2024, with the comparison it was made for still unpublished ²⁰. The honest answer is that the drugs that act on these receptors make an emetic animal vomit, and that whether these particular neurons are the ones responsible is unknown.

'The vagus' is not one nerve, and the correction matters

Almost everything the solitary nucleus hears from below the neck arrives on the vagus, and for most of the twentieth century the vagus was treated, in practice, as a single channel: stimulate it and you stimulate the body-to-brain connection. The genetic work of the last decade has made that description untenable. The vagus is a cable of many separate lines, each with its own molecular identity, its own organ, its own ending, its own conduction speed and its own target inside this nucleus.

The first clean demonstration came in 2016. Two genetically marked populations of vagal sensory neuron were traced and manipulated separately: GPR65-marked neurons innervated the gut lining and responded to nutrients, while GLP1R-marked neurons formed endings in the muscle wall and responded to distension of the stomach and intestine. The two had different peripheral anatomy, different central terminals and different conduction velocities ³⁶. In 2019 two papers widened this enormously. One used target-specific single-cell RNA sequencing to map the vagal cell types innervating the gastrointestinal tract and then tested them: food intake turned out to be most sensitive to stimulating intestinal stretch receptors, while nutrient-activated mucosal afferents had no effect on intake at all, and only the stretch receptors durably silenced the hunger-promoting AgRP neurons — named for agouti-related peptide, the signal they release — of the hypothalamus ⁵¹. That is a reversal of the intuitive idea that the gut stops a meal by detecting food. The other sequenced both vagal ganglia and found that the jugular ganglion, which develops from the neural crest, contains neuron types resembling skin sensory neurons — low-threshold mechanoreceptors, nociceptors, cold- and heat-sensitive C fibers — while the nodose ganglion, which develops from a placode, contains 18 distinct types dedicated to the internal organs ³⁷. Two different ganglia, two different developmental origins, two different kinds of sensing, bundled in one nerve.

The same approach then found populations too small to have been found any other way. About 100 neurons per mouse, marked by the gene P2RY1 and innervating the throat, turned out to run an entire coordinated airway defense: switching them on with light produced apnea, closure of the vocal folds, swallowing and expiratory reflexes together, and removing them abolished the protective response to water or acid placed in the larynx ⁶⁷. A hundred cells, one program. The 2022 review that collects this literature describes a 'tremendous diversity' of vagal sensory neuron types across the respiratory, cardiovascular and digestive systems, with many types still poorly understood and the signal transduction mechanisms of most of them unknown ⁶⁸. The reviews written since have not narrowed the picture. A 2025 synthesis argues against reading vagal afferent neurons as a passive relay for satiety signals at all: they encode stretch, nutrient content, metabolic state and even the metabolites of gut bacteria, and their subpopulations feed distinct brain circuits, including circuits for food reward and memory ⁶⁹. A 2025 review of the lung-innervating neurons alone describes striking molecular diversity, distinct projection patterns and specialized endings, and says plainly that how particular subtypes encode their inputs, and how those signals are integrated centrally, are unresolved ⁷⁰. New lines are also still being found. In 2026, mouse lung adenocarcinoma was shown to recruit vagal sensory neurons marked by the gene Npy2r, which signal to brainstem nuclei and drive sympathetic outflow that suppressed anti-tumor immunity through beta-adrenergic signaling in alveolar macrophages; blocking that sensory-to-sympathetic route genetically, pharmacologically or chemogenetically slowed tumor growth ⁷¹. That is a vagal line nobody was looking for, found in mice, not in people.

This changes how to read older work and how to read clinical practice. When a 1992 experiment stimulates 'the vagus' with a pulse train and measures retching, it is stimulating many lines at once, and the latency it measures — 387 milliseconds — is the latency of whichever line dominates ⁸. When a clinician stimulates the vagus in the neck to treat epilepsy, the same applies. And the sorting continues inside the nucleus: the 2022 imaging work showed that the spatial separation of organ representations is sharpened within the solitary nucleus beyond what the incoming axons achieve by themselves, and collapses when local inhibition is blocked ²¹. The 2026 output tracing shows the same principle on the far side: each molecularly defined population of solitary nucleus neurons projects to its own set of targets ³⁵, and the cross-species atlas shows that the catalog of cell types is not even identical between mouse and rat ⁴.

The honest summary is that the parts list is now good and the wiring diagram is improving, in rodents, while the function of most individual lines is unknown — the 2025 reviews say so in as many words ^69,70 — and almost none of it has been confirmed in a human. There is no human equivalent of these experiments, because they require genetic access to particular cell types in a living brain.

The other jobs: blood pressure, breathing, swallowing

Vomiting is the reason this nucleus appears in a podcast about nausea, but it is a minority of what the structure does. The stretch receptors in the walls of the carotid arteries and the aortic arch report blood pressure continuously, and their fibers end in the medial subnucleus. That connection is, in the phrase of the modelers who studied it most closely, 'the first synapse in the baroreceptor reflex' ²². It is not a passive relay: a detailed model fitted to recordings from rat medial solitary nucleus neurons predicted that above roughly 20 to 30 hertz the properties of the synapse itself — vesicle supply and receptor kinetics — limit how faithfully the neuron can follow its input ²². The same junction has a bedside consequence. In male mice given 50 to 200 micrograms per kilogram of dexmedetomidine, a sedative that commonly slows the heart, heart rate fell dose-dependently from 408.4 to 255.5 beats per minute; the drug directly excited glutamatergic neurons of the vagal nodose ganglion, raising their firing rate from 0.8 to 3.8 hertz, and increased glutamate release in the solitary nucleus, and chemogenetically silencing that nodose-to-nucleus line reversed the slowing ⁷². Mice, not patients — but the pathway is the one an anesthesiologist is managing. The same junction sits at the boundary between circulation and consciousness: the nucleus, together with the C1 adrenergic neurons and the parabrachial nucleus, mediates bidirectional interactions between the baroreflex and arousal, which is one reason sleep apnea, insomnia and hypertension travel together ⁷³. A 2025 review summarizes the nucleus's role in the central control of blood pressure through the baroreflex, sympathetic outflow, the renin-angiotensin-aldosterone system and oxidative stress, and is candid that the questions are not resolved ⁷⁴.

Breathing reports here too. Sensory fibers from the lungs and airways end in the ventrolateral part of the nucleus ³, and the whole catalog of respiratory vagal afferent types — stretch receptors, irritant receptors, chemical sensors — is reviewed in the modern vagal literature ⁶⁸. What arrives is also modulated by the rest of the brain and by the clock. In rats, orexin — a wake-promoting hypothalamic peptide whose neurons fire mainly during the animal's active hours — facilitates this reflex by way of hypothalamic corticotropin-releasing hormone neurons that project to the solitary nucleus. Silencing those neurons cut the breathing response to low oxygen by about half — but only during the animals' active phase, with no effect during the inactive phase, and activating the same pathway did nothing at all ³⁹. The same circuit, tested at a different hour, gives a different answer. A 2026 study from the same laboratory showed how a descending signal of this kind acts at the level of cell types. In slices from mice of both sexes, corticotropin-releasing hormone reduced the currents that incoming sensory fibers evoked in the inhibitory GABA neurons of the solitary nucleus while increasing them in non-GABA neurons; the receptor for it was more abundant in the GABA neurons of the medial subnucleus; and blocking that receptor on its own increased GABA-neuron currents, which means the peptide holds the brakes down even at rest. That is why its net effect on the nucleus is excitatory ⁷⁵. How much the traffic coming the other way shapes the breathing rhythm itself was measured in the same year: recording along the length of the ventral respiratory column in live mice while stimulating the GABAergic afferents of the solitary nucleus advanced or delayed the next breath, in a way quantitatively different from inhibiting neurons inside the column, and the authors read this as inhibition from lung stretch pathways transiently stabilizing an expiratory state — an 'inspiratory-off attractor' rather than a simple switch ⁷⁶. Mice, and a modeling argument fitted to recordings. What is at stake clinically is set out in a 2026 review of sudden unexpected death in epilepsy, which argues that seizure activity spreading into the brainstem respiratory centers, this nucleus and the pre-Bötzinger complex among them, is a route to fatal central apnea, and that astrocytes are part of that mechanism ⁷⁷.

Swallowing is the clearest case of the nucleus doing more than relaying. Within it is the dorsal swallowing group, a set of neurons that organize the sequence of a swallow. A 2025 study imaged this directly in eight perfused rat brainstems: stimulating the superior laryngeal nerve produced first a tightly confined sensory response on the stimulated side at 114.3 plus or minus 94.4 milliseconds, and then, when the stimulus was strong enough to trigger a swallow, a bilateral and anatomically distributed response at 200 plus or minus 145.2 milliseconds that preceded the motor bursts in the vagus ⁷⁸. The authors conclude that the solitary nucleus holds swallowing premotor neurons, while the first-order premotor pools that actually drive the throat muscles lie outside the dorsal swallowing group. Rhythm-making and relaying are mixed in the same small volume, which is exactly why the vomiting question took so long to settle.

This has already reached treatment. In a rat model of stroke-related swallowing difficulty, ten-hertz repetitive transcranial magnetic stimulation over the cortex improved swallowing, and the improvement disappeared when solitary nucleus excitability was suppressed or when one molecular step in the pathway was blocked ⁷⁹. The stimulation is applied to the cortex; the effect requires this nucleus in the brainstem. It is a rat model, not a trial, but the technique is already in clinical use for the same problem.

What happens when it is damaged

The commonest way the solitary nucleus is destroyed in a human being is a stroke in the side of the medulla, producing Wallenberg's syndrome — also called lateral medullary syndrome. The artery involved is usually the vertebral artery or the posterior inferior cerebellar artery that branches from it. The largest clean series followed 130 consecutive patients with pure lateral medullary infarction, correlating MRI with findings. Lesions nearer the head end of the medulla produced difficulty swallowing and facial weakness significantly more often than lesions nearer the tail end, and slurred speech more often, while the tail-end lesions caused severe gait unsteadiness and headache more often. Angiography in 123 of them showed vertebral artery disease in 67 percent and posterior inferior cerebellar artery disease in 10 percent, and the presumed causes were large vessel disease in 50 percent, arterial dissection in 15 percent, small vessel disease in 13 percent and cardiac embolism in 5 percent ¹⁶. That paper does not report how many of the patients vomited or hiccupped, so no such figure is quoted here.

Hiccups, vomiting and nausea do appear as the defining presentation of a different lesion in the same neighborhood. Area postrema syndrome is defined as hiccups, nausea or vomiting with no explanation in the gut, lasting at least 48 hours, and it is one of the core diagnostic attacks of neuromyelitis optica spectrum disorder, an autoimmune disease in which antibodies against the water channel aquaporin-4 injure astrocytes — with the dorsal medulla, and the area postrema in particular, a favored site ³¹. It can be the first manifestation of the disease, preceding optic neuritis or spinal cord inflammation; a similar picture occurs less often in myelin oligodendrocyte glycoprotein antibody disease and in glial fibrillary acidic protein (GFAP) astrocytopathy, an autoimmune attack on the skeleton protein that gives astrocytes their shape ⁸⁰. The clinically important point in both reviews is that these patients are routinely routed to gastroenterology first, delaying immunotherapy by weeks to months ⁸⁰. One hypothesis currently in circulation — that glucagon-like peptide-1 and its receptor track with the presence and the severity of the syndrome specifically in aquaporin-4-IgG-positive neuromyelitis optica, with receptor blockade normalizing neuronal overexcitability in animal models — comes from a single center, was presented at the 2025 American Neurological Association meeting and posted as a preprint, and has not been independently replicated or fully peer reviewed; the review that reports it calls it hypothesis-generating, and so does this page ⁸⁰.

Less dramatic damage to the same territory produces a recognizable cluster. In children with dorsal brainstem injury from oxygen deprivation at birth, lesions in the region of the solitary nucleus produced difficulty swallowing, reflux of stomach contents, noisy obstructed breathing from the larynx, facial weakness and abnormal breathing patterns; children with Leigh syndrome and similar lesions added hiccups and vomiting to that list ²⁷. Human neuropathology confirms the anatomy by direct inspection: in one patient with spinocerebellar ataxia type 3 and documented swallowing difficulty, serial sections showed neuron loss and astrogliosis in nearly all the swallowing-related lower brainstem nuclei, naming the gelatinous, medial, parvocellular and pigmented solitary nuclei among them ⁴⁴. That is a single case, which is how much of the human subnuclear detail rests.

Traffic also runs downward, and when it does it can injure the organs below. In rats subjected to restraint and water immersion, a standard stress model, silencing the glutamatergic pathway running from the insular cortex to the solitary nucleus reduced stomach lining damage and shifted the markers of barrier integrity in the stomach wall ⁸¹. The nucleus is not only a place the body reports to; it is also a place the forebrain acts through, which is one mechanistic account of how distress reaches the stomach. A 2026 study in mice narrows the claim. Restraint water-immersion stress raised GAD65, one of the two enzymes that make the inhibitory transmitter GABA, along with somatostatin, and lowered parvalbumin, a protein that marks a different class of inhibitory cell, in the insular cortex; silencing the insular inhibitory neurons, or the somatostatin ones alone, made the stomach injury worse, and activating the somatostatin cells made it better. But silencing only those somatostatin neurons that project to the solitary nucleus changed nothing, while silencing somatostatin and glutamatergic insular neurons together reduced the injury and raised the tight-junction proteins occludin and the claudins ⁸². The insula does act on the stomach through this nucleus; it does not do it by way of that one projection. Traffic from above can also protect. In a rat model of chronic migraine, oxytocin rose in the paraventricular hypothalamus and in the solitary nucleus during late pregnancy, viral tracing confirmed a direct oxytocin projection from the first to the second, and giving oxytocin into the nose or overexpressing it at either end lowered calcitonin gene-related peptide, neuronal activation and inflammatory cytokines in the nucleus and reduced pain behavior ⁸³. These are rodent models of stress and of migraine, not human stress or human migraine, and the step from one to the other has not been made.

Finally, the nucleus responds to its own injuries in ways that are not simply local. When the superior laryngeal nerve was cut on one side in mice, the motor nucleus showed a c-Fos response — c-Fos being a protein made when a neuron has recently been active — that shifted over three days from the cut side to the intact side, while the solitary nucleus responded on both sides from the start ⁸⁴. Losing input on one side is not a one-sided event in this structure, which may bear on why recovery after a medullary stroke is so variable.

Vagus nerve stimulation works through it, and that is most of what is known

Electrical stimulation of the vagus nerve in the neck is an approved treatment for drug-resistant epilepsy and for depression, and is used in heart failure ²³. The nerve carries the stimulation into the brain, and the first place it lands is the solitary nucleus. Beyond that, the mechanism is not established, and the research on it is unusually candid about saying so.

The electrophysiology has been done carefully in rats. At the mild intensities used clinically — the threshold that just slows the heart — stimulation raised firing in about a third of solitary nucleus neurons, but it directly activated only myelinated vagal fibers: 3 of 4 neurons receiving a single-synapse A-fiber input responded, 6 of 42 receiving a single-synapse C-fiber input, and 16 of 21 neurons connected only through intervening neurons ²³. Most of the extra activity was not time-locked to the stimulus pulses at all, and it continued into the periods between stimulation. Two of that paper's authors were employees of Cyberonics, which manufactures vagus nerve stimulators; the finding is not thereby wrong, but it should be read knowing it.

A follow-up study from the same laboratory asked how much of this depends on the rest of the brain. Cutting the brainstem away from everything above it reduced the resting firing of second-order solitary nucleus neurons receiving myelinated vagal input by 85 percent, and abolished the stimulation-driven activation of the higher-order neurons and of the second-order neurons receiving unmyelinated input ³⁸. Acute stimulation still activated the directly connected cells. The interpretation is that descending projections from above the medulla are needed to amplify what the periphery sends — so vagus nerve stimulation is not a one-way signal into a relay, and the brain's own state determines how much of it registers.

In humans the evidence is imaging, and it is thinner. A study in 37 healthy volunteers compared four ear locations for non-invasive stimulation of the vagus nerve's auricular branch, using 500-microsecond pulses at 25 hertz, and found that only stimulation at the cymba conchae — the hollow of the upper ear — activated both the solitary nucleus and the locus coeruleus significantly more than an earlobe sham ⁸⁵. That result is a group average in a structure a few millimeters wide, at the limits of what brainstem functional MRI can resolve. A 2026 randomized trial brings the caution into focus: 70 patients with chronic low back pain received four weeks of either auricular vagus stimulation or stimulation of the great auricular nerve, which is not vagal; 51 completed; pain improved in both groups with no significant difference between them, while the connectivity changes in the solitary nucleus and other brainstem nuclei differed between the arms ⁸⁶. A measurable change in this nucleus is not, by itself, evidence that the vagal route is what helped. None of this work has shown, in a person, which cells in the nucleus are being activated or what they do next; what is known clinically is that the treatment helps some patients with epilepsy and depression, and that the nucleus is unavoidably in the path ¹. Work on neighboring methods makes the same point about how much of this is inference. In 2026, focused ultrasound aimed at the spleen in mice with chemically induced colitis reduced inflammation in the gut and the brain, restored the intestinal barrier and improved anxiety-like behavior and cognition — and all of it disappeared when the vagus was cut below the diaphragm, with activation reported along a solitary nucleus to locus coeruleus circuit ⁸⁷. Something is traveling the vagal route; which cells in this nucleus carry it, and what they do next, is not shown there either.

The species problem, which runs under everything above

Rats and mice cannot vomit. They have the nucleus, the area postrema, the vagal inputs and most of the receptors, but no emetic reflex. This is not a detail; it is the structural weakness of the entire modern literature on this nucleus, because the modern tools — cell-type-specific sequencing, optogenetics, chemogenetics, calcium imaging — exist almost exclusively in mice. Every mouse experiment about nausea measures something else instead: conditioned flavor avoidance, reduced eating, a behavioral proxy for malaise ^13,18. The inability was tested directly rather than assumed. Prototypical emetic drugs — apomorphine, veratrine and copper sulfate — produced neither retching nor vomiting in rats, mice, voles, beavers, guinea pigs, nutria or mountain beavers, which between them cover all three major groups of Rodentia. The same animals also had a less muscular diaphragm, and a stomach shape worse suited to moving contents toward the esophagus, than the cat, the ferret and the musk shrew. And in an isolated brainstem preparation, mice and rats showed none of the coordinated mouth, esophagus, shoulder and phrenic nerve activity that musk shrews showed to the same stimuli. The authors concluded that the inability is general to the rodents, and that the likeliest reason is a missing brainstem component rather than the anatomy alone ¹⁷: the missing piece is thought to be neural, and in the medulla, rather than mechanical.

The species that can vomit are used for the reflex itself, and they do not agree with one another. In one study, 34 ferrets and 36 musk shrews were tested with isoflurane, the ferrets also receiving morphine and a positive control injection of cisplatin. Cisplatin produced 129.8 plus or minus 22.0 retches and 13.7 plus or minus 2.3 vomits in the ferrets, and morphine produced vomiting with a maximum at 0.9 milligrams per kilogram; but isoflurane at 2 to 4 percent, an anesthetic that causes vomiting in people, never made a ferret vomit at all, even combined with a low dose of morphine at 0.1 milligrams per kilogram. Musk shrews, by contrast, vomited readily at 0.5 to 3 percent isoflurane, up to 11.8 plus or minus 2.4 episodes ¹⁹. The contrast rests on the one agent both species received, which is the right way to read it: the same drug, opposite answers. Which one models the patient depends on which drug is being studied, and that cannot be known in advance.

The musk shrew has become the standard small emetic animal, and the work on it illustrates both the value and the fragility. One study sequenced the whole transcriptome of the solitary nucleus in a shrew model of postoperative vomiting, found evidence of suppressed GABA-B receptor signaling, and showed that baclofen, a GABA-B agonist, eliminated the vomiting ⁸⁸. The authors also report, honestly, that the sham group and the surgery group both had a median of six emetic episodes — that is, the surgery added nothing to the vomiting caused by handling and anesthesia — which considerably weakens the model the conclusion rests on. Emesis in this animal is now scored by machine: a deep learning system reached 98.92 percent overall accuracy against human observers, validated across eight different emetic drugs ⁸⁹. The measurement has become excellent. What it measures is still a shrew.

The bridge being built is molecular. A 2024 paper characterized a toxin tethered to a GLP-1-receptor ligand that kills GLP-1-receptor cells in human tissue, in mice and in the musk shrew, specifically so that the same cell population can be removed in a species that vomits and in a species that does not ²⁰. The cross-species cell atlas of the dorsal vagal complex makes the same point from the other direction: mouse and rat, two rodents, already differ in their cell populations ⁴, which sets a floor on how much confidence any single-species result deserves. And on the human side, the receptor pharmacologists issued the warning thirty years ago, after finding that 5-HT3 receptors sit in parts of the human forebrain where animals do not have them and bind drugs differently: predicting a clinical response from animal data, they wrote, requires caution ¹⁵.

What is still unknown, and what would settle it

The largest gap is nausea. Vomiting is a motor act that can be recorded from nerves in an anesthetized animal; nausea is a feeling that only a person can report. The two come apart pharmacologically: NK1 receptor antagonists block vomiting from many causes while doing much less, and for some stimuli apparently nothing, for the sensation — at doses that occupy nearly all the receptors in the relevant brain regions ²⁵. The accepted explanation is that nausea requires the solitary nucleus to project upward into forebrain circuits, and that those circuits are what we do not understand ^25,61. Nothing about this is settled; the 2004 review that set out the serotonin story also said, in plain terms, that the precise role of serotonin in the occurrence of vomiting has not been fully elucidated ⁵³, and that remains closer to true than most textbook accounts suggest.

The second gap is human anatomy at the cellular level. The subnuclei can be recognized in human sections and the 5-HT3 receptors can be counted in human tissue ^14,15,44, but the cell-type catalogs, the wiring diagrams and the functional maps are all rodent, and the cross-species atlas shows that even mouse and rat differ ⁴. There is no human counterpart to watching organ-tuned neurons light up ²¹, and the brainstem functional MRI that exists in people resolves a region, not a cell type ^85,86. A human single-nucleus atlas of the dorsal vagal complex, done on postmortem tissue, is technically possible and would be the single most useful thing anyone could add.

The third gap is that almost everything known is static. The work on orexin — the wake-promoting hypothalamic peptide — and the oxygen-sensing chemoreflex shows that the same pathway gives a different answer depending on the time of day ³⁹; the vagus nerve stimulation work shows that most of what reaches this nucleus from the periphery depends on descending traffic from above ³⁸; the stress work shows the forebrain acting down through it onto the stomach ⁸¹; and the astrocyte work shows that even the supporting cells respond in several directions at once ³². A structure described as a relay is in fact a place where the state of the whole organism determines what a signal means. Even the resident bacteria of the gut appear in the ledger: in 2026, mice lacking the Sirtuin 3 gene had an altered gut microbiome, raised choline in the blood, lowered GABA in the solitary nucleus and changed mood-related behavior, and cutting the vagus below the diaphragm returned those measures toward those of normal mice — a correlation in mice with a surgical test attached, not a demonstrated mechanism ⁹⁰. None of the standard descriptions capture this, and no experiment has yet mapped how the state is set.

The fourth is the status of claims that get repeated. The idea that the solitary nucleus contains the vomiting pattern generator was held for a decade and is wrong ^3,8,26. The idea that substance P is the transmitter at the vagal synapse there was reasonable and is wrong ^9,11. The idea that the brain's GLP-1 neurons are the relay for the gut's GLP-1, and the route by which GLP-1 drugs suppress appetite, was the working model and is wrong in mice ^12,63. The idea that somatostatin neurons in the taste end of the nucleus are bitter-specific is currently being walked back in a preprint ⁷. The idea that the serotonin route carries the whole of chemotherapy sickness has just been narrowed as well: mice bred without the 5-HT3 receptor are made as ill by cisplatin as mice that have it ⁵². The hypothesis that GLP-1 signaling explains area postrema syndrome is one center's unreplicated preprint and should be treated as a question ⁸⁰. The nucleus has a long record of inviting tidy stories, and a longer record of not containing them. Even the injury response is less local than it looks: cutting one laryngeal nerve in a mouse produced a response on both sides of this nucleus ⁸⁴.

The tests, one by one

• MRI of the dorsal medullaThe only routine way a clinician looks at this structure in a living person; lesions are inferred from the shape and position of the damaged area ^16,31.Suspected lateral medullary stroke; unexplained hiccups, nausea or vomiting lasting at least two daysGood enough to classify lateral medullary infarcts by level and to predict which symptoms a patient will have: in 130 consecutive patients, rostral lesions gave dysphagia and facial weakness significantly more often than caudal ones ¹⁶.It shows a territory, not a nucleus, and never a cell type. Area postrema syndrome is routinely mistaken for gastrointestinal disease, delaying immunotherapy by weeks to months ⁸⁰.
• Single-cell and single-nucleus RNA sequencingReads which genes are switched on in each individual cell, so cell types are defined by contents rather than appearance ^4,37.Building the parts list of the nucleus and of the vagal ganglia that feed itHighly reproducible and now cross-validated: 18 distinct neuron types in the mouse nodose ganglion ³⁷; 123 cell identities across more than 180,000 harmonized mouse and rat cells of the dorsal vagal complex ⁴.It is a snapshot of gene expression, not of activity or function, and the catalogs differ between mouse and rat ⁴, which sets a limit on what they say about humans.
• Two-photon calcium imaging in the living brainstemWatches hundreds of neurons fluoresce as their internal calcium rises with activity, while organs are stimulated ²¹.Mapping which neuron answers to which organ, in anesthetized miceResolved organ-specific tuning and a topographic arrangement, and detected the blurring of that map when inhibition was blocked ^21,34.Mice only, surgically exposed brainstem, and limited so far to the gut and upper airway. There is no human version.
• c-Fos immunolabelingStains for a protein made when a neuron has recently been active, giving a map of what responded to something hours ago ^18,84.The standard cheap survey of which parts of the nucleus a stimulus engagedReliable as a marker of recent activity, and sensitive enough to show that cutting one laryngeal nerve in a mouse activates both sides of this nucleus ⁸⁴.It cannot tell excitation from the activation of an inhibitory cell, gives no timing within the window, and misses neurons that respond without making c-Fos.
• Brainstem functional MRI during ear stimulationThe only method that looks at this nucleus working in a conscious human being ^85,86.Testing whether non-invasive vagus nerve stimulation reaches the brainstem, and where on the ear to place itIn 37 healthy volunteers, only stimulation at the cymba conchae activated both the solitary nucleus and the locus coeruleus significantly more than an earlobe sham ⁸⁵.A group average in a structure a few millimeters wide, at the limit of resolution. In a 2026 randomized trial, connectivity changes differed between vagal and non-vagal ear stimulation while the pain relief did not, so an imaging change here does not establish the route ⁸⁶. The feeling itself is also hard to provoke in a scanner: of 17 people given 21 minutes of sinusoidal galvanic vestibular stimulation in 2026, only four felt any nausea, and the signal changes that tracked the stimulation were in the nucleus accumbens and locus coeruleus, not in this nucleus ⁵⁸.

Where it connects

Sources

90 sources, numbered as they are cited. Every one was checked against PubMed or its publisher before it was cited here; the note under each says what it shows and what it does not.

1. 1
   Cutsforth-Gregory JK, Benarroch EE. Nucleus of the solitary tract, medullary reflexes, and clinical implications.Neurology · 2017 · 88(12):1187-1196doi:10.1212/WNL.0000000000003751 · PMID 28202704

   A neurologist's overview of the solitary nucleus as the hub of the medullary reflexes, written around a case. The PubMed record carries no abstract, so it is used here only for the broad clinical framing it is known for and not for any specific number.

2. 2
   Altschuler SM, Bao XM, Bieger D, Hopkins DA, Miselis RR. Viscerotopic representation of the upper alimentary tract in the rat: sensory ganglia and nuclei of the solitary and spinal trigeminal tracts.J Comp Neurol · 1989 · 283(2):248-68doi:10.1002/cne.902830207 · PMID 2738198

   The tracing study that drew the body map. In 172 rats, 0.5 to 65 microliters of horseradish peroxidase and related tracers were injected into the ninth and tenth cranial nerves and into the wall of the upper gut. Soft palate and pharynx labeled the interstitial and intermediate subnuclei; the esophagus labeled subnucleus centralis; the stomach wall labeled subnucleus gelatinosus. Rat anatomy, not human.

3. 3
   Hornby PJ. Central neurocircuitry associated with emesis.Am J Med · 2001 · 111 Suppl 8A:106S-112Sdoi:10.1016/s0002-9343(01)00849-x · PMID 11749934

   The review that states the correction plainly: 'there is not an isolated vomiting center, but rather groups of loosely organized neurons throughout the medulla that may be activated in sequence by a CPG.' Also lists the functional subnuclei (centralis for swallowing, gelatinosus for gastric sensation, intermediate and interstitial for laryngeal and pharyngeal sensation, medial for baroreceptors, ventrolateral for respiration) and places the action of NK1 antagonists and of THC in the dorsal vagal complex. A review, not new data.

4. 4
   Hes C, Tomlinson AJ, Michielsen L, Murdoch HJ, Soltani F, Kokoeva MV, Sabatini PV. A unified rodent atlas reveals the cellular complexity and evolutionary divergence of the dorsal vagal complex.eLife · 2025 · 14:e106217doi:10.7554/eLife.106217 · PMID 41055205

   Single-nucleus RNA sequencing harmonized across mouse and rat: more than 180,000 cells and 123 cell identities at five levels of resolution, including Gfral-expressing cells, calcium-permeable astrocytes and a leptin-receptor neuron population in the rat area postrema. It also shows that mouse and rat differ, which is the point most relevant to humans.

5. 5
   Lang IM. Physiology of the Digestive Tract Correlates of Vomiting.J Neurogastroenterol Motil · 2023 · 29(1):20-30doi:10.5056/jnm22105 · PMID 36606433

   A current review of the motor sequence of vomiting, which says the pattern generator 'may be in the Bötzinger nucleus.' The hedge is the author's own and is worth keeping.

6. 6
   Travers S, Breza J, Harley J, Zhu J, Travers J. Neurons with diverse phenotypes project from the caudal to the rostral nucleus of the solitary tract.J Comp Neurol · 2018 · 526(14):2319-2338doi:10.1002/cne.24501 · PMID 30325514

   Viral tracing in mice showing that the gut half of the nucleus talks to the taste half: glutamatergic, GABAergic and catecholaminergic projections all run from caudal to rostral, densest for the glutamatergic and sparsest for the catecholaminergic. Anatomy only.

7. 7
   Kalyanasundar B, Klimovich C, Travers S. Somatostatin neurons in the rostral nucleus of the solitary tract are functionally heterogeneous.bioRxiv · 2026 · preprintdoi:10.64898/2026.04.21.719923 · PMID 42079276

   A 2026 preprint, not yet peer reviewed, that walks back an earlier claim: somatostatin neurons of the taste end of the nucleus are not bitter-specific. Most responded best to bitter but many responded to other tastes, some non-somatostatin neurons responded to bitter, and roughly equal numbers of these cells carried excitatory and inhibitory markers. Mice.

8. 8
   Fukuda H, Koga T. Non-respiratory neurons in the Bötzinger complex exhibiting appropriate firing patterns to generate the emetic act in dogs.Neurosci Res · 1992 · 14(3):180-94doi:10.1016/0168-0102(92)90079-r · PMID 1331922

   The experiment that moved the generator out of the solitary nucleus. In decerebrate dogs, 78 non-respiratory neurons in the Bötzinger complex responded to vagal stimulation at a mean latency of 387 milliseconds; half burst in time with each retch. Cooling the medial solitary nucleus abolished retching and the bursts in all 11 such neurons tested, while stimulating the opposite vagus still worked. The solitary nucleus is the gate; the rhythm is made further forward. Dogs, decerebrate and paralyzed.

9. 9
   Furukawa N, Hatano M, Fukuda H, Koga T. Non-N-methyl-D-aspartate receptors may mediate the transmission of emetic signals between visceral vagal afferents and the solitary nucleus in dogs.Neurosci Lett · 1998 · 258(1):53-6doi:10.1016/s0304-3940(98)00859-3 · PMID 9876050

   In decerebrate dogs the NMDA blocker MK-801 changed nothing, while the non-NMDA blocker NBQX abolished both the salivation and the retching produced by vagal stimulation. Small study, one species, but it names the transmitter class at the first synapse.

10. 10
    Shiroshita Y, Koga T, Fukuda H. Capsaicin in the 4th ventricle abolishes retching and transmission of emetic vagal afferents to solitary nucleus neurons.Eur J Pharmacol · 1997 · 339(2-3):183-92doi:10.1016/s0014-2999(97)01370-8 · PMID 9473134

    Capsaicin at 33 millimolar and resiniferatoxin at 160 micromolar applied to the floor of the fourth ventricle first excited solitary nucleus neurons, sometimes producing retching, then silenced both, while direct stimulation of the nucleus still provoked retching. Evidence that the emetic vagal fibers are capsaicin-sensitive C fibers. Decerebrate dogs.

11. 11
    Fukuda H, Koga T, Furukawa N, Nakamura E, Shiroshita Y. The tachykinin NK1 receptor antagonist GR205171 prevents vagal stimulation-induced retching but not neuronal transmission from emetic vagal afferents to solitary nucleus neurons in dogs.Brain Res · 1998 · 802(1-2):221-31doi:10.1016/s0006-8993(98)00630-1 · PMID 9748589

    The reversal of the authors' own hypothesis. GR205171 at 0.05 to 0.7 milligrams per kilogram stopped retching within five minutes yet left solitary nucleus firing and evoked potentials unchanged, so substance P is not the transmitter carrying the emetic signal into the nucleus in dogs. Dogs.

12. 12
    Brierley DI, Holt MK, Singh A, de Araujo A, McDougle M, Vergara M, et al. Central and peripheral GLP-1 systems independently suppress eating.Nat Metab · 2021 · 3(2):258-273doi:10.1038/s42255-021-00344-4 · PMID 33589843

    The reversal of the unified gut-brain GLP-1 circuit. In mice, the GLP-1-making neurons of the solitary nucleus encode satiation but get their vagal input mainly from oxytocin-receptor neurons rather than GLP-1-receptor neurons, and they are not needed for a GLP-1 drug to suppress eating; activating them alongside semaglutide suppressed eating more than semaglutide alone. Mice.

13. 13
    Jiang W, Skoug C, Rodrigues I, Ciabatti E, Gribble FM, Reimann F, Brierley DI, Holt MK, Trapp S. Brainstem GLP-1 neurons modulate physiological satiation and drive sustained weight loss in obese mice.Mol Metab · 2026 · 107:102347doi:10.1016/j.molmet.2026.102347 · PMID 41786245

    The most recent test of the same cells. The GLP-1 neurons of the solitary nucleus and of the intermediate reticular nucleus set meal size, their total number runs inverse to weight gain, ablating both delays satiation enough to cause weight gain on ordinary chow, and daily chemogenetic activation drove sustained weight loss in obese mice without notable adverse effects. Mice; 'without notable adverse effects' in a species that cannot vomit is a weak reassurance about nausea in people.

14. 14
    Barnes JM, Barnes NM, Costall B, Deakin JF, Ironside JW, Kilpatrick GJ, Naylor RJ, Rudd JA, Simpson MD, Slater P. Identification and distribution of 5-HT3 recognition sites within the human brainstem.Neurosci Lett · 1990 · 111(1-2):80-6doi:10.1016/0304-3940(90)90348-d · PMID 2336196

    Human postmortem autoradiography. Binding of the 5-HT3 ligand [3H]GR65630 was highest in the area postrema and area subpostrema at 112.1 femtomoles per milligram of protein, lower in the dorsal vagal complex at 22.9, and 3.9 in the olivary nucleus, with none detectable in the reticular formation beneath. Binding, not function.

15. 15
    Parker RM, Barnes JM, Ge J, Barber PC, Barnes NM. Autoradiographic distribution of [3H]-(S)-zacopride-labelled 5-HT3 receptors in human brain.J Neurol Sci · 1996 · 144(1-2):119-27doi:10.1016/s0022-510x(96)00211-0 · PMID 8994113

    Human brain autoradiography with a second ligand. The densest 5-HT3 receptor labeling was in the solitary nucleus, area postrema and spinal trigeminal nucleus, at 50 to 200 femtomoles per milligram tissue equivalent, against 4 to 17 in forebrain regions. The authors warn explicitly that human and animal 5-HT3 pharmacology differ enough to make animal predictions unsafe.

16. 16
    Kim JS. Pure lateral medullary infarction: clinical-radiological correlation of 130 acute, consecutive patients.Brain · 2003 · 126(Pt 8):1864-72doi:10.1093/brain/awg169 · PMID 12805095

    The largest clean series of Wallenberg's syndrome. In 130 consecutive patients, rostral lesions produced dysphagia and facial weakness significantly more often than caudal ones, angiography in 123 showed vertebral artery disease in 67 percent and posterior inferior cerebellar artery disease in 10 percent, and the causes were large vessel disease in 50 percent, arterial dissection in 15 percent, small vessel disease in 13 percent and cardiac embolism in 5 percent. The paper does not report vomiting or hiccup rates, so none are quoted here.

17. 17
    Horn CC, Kimball BA, Wang H, Kaus J, Dienel S, Nagy A, Gathright GR, Yates BJ, Andrews PLR. Why can't rodents vomit? A comparative behavioral, anatomical, and physiological study.PLoS One · 2013 · 8(4):e60537doi:10.1371/journal.pone.0060537 · PMID 23593236

    Behavioral testing across all three major groups of Rodentia, anatomical comparison with cat, ferret and musk shrew, and an in situ brainstem preparation. Apomorphine, veratrine and copper sulfate produced no retching or vomiting in any rodent tested; the authors attribute the absence chiefly to a missing brainstem neurological component rather than to anatomy alone. The study behind a sentence this page repeats throughout.

18. 18
    Zhang C, Kaye JA, Cai Z, Wang Y, Prescott SL, Liberles SD. Area Postrema Cell Types that Mediate Nausea-Associated Behaviors.Neuron · 2021 · 109(3):461-472.e5doi:10.1016/j.neuron.2020.11.010 · PMID 33278342

    Single-nucleus sequencing of the area postrema found only a few neuron types; genetically targeted excitatory neurons produced nausea-related behavior, and one type carried the aversion caused by several different poisons. Mice cannot vomit, so the readout is aversion and malaise behavior, not emesis.

19. 19
    Horn CC, Meyers K, Pak D, Nagy A, Apfel CC, Williams BA. Post-anesthesia vomiting: impact of isoflurane and morphine on ferrets and musk shrews.Physiol Behav · 2012 · 106(4):562-8doi:10.1016/j.physbeh.2012.03.031 · PMID 22504494

    Thirty-four ferrets and 36 musk shrews. Cisplatin produced 129.8 plus or minus 22.0 retches and 13.7 plus or minus 2.3 vomits in ferrets and morphine peaked at 0.9 milligrams per kilogram, but isoflurane at 2 to 4 percent never made a ferret vomit, while musk shrews vomited readily at 0.5 to 3 percent. Two species, opposite answers, same stimulus.

20. 20
    Fulton S, Horn CC, Zhang C. Characterizing a new tool to manipulate area postrema GLP1R neurons across species.Physiol Behav · 2024 · 276:114474doi:10.1016/j.physbeh.2024.114474 · PMID 38272107

    A ligand-conjugated saporin that kills GLP-1-receptor cells in human tissue, mice and the musk shrew, which can vomit. A methods paper: it provides the means to test the circuit in a vomiting species, not yet the result.

21. 21
    Ran C, Boettcher JC, Kaye JA, Gallori CE, Liberles SD. A brainstem map for visceral sensations.Nature · 2022 · 609(7926):320-326doi:10.1038/s41586-022-05139-5 · PMID 36045291

    Two-photon calcium imaging of the solitary nucleus in living mice. Individual neurons were tuned to signals from particular organs and were arranged topographically by body position; some mechanical and chemical inputs from one organ converged centrally; blocking brainstem inhibition broadened tuning and disorganized the map. Mice, gut and upper airway only.

22. 22
    Schild JH, Clark JW, Canavier CC, Kunze DL, Andresen MC. Afferent synaptic drive of rat medial nucleus tractus solitarius neurons: dynamic simulation of graded vesicular mobilization, release, and non-NMDA receptor kinetics.J Neurophysiol · 1995 · 74(4):1529-48doi:10.1152/jn.1995.74.4.1529 · PMID 8989391

    A mathematical model, fitted to the authors' recordings, of 'this first synapse in the baroreceptor reflex.' It predicts that above roughly 20 to 30 hertz the synapse itself limits how much a solitary nucleus neuron can follow its input. A simulation, not an experiment.

23. 23
    Beaumont E, Campbell RP, Andresen MC, Scofield S, Singh K, Libbus I, KenKnight BH, Snyder L, Cantrell N. Cervical vagus nerve stimulation augments spontaneous discharge in second- and higher-order sensory neurons in the rat nucleus of the solitary tract.Am J Physiol Heart Circ Physiol · 2017 · 313(2):H354-H367doi:10.1152/ajpheart.00070.2017 · PMID 28476920

    At the mild intensities used clinically, vagus nerve stimulation raised firing in about a third of rat solitary nucleus neurons, but directly activated only myelinated fibers: 3 of 4 neurons with single-synapse A-fiber input, 6 of 42 with single-synapse C-fiber input, and 16 of 21 multi-synapse neurons. Commercial conflict of interest: two authors were employees of Cyberonics, which makes vagus nerve stimulators.

24. 24
    Lindley C, Blower P. Oral serotonin type 3-receptor antagonists for prevention of chemotherapy-induced emesis.Am J Health Syst Pharm · 2000 · 57(18):1685-97doi:10.1093/ajhp/57.18.1685 · PMID 11006796

    Clinical review. Oral granisetron, 2 milligrams once daily or 1 milligram twice daily, had been studied in more than 4,500 patients receiving highly or moderately emetogenic chemotherapy, with total control of vomiting in 44 to 60 percent and complete response in 70 to 94 percent. A review of trials, not a single trial, and the endpoint is vomiting rather than nausea.

25. 25
    Andrews PLR, Golding JF, Sanger GJ. An assessment of the effects of neurokinin1 receptor antagonism against nausea and vomiting: Relative efficacy, sites of action and lessons for future drug development.Br J Clin Pharmacol · 2023 · 89(12):3468-3490doi:10.1111/bcp.15852 · PMID 37452618

    The honest accounting: NK1 antagonists block vomiting but do much less, and possibly nothing, for the feeling of nausea, even though imaging shows very high receptor occupancy at the doses used in patients. The authors place the drug action in the pattern generator and the solitary nucleus, and conclude that we do not understand nausea. A review.

26. 26
    Carpenter DO. Neural mechanisms of emesis.Can J Physiol Pharmacol · 1990 · 68(2):230-6doi:10.1139/y90-036 · PMID 2178747

    The clearest statement of the belief that was later corrected: 'it is likely that the sequence of muscle excitation and inhibition is controlled by a central pattern generator located in the nucleus tractus solitarius.' A review proposing a hypothesis, not an experiment.

27. 27
    Saito Y. Reflections on the brainstem dysfunction in neurologically disabled children.Brain Dev · 2009 · 31(7):529-36doi:10.1016/j.braindev.2009.01.009 · PMID 19329267

    Clinical review of children with dorsal brainstem injury. Damage in the territory of the solitary nucleus after hypoxic-ischemic injury produced dysphagia, gastroesophageal reflux, laryngeal stridor and abnormal breathing; children with Leigh syndrome and similar lesions also had hiccups and vomiting. Observational, and in a mixed group of diagnoses.

28. 28
    Minami M, Endo T, Hirafuji M. [Role of serotonin in emesis].Nihon Yakurigaku Zasshi · 1996 · 108(5):233-42doi:10.1254/fpj.108.233 · PMID 8974084

    Japanese-language review with English abstract. States that enterochromaffin cells of the gut lining make and release about 80 percent of all the serotonin in the body, that cisplatin raises serotonin in the intestinal lining and the area postrema, and that cutting the abdominal vagus suppresses cisplatin-induced vomiting. Animal work.

29. 29
    Balaban CD, Beryozkin G. Vestibular nucleus projections to nucleus tractus solitarius and the dorsal motor nucleus of the vagus nerve: potential substrates for vestibulo-autonomic interactions.Exp Brain Res · 1994 · 98(2):200-12doi:10.1007/BF00228409 · PMID 8050507

    Tract tracing in rabbits found axons running from the caudal medial and inferior vestibular nuclei to the solitary nucleus on both sides, with fewer reaching the dorsal motor nucleus. An anatomical substrate for motion sickness, not a demonstration of it.

30. 30
    Gagliuso AH, Chapman EK, Martinelli GP, Holstein GR. Vestibular neurons with direct projections to the solitary nucleus in the rat.J Neurophysiol · 2019 · 122(2):512-524doi:10.1152/jn.00082.2019 · PMID 31166818

    The chemistry of that pathway in rats: the vestibular cells projecting to the solitary nucleus are glutamatergic, carry vesicular glutamate transporter 2 and the modulator imidazoleacetic acid ribotide, and are wrapped in inhibitory terminals. Anatomy and immunolabeling; no behavior tested.

31. 31
    Eliaçık S. Intractable hiccups as a brainstem red flag: demyelinating disorders with emphasis on neuromyelitis optica spectrum disorder and area postrema syndrome.Neurol Sci · 2026 · 47(9)doi:10.1007/s10072-026-09288-7 · PMID 42603218

    Narrative review searched to January 2026. Area postrema syndrome is defined as otherwise unexplained hiccups, nausea or vomiting lasting at least 48 hours and is a core diagnostic feature of neuromyelitis optica spectrum disorder, where the injury is to astrocytes in the dorsal medulla. A narrative review, so no pooled frequencies.

32. 32
    Souza MM de, Zacharias LR, Leão RM. Complex dynamics of glutamate-induced calcium responses in astrocytes from the nucleus of the solitary tract of mice.Braz J Med Biol Res · 2026 · 59:e15017doi:10.1590/1414-431X2025e15017 · PMID 41637284

    Only a subset of astrocytes in the subpostremal solitary nucleus raised their internal calcium when given glutamate, a smaller fraction lowered it, and blocking action potentials with tetrodotoxin halved the proportion that rose. The nucleus is not only neurons, and its non-neuronal cells are not uniform either. Mouse brain slices.

33. 33
    Huang TX, Wang S, Ran C. Interoceptive processing in the nucleus of the solitary tract.Curr Opin Neurobiol · 2025 · 93:103021doi:10.1016/j.conb.2025.103021 · PMID 40239364

    Current review of the nucleus as a sensory structure: organ signals form a topographic map, are computed locally, are modulated by descending input from higher brain regions, and are distributed by projection neurons. Written from rodent work; it makes no human claims.

34. 34
    Ran C, Boettcher JC, Kaye JA, Gallori CE, Liberles SD. Publisher Correction: A brainstem map for visceral sensations.Nature · 2022 · 611(7934):E6doi:10.1038/s41586-022-05414-5 · PMID 36241717

    A publisher's correction to the 2022 paper above. It is a correction notice, not a retraction and not an expression of concern; the findings stand.

35. 35
    Gasparini S, Zhu H, Peraza Munuzuri AS, Leuang ML, Fazan FS, Geerling JC. Molecular Ontology Predicts Output Connections From the Nucleus of the Solitary Tract.J Comp Neurol · 2026 · 534(8):e70188doi:10.1002/cne.70188 · PMID 42561032

    Cell-type-specific tracing in mice. Lmx1b-expressing excitatory neurons project widely to brainstem and forebrain while inhibitory neurons project mostly within the brainstem, and the Th, Cck, Npff, Pdyn and Hsd11b2 subpopulations each have a distinct output pattern. Anatomy in mice; it does not test what any of these pathways do.

36. 36
    Williams EK, Chang RB, Strochlic DE, Umans BD, Lowell BB, Liberles SD. Sensory Neurons that Detect Stretch and Nutrients in the Digestive System.Cell · 2016 · 166(1):209-21doi:10.1016/j.cell.2016.05.011 · PMID 27238020

    The paper that split the vagus by gene. GPR65-marked neurons sense nutrients in the gut lining; GLP1R-marked neurons sense stretch of the stomach and intestine; the two have different endings, different conduction speeds and different central targets. Mice.

37. 37
    Kupari J, Häring M, Agirre E, Castelo-Branco G, Ernfors P. An Atlas of Vagal Sensory Neurons and Their Molecular Specialization.Cell Rep · 2019 · 27(8):2508-2523.e4doi:10.1016/j.celrep.2019.04.096 · PMID 31116992

    Single-cell sequencing of both vagal ganglia in mice. The jugular ganglion, which comes from the neural crest, holds neuron types that resemble skin sensory neurons; the nodose ganglion, which comes from a placode, holds 18 distinct types dedicated to the internal organs. The anatomical basis for saying the vagus is not one nerve.

38. 38
    Cooper CM, Farrand AQ, Andresen MC, Beaumont E. Vagus nerve stimulation activates nucleus of solitary tract neurons via supramedullary pathways.J Physiol · 2021 · 599(23):5261-5279doi:10.1113/JP282064 · PMID 34676533

    Cutting the rat brainstem away from everything above it dropped the resting firing of second-order solitary nucleus neurons by 85 percent and abolished the stimulation-driven activation of higher-order neurons, so much of what vagus nerve stimulation does to this nucleus depends on descending traffic from the rest of the brain. Anesthetized rats.

39. 39
    Ben Musa R, Khodadadi-Mericle F, Kline DD, Hasser EM, Cummings KJ. Orexin facilitates the peripheral chemoreflex in the active phase via corticotropin-releasing hormone neurons that project to the nucleus of the solitary tract.Function (Oxf) · 2025 · 7(1):e0842025doi:10.1152/function.084.2025 · PMID 41401425

    In rats, silencing the hypothalamic corticotropin-releasing hormone neurons that project to the solitary nucleus cut the breathing response to low oxygen by about half, but only during the animals' active phase; activating the same pathway did nothing. Evidence that what this nucleus does with a signal depends on the time of day.

40. 40
    Kalia M, Mesulam MM. Brain stem projections of sensory and motor components of the vagus complex in the cat: I. The cervical vagus and nodose ganglion.J Comp Neurol · 1980 · 193(2):435-65doi:10.1002/cne.901930210 · PMID 7440777

    Thirty-two cats, horseradish peroxidase and tritiated amino acid tracing. Vagal afferent fibers reached most of the subnuclei of the solitary nucleus bilaterally, more intensely on the injected side, and also ended in the area postrema and around neurons of the dorsal motor nucleus of the vagus, which the authors read as a substrate for monosynaptic vagal reflexes. The pre-1989 anatomical groundwork for the body map; cat, not human.

41. 41
    Altschuler SM, Escardo J, Lynn RB, Miselis RR. The central organization of the vagus nerve innervating the colon of the rat.Gastroenterology · 1993 · 104(2):502-9doi:10.1016/0016-5085(93)90419-d · PMID 8425692

    Tracer injected into the colon of 29 rats labeled sensory endings only in the commissural and medial subnuclei of the solitary nucleus, and labeling from the rectum did not reach the brainstem at all. Shows that the map is partial: the lower bowel is a spinal, not a vagal, story.

42. 42
    Jin H, Fishman ZH, Ye M, Wang L, Zuker CS. Top-Down Control of Sweet and Bitter Taste in the Mammalian Brain.Cell · 2021 · 184(1):257-271.e16doi:10.1016/j.cell.2020.12.014 · PMID 33417862

    Identified the brainstem neurons that carry sweet and bitter from the tongue toward the cortex in mice, the somatostatin population of the rostral solitary nucleus among them, and showed strong positive and negative feedback onto those incoming signals from taste cortex and amygdala. The bitter-specificity part of this account is what the 2026 preprint walks back; named here because the reversal is only legible if the superseded claim is cited.

43. 43
    Rodriguez HR, Park G, Vinson LT, Lee H. Gustatory Connectivity and Central Projections of the Mouse Petrosal-Nodose-Jugular Ganglia Complex.eNeuro · 2026doi:10.1523/ENEURO.0227-26.2026 · PMID 42844011

    An improved surgical route for delivering viral vectors to the mouse petrosal, nodose and jugular ganglia. Gustatory neurons of the complex terminate in the rostral solitary nucleus; the same complex also projects to the area postrema, nucleus ambiguus, lateral reticular nucleus, spinal trigeminal tract and cervical spinal cord. Mouse anatomy, and a methods paper.

44. 44
    Rüb U, Brunt ER, Del Turco D, de Vos RAI, Gierga K, Paulson H, Braak H. Guidelines for the pathoanatomical examination of the lower brain stem in ingestive and swallowing disorders and its application to a dysphagic spinocerebellar ataxia type 3 patient.Neuropathol Appl Neurobiol · 2003 · 29(1):1-13doi:10.1046/j.1365-2990.2003.00437.x · PMID 12581335

    Human neuropathology. Serial 100-micrometer sections of one patient with spinocerebellar ataxia type 3 showed neuron loss and astrogliosis in nearly all the swallowing-related lower brainstem nuclei, the gelatinous, medial, parvocellular and pigmented solitary nuclei among them. A single case, but it names the human subnuclei by sight rather than by analogy with the rat.

45. 45
    Wang SC, Borison HL. The vomiting center; a critical experimental analysis.Arch Neurol Psychiatry · 1950 · 63(6):928-41doi:10.1001/archneurpsyc.1950.02310240087005 · PMID 15426437

    The paper that named a vomiting center in the medulla, from stimulation and ablation experiments in cats and dogs. The PubMed record carries no abstract, so it is cited here only for the doctrine it is universally credited with establishing, which the rest of this page spends its history correcting.

46. 46
    Borison HL, Brizzee KR. Morphology of emetic chemoreceptor trigger zone in cat medulla oblongata.Proc Soc Exp Biol Med · 1951 · 77(1):38-42doi:10.3181/00379727-77-18670 · PMID 14844387

    The companion claim: a chemoreceptor trigger zone in the area postrema, anatomically separate from the supposed vomiting center, able to sample the blood. No abstract in the PubMed record; cited for the framing only.

47. 47
    Borison HL, Wang SC. Physiology and pharmacology of vomiting.Pharmacol Rev · 1953 · 5(2):193-230PMID 13064033

    The review that fixed the doctrine this page's history exists to correct: a vomiting center in the medulla, driven in part by a separate chemoreceptor trigger zone in the area postrema. The PubMed record carries no abstract and no DOI, so it is cited for that framing and for nothing quantitative.

48. 48
    Lang IM. Digestive tract motor correlates of vomiting and nausea.Can J Physiol Pharmacol · 1990 · 68(2):242-53doi:10.1139/y90-038 · PMID 2178749

    The companion review on what the gut does during vomiting: the proximal stomach relaxes, a retrograde giant contraction sweeps the small intestine toward the mouth, and the esophagus pulls the stomach into the chest. It is explicit that nausea in animals has never been demonstrated, only inferred.

49. 49
    Fukuda H, Koga T, Furukawa N, Nakamura E, Shiroshita Y. The tachykinin NK1 receptor antagonist GR205171 abolishes the retching activity of neurons comprising the central pattern generator for vomiting in dogs.Neurosci Res · 1999 · 33(1):25-32doi:10.1016/s0168-0102(98)00106-0 · PMID 10096468

    GR205171 at 25 to 50 micrograms per kilogram intravenously abolished the slow build-up and the retching bursts of pattern generator neurons but left the fast initial response intact, placing the antiemetic action at the generator or on the path from the solitary nucleus to it. One neuron in each of nine decerebrate dogs.

50. 50
    Hsu JY, Crawley S, Chen M, Ayupova DA, Lindhout DA, Higbee J, et al. Non-homeostatic body weight regulation through a brainstem-restricted receptor for GDF15.Nature · 2017 · 550(7675):255-259doi:10.1038/nature24042 · PMID 28953886

    Identifies GFRAL as the receptor for GDF15 and places GFRAL neurons exclusively in the area postrema and solitary nucleus of the mouse brainstem; mice lacking Gfral were resistant to chemotherapy-induced anorexia and weight loss. Commercial conflict of interest: nearly all authors were employees of NGM Biopharmaceuticals, with co-authors from Merck Research Labs.

51. 51
    Bai L, Mesgarzadeh S, Ramesh KS, Huey EL, Liu Y, Gray LA, et al. Genetic Identification of Vagal Sensory Neurons That Control Feeding.Cell · 2019 · 179(5):1129-1143.e23doi:10.1016/j.cell.2019.10.031 · PMID 31730854

    Target-specific single-cell sequencing of the gut-innervating vagus in mice. Food intake was most sensitive to stimulating intestinal stretch receptors, while nutrient-activated mucosal fibers had no effect on intake, and only the stretch receptors silenced the hunger-promoting AgRP neurons of the hypothalamus. A direct contradiction of the simple idea that nutrient sensing in the gut lining is what stops a meal.

52. 52
    Fujii R, Kawai H, Kai M, Kondo M. Integrated behavioral and biological assessment reveals 5-HT3 receptor-independent nausea/malaise-like responses in cisplatin-treated mice.Biochem Biophys Res Commun · 2026 · 819:153812doi:10.1016/j.bbrc.2026.153812 · PMID 42025017

    5-HT3 receptor knockout mice showed cisplatin-induced reduction in food intake, weight loss, reduced spontaneous activity, pica and conditioned taste aversion to the same extent as wild-type, with increased c-Fos-positive and c-Fos/GFRAL double-positive cells in the area postrema and solitary nucleus of both genotypes and plasma GDF15 elevated for at least a day. Mice, so the readout is malaise behavior and not vomiting.

53. 53
    Minami M, Endo T, Hamaue N, Hirafuji M. [Serotonin and anticancer drug-induced emesis].Yakugaku Zasshi · 2004 · 124(8):491-507doi:10.1248/yakushi.124.491 · PMID 15297719

    A later review by the same group, notable for saying outright that 'the precise role of 5-HT in the occurrence of vomiting has not been fully elucidated.' Japanese-language, English abstract.

54. 54
    Borison HL. Area postrema: chemoreceptor circumventricular organ of the medulla oblongata.Prog Neurobiol · 1989 · 32(5):351-90doi:10.1016/0301-0082(89)90028-2 · PMID 2660187

    Borison's own later review of the area postrema as a circumventricular organ — a region outside the blood-brain barrier — with roles in vomiting, breathing and circulation. The PubMed record carries no abstract; cited for the standing description of the region's permeability rather than for a number.

55. 55
    Najjar NR, De Jonghe BC, Hayes MR, Doyle RP. From stress cytokine to drug target: The translational landscape of growth differentiation factor 15.J Pharmacol Exp Ther · 2026 · 393(11):105036doi:10.1016/j.jpet.2026.105036 · PMID 42880196

    Medicinal-chemistry review of the GDF15-GFRAL-RET axis: structure-activity work, biologics and small molecules, patents and trials, across cancer anorexia-cachexia, chemotherapy-induced nausea and vomiting, obesity and hyperemesis gravidarum. Its authors are the group that developed the brain-penetrance-reduced GLP-1 conjugates cited elsewhere on this page, so its enthusiasm should be read with that in mind.

56. 56
    Borner T, Workinger JL, Tinsley IC, Fortin SM, Stein LM, Chepurny OG, et al. Corrination of a GLP-1 Receptor Agonist for Glycemic Control without Emesis.Cell Rep · 2020 · 31(11):107768doi:10.1016/j.celrep.2020.107768 · PMID 32553160

    Exendin-4 attached to dicyanocobinamide, a corrin-ring precursor of vitamin B12, to reduce brain penetrance. In the musk shrew, which vomits, the conjugate improved glucose tolerance with a profound reduction in emetic events compared with unmodified exendin-4. Evidence that the emesis of this drug class is driven from inside the brain.

57. 57
    Borner T, Shaulson ED, Tinsley IC, Stein LM, Horn CC, Hayes MR, Doyle RP, De Jonghe BC. A second-generation glucagon-like peptide-1 receptor agonist mitigates vomiting and anorexia while retaining glucoregulatory potency in lean diabetic and emetic mammalian models.Diabetes Obes Metab · 2020 · 22(10):1729-1741doi:10.1111/dom.14089 · PMID 32410372

    The clearest localization of the side effect. Systemic exendin-4 reliably produced emesis in musk shrews while an equimolar dose of the vitamin B12 conjugate produced almost none; given directly into the brain, the two were equally emetic. Shrews and Goto-Kakizaki rats, not patients.

58. 58
    McCarthy B, Rim D, Sesa-Ashton G, Crawford LS, Henderson LA, Macefield VG. Cortical and subcortical correlates of perceptual responses during sinusoidal galvanic vestibular stimulation: a BOLD fMRI study.Exp Brain Res · 2026 · 244(12)doi:10.1007/s00221-026-07411-w · PMID 42878889

    Seventeen participants, 21 minutes of plus or minus 2 milliamp, 0.08 hertz sinusoidal galvanic vestibular stimulation during functional MRI. Only four reported incipient or overt nausea; the signal changes associated with the sustained stimulation were in the nucleus accumbens and locus coeruleus. The authors suggest the supine, head-fixed posture limited the nausea. A small human study, and a negative result for the solitary nucleus.

59. 59
    Singh P, Kuo B. Central Aspects of Nausea and Vomiting in GI Disorders.Curr Treat Options Gastroenterol · 2016 · 14(4):444-451doi:10.1007/s11938-016-0107-x · PMID 27734216

    Clinical review setting out the convergence of vagal, vestibular and chemoreceptor trigger zone input on the solitary nucleus, and arguing that chronic nausea travels a different road from acute nausea, one that resembles neuropathic pain. The authors call our understanding of central nausea pathways 'still very basic.'

60. 60
    Cangemi DJ, Kuo B. Practical Perspectives in the Treatment of Nausea and Vomiting.J Clin Gastroenterol · 2019 · 53(3):170-178doi:10.1097/MCG.0000000000001164 · PMID 30614944

    The clinical consequence of the same argument: conventional antiemetics work poorly for chronic nausea, and neuromodulators are used instead. The authors note the need for proper studies of these agents in chronic functional nausea, which is an admission that the evidence is thin.

61. 61
    Horn CC, Wallisch WJ, Homanics GE, Williams JP. Pathophysiological and neurochemical mechanisms of postoperative nausea and vomiting.Eur J Pharmacol · 2014 · 722:55-66doi:10.1016/j.ejphar.2013.10.037 · PMID 24495419

    Review of postoperative nausea and vomiting, which it attributes mainly to inhaled anesthetics and opioids, with young age, female sex, not smoking and a history of motion sickness as risk predictors. It states plainly that we still lack fundamental knowledge of the hindbrain generator and the forebrain pathways for nausea.

62. 62
    Holt MK, Cook DR, Brierley DI, Richards JE, Reimann F, Gourine AV, Marina N, Trapp S. PPG neurons in the nucleus of the solitary tract modulate heart rate but do not mediate GLP-1 receptor agonist-induced tachycardia in mice.Mol Metab · 2020 · 39:101024doi:10.1016/j.molmet.2020.101024 · PMID 32446875

    A second dissociation. Exendin-4 raised heart rate in mice through beta-adrenoceptors, and it did so just as well after the solitary nucleus GLP-1 neurons were destroyed, without ever activating them — yet switching those neurons on artificially did raise heart rate. The same cells can do a thing without being the ones a drug uses to do it.

63. 63
    Lewis JE, Montaner M, Nuzzaci D, James-Okoro PP, Harada N, Inagaki N, et al. Distinct brain regions mediate regulation of food intake in response to GIPR agonism or antagonism.Nat Metab · 2026 · 8(8):1669-1678doi:10.1038/s42255-026-01575-z · PMID 42498824

    Knocking the GIP receptor out of the mouse area postrema left the animals partly protected against diet-induced obesity, less responsive to the appetite-suppressing effect of acyl-GIP, and less able to have avoidance triggered by peptide YY prevented by that drug — a reduction, not an abolition. The extra weight loss from a GIP blocker added to liraglutide depended instead on hypothalamic receptors and, the authors state, was not mediated by solitary nucleus preproglucagon neurons. Mice; co-authors from the Indiana Biosciences Research Institute.

64. 64
    Holt MK, Valderrama N, Polanco MJ, Hayter I, Badenoch EG, Trapp S, Rinaman L. Modulation of stress-related behaviour by preproglucagon neurons and hypothalamic projections to the nucleus of the solitary tract.Mol Metab · 2024 · 91:102076doi:10.1016/j.molmet.2024.102076 · PMID 39603502

    These neurons also respond to psychological stress. Activating them moderately increased anxiety-like behavior in mice, and the effect differed by sex; the paper states that their role in anxiety-like behavior is controversial. Input came from the paraventricular hypothalamus, but not through corticotropin-releasing hormone receptors on the neurons themselves. One co-author was employed by GlaxoSmithKline Pharmaceuticals.

65. 65
    Ritchey CR, Rossi DJ, Peters JH. Characterization of NTS-to-VTA projection neurons reveals higher-order synaptic organization and distinct responsiveness to cholecystokinin.J Physiol · 2025 · 603(17):4817-4843doi:10.1113/JP288729 · PMID 40785037

    Mouse neurons projecting from the solitary nucleus to the ventral tegmental area were largely catecholaminergic and GLP-1 containing, were mostly connected to the vagus only indirectly through other neurons, and were mostly inhibited rather than excited by cholecystokinin. A reminder that most of the nucleus is not a simple relay.

66. 66
    Baumer-Harrison C, Aldaghma D, White AD, Applebey SV, Pataro AM, Mohiby AZ, et al. GLP-1R biased cAMP agonism maintains glycemic control with reduced malaise and emesis in preclinical mammalian models.Diabetes Obes Metab · 2026 · 28(3):2317-2328doi:10.1111/dom.70427 · PMID 41508705

    Exendin-4-Phe1, which favors cyclic AMP signaling over beta-arrestin recruitment, in 43 mice, 54 rats and 30 musk shrews. Fewer emetic episodes than exendin-4 in shrews and little to no pica in rats, with glucose tolerance equally improved in all three species, and c-Fos raised in nausea- and emesis-related regions in all of them. Commercial conflict of interest: several authors were employees of Lilly Research Laboratories.

67. 67
    Prescott SL, Umans BD, Williams EK, Brust RD, Liberles SD. An Airway Protection Program Revealed by Sweeping Genetic Control of Vagal Afferents.Cell · 2020 · 181(3):574-589.e14doi:10.1016/j.cell.2020.03.004 · PMID 32259485

    About 100 P2RY1-marked throat neurons per mouse; switching them on with light produced a whole coordinated defense — apnea, vocal fold closure, swallowing and expiratory reflexes — and removing them abolished the response to water and acid in the larynx. Mice.

68. 68
    Prescott SL, Liberles SD. Internal senses of the vagus nerve.Neuron · 2022 · 110(4):579-599doi:10.1016/j.neuron.2021.12.020 · PMID 35051375

    The standard current review of vagal sensory diversity across the respiratory, cardiovascular and digestive systems. Its closing admission is the useful one: the transduction mechanisms of most vagal sensory neurons are not known, and interoception lags far behind the external senses.

69. 69
    Mendez-Hernandez R, Braga I, Bali A, Yang M, de Lartigue G. Vagal Sensory Gut-Brain Pathways That Control Eating-Satiety and Beyond.Compr Physiol · 2025 · 15(2):e70010doi:10.1002/cph4.70010 · PMID 40229922

    Review integrating single-cell transcriptomics, circuit mapping and functional imaging, arguing directly against the passive-relay view of vagal afferent neurons: they encode stretch, nutrient content, metabolic state and microbial metabolites, and feed distinct brain circuits including those for reward and memory. A review, mostly of rodent work.

70. 70
    Su Y, Zhu Z, Sun X. Lung-Innervating Neurons in Lung Homeostasis and Diseases.Physiology (Bethesda) · 2025 · 41(4)doi:10.1152/physiol.00036.2025 · PMID 41259123

    Review of airway vagal sensory neurons: molecular diversity, projection patterns, terminal morphologies, and an explicit statement that how subtypes encode their inputs and how those signals are integrated centrally remain unresolved. A review.

71. 71
    Wei HK, Yu CD, Hu B, Zeng X, Ichise H, Li L, et al. Tumour-brain crosstalk restrains cancer immunity via a sensory-sympathetic axis.Nature · 2026 · 650(8103):1007-1016doi:10.1038/s41586-025-10028-8 · PMID 41639447

    Mouse lung adenocarcinoma recruits Npy2r-expressing vagal sensory neurons that signal to brainstem nuclei; the resulting sympathetic outflow suppressed anti-tumor immunity through beta-adrenergic signaling in alveolar macrophages, and disrupting the route genetically, pharmacologically or chemogenetically slowed tumor growth. Mice; a genetically defined vagal line, not a human finding.

72. 72
    Sun J, Fang Y, Tao W, Kang Y, Chen T, Zhong Q, Zhu J, Ma D, Zhang Z, Chen C. Dexmedetomidine induces bradycardia through vagal-nucleus tractus solitarius glutamatergic circuit.Neuropharmacology · 2026 · 301:111191doi:10.1016/j.neuropharm.2026.111191 · PMID 42749214

    Male C57BL/6J mice, 50 to 200 micrograms per kilogram of dexmedetomidine. Dose-dependent fall in heart rate from 408.4 to 255.5 beats per minute, more phosphorylated ERK in nodose glutamatergic neurons, firing rate up from 0.8 to 3.8 hertz, increased glutamate release in the solitary nucleus, and chemogenetic inactivation of the nodose-to-nucleus glutamatergic circuit reversing the bradycardia. Mice, not patients.

73. 73
    Silvani A, Calandra-Buonaura G, Benarroch EE, Dampney RAL, Cortelli P. Bidirectional interactions between the baroreceptor reflex and arousal: an update.Sleep Med · 2015 · 16(2):210-6doi:10.1016/j.sleep.2014.10.011 · PMID 25616389

    Review placing the solitary nucleus, the C1 adrenergic neurons and the parabrachial nucleus at the junction between blood pressure control and waking up. Mechanistic claims are drawn from animal models; the human data are correlational.

74. 74
    Tian Y, Li N, Zhang Y, Wang HJ. [Research progress of nucleus tractus solitarius involved in central regulation of hypertension].Sheng Li Xue Bao · 2025 · 77(1):85-94PMID 40065617

    Chinese-language review with English abstract; the PubMed record carries no DOI. Summarizes the nucleus as the first receiving station for cardiovascular afferents and its role in baroreflex, sympathetic activity, the renin-angiotensin-aldosterone system and oxidative stress in hypertension. A review, and it states the unresolved questions rather than settling them.

75. 75
    Gama de Barcellos Filho P, Dantzler HA, Brumfield J, Wright HR, Cummings KJ, Hasser EM, Kline DD. Corticotropin-Releasing Hormone Differentially Modulates GABA and Non-GABA Nucleus Tractus Solitarii Neurons.J Neurosci · 2026 · 46(41)doi:10.1523/JNEUROSCI.0525-26.2026 · PMID 42778371

    Patch-clamp recording and calcium imaging in Gad1-EGFP mice of both sexes. Corticotropin-releasing hormone reduced solitary-tract-evoked excitatory currents in GABA neurons and increased them in non-GABA neurons; the receptor was more abundant in medial GABA neurons; blocking the receptor alone increased GABA-neuron currents, indicating a standing restraint. Mice, brain slices.

76. 76
    Bush NE, Oliveira LM, Glovak ZT, Ramirez JM. An inspiration-off attractor supports the robust and flexible control of breathing.Sci Adv · 2026 · 12(42):eaed2474doi:10.1126/sciadv.aed2474 · PMID 42873882

    High-density extracellular recording along the ventral respiratory column with optogenetic and physiological manipulation in living mice. Stimulating the GABAergic afferents of the solitary nucleus advanced or delayed breathing, quantitatively distinct from inhibition local to the column. The attractor account comes from generative dynamical models fitted to the recordings, which is a different kind of claim from the recordings themselves.

77. 77
    Fukushi I, Ozaki I, Pokorski M, Okada Y. Astrocytes at the interface of peri-ictal breathing dysfunction and SUDEP.Front Netw Physiol · 2026 · 6:1955271doi:10.3389/fnetp.2026.1955271 · PMID 42889471

    Review arguing that seizure activity or spreading depolarization reaching the brainstem respiratory centers, the pre-Bötzinger complex and the solitary nucleus among them, is a route to fatal central apnea in sudden unexpected death in epilepsy, with astrocytes implicated in the mechanism. A review of animal and neuropathological work.

78. 78
    Fuse S, Sugiyama Y, Dhingra RR, Hirano S, Dutschmann M, Okada Y, Oku Y. Spatio-temporal segregation between sensory relay and swallowing pre-motor population activities by optical imaging in the rat nucleus of the solitary tract.Pflugers Arch · 2025 · 477(5):719-727doi:10.1007/s00424-025-03065-9 · PMID 39862246

    Calcium imaging in eight perfused rat brainstems separated two things in time and space: a sensory relay confined to the stimulated side at 114.3 plus or minus 94.4 milliseconds, and a bilateral, distributed swallowing signal at 200 plus or minus 145.2 milliseconds. The authors conclude the first-order premotor neurons sit outside the dorsal swallowing group.

79. 79
    Chen J, Zhao F, Hong J, Zhang J, Ye Q, Hu J, Dai Y, Shan Y, Li C, Wen H. HF-rTMS improves swallowing function in rats with poststroke dysphagia by increasing nucleus tractus solitarius excitability through the NMDAR1-Npas4-Nav1.1 pathway.Exp Neurol · 2025 · 398:115606doi:10.1016/j.expneurol.2025.115606 · PMID 41448525

    Ten-hertz magnetic stimulation over the cortex improved swallowing in a rat stroke model, and the benefit disappeared when solitary nucleus excitability or the Npas4 step was blocked. A mechanism in rats for a treatment already used in patients; it is not itself a clinical trial.

80. 80
    Shosha E, Madani J. Recognizing and Differentiating Area Postrema Syndrome Across NMOSD, MOGAD, and GFAP Astrocytopathy.Int J Mol Sci · 2026 · 27(19):8661doi:10.3390/ijms27198661 · PMID 42883907

    Review that flags how often this syndrome is sent to gastroenterology first, delaying treatment by weeks to months. It also reports a single-center study presented at the 2025 American Neurological Association meeting and posted as a preprint, proposing that GLP-1 and its receptor track with the syndrome; the review's own authors call it hypothesis-generating, not replicated and not fully peer reviewed, and that caution is repeated here.

81. 81
    Wang Z, Li X, Li Y, Sun X, Wang Y, Lu T, Zhao D, Ma X, Sun H. The insular cortex-nucleus tractus solitarius glutamatergic pathway involved in acute stress-induced gastric mucosal damage in rats.Neurobiol Stress · 2025 · 36:100723doi:10.1016/j.ynstr.2025.100723 · PMID 40242326

    Silencing the pathway running from the insular cortex down to the solitary nucleus reduced stress-induced stomach injury in rats. Evidence that the traffic runs downward as well as upward; a rat model of restraint and water immersion, not human stress.

82. 82
    Wang Y, Dong F, Li M, Wang Z, Li N, Ma X, Sun H. GABAergic Interneurons in the Insular Cortex Alleviate Acute Stress-Induced Gastric Injury by Suppressing the IC-NTS Glutamatergic Pathway.Biomed J · 2026 · 101055doi:10.1016/j.bj.2026.101055 · PMID 42892498

    Chemogenetic work in mice under restraint water-immersion stress. Inhibiting insular somatostatin neurons projecting specifically to the solitary nucleus did not alter gastric mucosal injury; inhibiting somatostatin and glutamatergic insular neurons together reduced injury and raised the tight-junction proteins occludin and the claudins along with a marker of cell proliferation. A mouse model, and a correction to the simple projection account.

83. 83
    Wang L, Zeng Z, Zhou Y, Gu C, Zhang Y, Yang L, et al. Late-pregnancy oxytocin alleviates neuroinflammation and hyperalgesia in chronic migraine model via PVN to NTS oxytocinergic projections and ECM-receptor interaction signaling.Brain Behav Immun · 2026 · 138:106919doi:10.1016/j.bbi.2026.106919 · PMID 42472609

    A pregnant rat model of chronic migraine. Oxytocin rose in the paraventricular hypothalamus and the solitary nucleus in late pregnancy, viral tracing confirmed a direct oxytocin projection between them, and intranasal oxytocin or viral overexpression at either site lowered calcitonin gene-related peptide, neuronal activation and inflammatory cytokines in the nucleus and reduced pain behavior. Rats; a model of why migraine often eases late in pregnancy, not a clinical result.

84. 84
    Sakamoto H, Akiyoshi R, Tochigi K, Nishijima Y, Tanaka Y. Time-dependent and side-specific c-Fos responses of the nucleus tractus solitarius and hypoglossal nucleus after unilateral nerve transection in mice.Neurosci Res · 2026 · 231:105124doi:10.1016/j.neures.2026.105124 · PMID 42763083

    Cutting the superior laryngeal nerve on one side in mice produced a c-Fos response in the solitary nucleus on both sides, while the motor nucleus responded on one side and then the other. Shows that losing input on one side is not a one-sided event in this nucleus. Mice, three time points, activity marker only.

85. 85
    Yakunina N, Kim SS, Nam EC. Optimization of Transcutaneous Vagus Nerve Stimulation Using Functional MRI.Neuromodulation · 2017 · 20(3):290-300doi:10.1111/ner.12541 · PMID 27898202

    Thirty-seven healthy volunteers, four ear sites, 500-microsecond pulses at 25 hertz. Only stimulation at the cymba conchae activated both the solitary nucleus and the locus coeruleus significantly more than the earlobe sham. Brainstem functional MRI at these scales is technically hard, and the result is a group average, not a demonstration in any one person.

86. 86
    Li Y, Li T, Yan K, Wu J, Reddy S, Sacca V, et al. Both Transcutaneous Auricular Vagus Nerve Stimulation and Great Auricular Nerve Stimulation Modulate Functional and Structural Connectivity of Brainstem Nuclei in Chronic Low Back Pain.Neuromodulation · 2026 · 29(4):597-606doi:10.1016/j.neurom.2025.08.415 · PMID 41091086

    A randomized trial in 70 patients with chronic low back pain, 51 completing. Pain improved in both arms with no significant difference between vagal and non-vagal ear stimulation, while the imaging changes differed. Registered as NCT03959111. The negative between-group result is the important part: an imaging change in the solitary nucleus is not proof that the vagal route is what helped.

87. 87
    Pan ZY, Wu MT, Lau CI, Yang FY. Spleen-targeted ultrasound restores gut-brain homeostasis in colitis via vagal neuroimmune modulation.Brain Behav Immun · 2026 · 139:107045doi:10.1016/j.bbi.2026.107045 · PMID 42838463

    Focused ultrasound over the spleen in mice with dextran-sulfate colitis reduced gut and brain inflammation, restored tight-junction proteins and improved anxiety-like behavior and cognition; the benefit was abolished by subdiaphragmatic vagotomy, with activation reported along a solitary nucleus to locus coeruleus circuit. Mice; the central mechanism is inferred from markers, not recorded.

88. 88
    Konno D, Sugino S, Shibata TF, Misawa K, Imamura-Kawasawa Y, Suzuki J, Kido K, Nagasaki M, Yamauchi M. Antiemetic effects of baclofen in a shrew model of postoperative nausea and vomiting: Whole-transcriptome analysis in the nucleus of the solitary tract.CNS Neurosci Ther · 2022 · 28(6):922-931doi:10.1111/cns.13823 · PMID 35238164

    Twenty-one female musk shrews. Sequencing the solitary nucleus pointed to suppressed GABA-B receptor signaling, and baclofen, a GABA-B agonist, eliminated vomiting. Note the control result the authors report honestly: the sham and surgery groups both had a median of six emetic episodes, so the surgery itself added nothing, which weakens the model.

89. 89
    Lu Z, Qiao Y, Huang X, Cui D, Liu JYH, Ngan MP, et al. A deep learning-based system for automatic detection of emesis with high accuracy in Suncus murinus.Commun Biol · 2025 · 8(1):209doi:10.1038/s42003-025-07479-0 · PMID 39930110

    An automated video system scored vomiting in the musk shrew with 98.92 percent overall accuracy, validated across eight emetic drugs. Useful as a measure of how emesis research is actually done, and of how much of it rests on one small insectivore.

90. 90
    Lee MJ, Kim IS, Ahn SH, Chung W, Kang YE, Jeon SM, et al. Vagal sensing of altered gut microbiota links to externalizing behaviors.Gut Microbes · 2026 · 18(1):2743919doi:10.1080/19490976.2026.2743919 · PMID 42860590

    Mice lacking Sirtuin 3 had an altered gut microbiome, raised plasma choline, reduced GABA in the solitary nucleus and mood-related behavioral change; subdiaphragmatic vagotomy returned those measures toward wild-type. Mice, and the authors describe the link as correlation sensed through vagal signaling rather than a demonstrated mechanism.