Serotonin (5-HT) receptors

Serotonin was isolated from blood serum in 1948 as a substance that raised vascular tone; the name is literally serum plus tone ^[5]. The receptor story began nine years later, when a pair of pharmacologists working on guinea pig ileum found that serotonin's actions there split cleanly into two, one blocked by dibenzyline and one blocked by morphine, and named them D and M ^[6]. Those two are the distant ancestors of what are now called 5-HT2 and 5-HT3. Everything since has been the same exercise repeated with better tools.

By the late 1980s radioligand binding had multiplied the count enough that the field needed a filing system, and receptors were sorted into families by which drugs bound them ^[4]. Cloning then changed the basis of the classification entirely. Once cDNA sequences arrived, membership was decided by three things together: gene sequence, pharmacology (which ligands bind and how tightly), and signalling (which G protein or channel is used). That is the scheme the International Union of Pharmacology formalised in 1994 ^[3] and updated comprehensively in 2021 ^[2]; the four decade arc from two receptors to fourteen is a good reminder of how much of this stayed invisible until the method arrived ^[7].

One convention in that scheme trips people up. Receptors written in lower case, such as 5-ht1E and 5-ht5A, are ones whose functional role in native tissue is not established to the classification committee's satisfaction. The lower case is not a typo; it is an admission ^[1][2]. Fourteen receptors have been described, and thirteen make a functional protein in humans, because the 5-ht5B gene is interrupted by stop codons in people even though it is expressed in mouse and rat ^[25].

It also helps to know where the transmitter itself is, because it is mostly not in the brain. Roughly 90% of the body's serotonin sits in enterochromaffin cells lining the gut, acting as a paracrine sensory signal that triggers peristaltic and secretory reflexes; platelets carry more, though they do not make it and simply take it up from plasma ^[8][9]. Brain serotonin is manufactured locally from tryptophan by a different enzyme isoform, because 5-HT does not cross the blood brain barrier. That one fact is why swallowing serotonin does nothing centrally, why 5-HTP and L-Tryptophan are used as precursors instead, and why a drug acting on 5-HT receptors is very often acting mostly on the gut.

The practical upshot for anyone reading a supplement label or a drug monograph: serotonergic tells you almost nothing on its own. What matters is which 5-HT receptor is being turned up or down, in which direction, and where in the body that receptor sits.

Here is the whole family tree in one view. Read the coupling column first, because it sets the sign of everything downstream. Gi/o receptors quiet a cell (they lower cAMP and open potassium channels), Gs receptors rev it up (they raise cAMP), Gq/11 receptors trigger phospholipase C and a calcium and inositol-phosphate cascade, and the lone ion channel simply opens a pore ^[1][2].

Three structural notes that do not fit in a cell. The 5-HT3 receptor is pentameric, built from five subunits around a central pore, and belongs to the same superfamily as nicotinic acetylcholine and GABA-A receptors rather than to the serotonin GPCRs at all; it responds in milliseconds where they take seconds ^[10]. The 5-HT1 family is the largest, five members all coupling to Gi/o and all reducing excitability, which is why it supplies both of the system's autoreceptors. And the 5-HT2 family is the one with a safety problem, because its middle member sits on heart valve tissue; that gets its own section below.

Two members are genuinely under-characterised rather than merely obscure. 5-ht1E has no selective ligand and is absent from rat and mouse, which closed the usual animal route to studying it. 5-ht5A stayed among the least understood receptors of the whole set for the same reason; what is known is that it couples to Gi/o, that its highest affinity ligand is LSD, and that mice lacking it show a blunted locomotor response to LSD ^[25].

Three signalling routes cover almost the whole family. Gi/o receptors (5-HT1, 5-HT5) inhibit adenylyl cyclase, lower cAMP and open inwardly rectifying potassium channels, hyperpolarising the cell and making it harder to fire. Gs receptors (5-HT4, 5-HT6, 5-HT7) do the opposite. Gq/11 receptors (5-HT2) do something different in kind: they activate phospholipase C, which splits a membrane lipid into IP3 and diacylglycerol, releasing intracellular calcium and activating protein kinase C. A Gq receptor is therefore not simply an on switch relative to a Gi one; it starts a different cascade with different targets and a different timecourse ^[1][2].

Autoreceptors are what make serotonin pharmacology counterintuitive. Two of the 5-HT1 receptors sit on serotonin neurons themselves. 5-HT1A on the raphe cell bodies slows the neuron's firing rate when serotonin builds up around it; 5-HT1B on the terminals cuts how much is released per action potential. Raise synaptic serotonin with a reuptake inhibitor and both brakes engage at once, which is why an SSRI's first effect on serotonin neuron firing is to reduce it. Only after those autoreceptors desensitise over two to three weeks does the intended rise in transmission arrive ^[19]. Almost every attempt to make antidepressants act faster has been an attempt to get around that brake.

5-HT2C carries a mechanism no other receptor in the human body uses this way. Its pre-messenger RNA is edited at up to five adenosine positions before translation, changing three amino acids in the second intracellular loop. The edited forms differ in how efficiently they couple to G protein and in how much constitutive activity they show, so a single gene yields a spectrum of receptors whose signalling strength is set after transcription and can vary by brain region ^[11]. Constitutive activity means the receptor signals with nothing bound, so at 5-HT2C the difference between a neutral antagonist and an inverse agonist is a real pharmacological difference rather than a semantic one.

Binding kinetics turn out to matter as much as affinity. The structure of 5-HT2B with LSD bound showed the ligand held under a lid formed by extracellular loop 2, giving it an unusually slow off-rate; mutating that lid sped dissociation and changed the signalling profile, including how strongly beta-arrestin was recruited ^[15]. That is the clearest structural account available of why LSD acts for many hours from a very small dose, and it is a concrete example of biased agonism, where two ligands at one receptor favour different downstream arms. Whether that bias is what separates a hallucinogenic 5-HT2A agonist from a non-hallucinogenic one is an open and heavily worked question ^[16].

5-HT1A is the poster child for the mood side of serotonin, and it is doing two different jobs at once. As an autoreceptor it is a brake on the serotonin system; as a postsynaptic receptor in hippocampus and cortex it mediates much of the anxiolytic and antidepressant effect. Partial agonists such as Buspirone, Gepirone and Tandospirone are built on that second role, and the 5-HT1A activity of CBD and of the non-hallucinogenic psychedelic analogue Tabernanthalog is a large part of why either draws interest.

5-HT2A is where the vivid effects live, and the evidence is unusually direct. In 1984 a simple correlation made the case: across a set of hallucinogens, affinity for the 5-HT2 site tracked human potency ^[12]. In 1998 it was tested the hard way in people. The psychotomimetic effects of Psilocybin in healthy volunteers were blocked dose-dependently by the 5-HT2A antagonist ketanserin and by Risperidone, and were increased by the dopamine antagonist Haloperidol ^[13]. Blocking dopamine made it worse; blocking 5-HT2A made it stop. That one experiment settles the receptor question more cleanly than any binding table ^[14]. See psychedelics and 5-HT2A for the rest.

5-HT3, the ion channel, is the reason a whole antiemetic class exists. It sits in the area postrema and on vagal afferents, both of which report to the vomiting reflex. Blocking it with Ondansetron or Tropisetron blunts that reflex, which is why these are staples during chemotherapy and after surgery ^[1][9]. The same block slows the gut, which is why constipation is their signature side effect; it is the intended action landing in an unintended place.

5-HT4 is the gut's accelerator, and possibly more. Agonists speed transit and are used for constipation and functional gut disorders ^[9]; Prucalopride is the current selective example, while Cisapride and Tegaserod were withdrawn or restricted over cardiac problems unrelated to 5-HT4 itself. The interesting claim is central: in rats, three days of a 5-HT4 agonist produced the markers of antidepressant action that normally take two to three weeks, including desensitised 5-HT1A autoreceptors, where three days of citalopram produced none ^[27]. That is a rodent result, and it has not yet become a drug.

5-HT6 and 5-HT7 are the newer targets, with very different records. 5-HT6 antagonism looked like a clean cognition mechanism for two decades and then failed decisively: three randomised trials of Idalopirdine added to a cholinesterase inhibitor in 2,525 patients with Alzheimer's disease found no cognitive benefit at any dose ^[28]. 5-HT7 antagonism has a better preclinical case in depression and circadian regulation ^[26] and is one component of Vortioxetine, which blocks the serotonin transporter while also acting as a 5-HT1A agonist, a 5-HT1B partial agonist and a 5-HT3, 5-HT1D and 5-HT7 antagonist ^[29]. That combination is the explanation offered for its cognitive effects, and it illustrates the attribution problem such drugs create: with six actions and one clinical result, no trial design separates them.

Because these receptors sit in different tissues doing different jobs, subtype selectivity is the whole game. A drug that hits only the receptor you want tends to be clean; a drug that sprays across the family brings side effects, some of them serious ^[1].

The sharpest cautionary tale is 5-HT2B. This receptor sits on the interstitial cells of heart valves, and chronic agonism drives those cells to proliferate and deposit matrix, stiffening the valve; the condition is valvulopathy, and it is not reversible. It surfaced clinically in 1997 as an unexplained cluster of valve disease in women taking fenfluramine with phentermine for weight loss ^[22]. The mechanism was then established by a screen designed to answer exactly one question: drugs known to cause valve disease were compared with drugs known not to, across eleven cloned serotonin receptors. Every positive control, including norfenfluramine (the active metabolite of Fenfluramine), ergotamine and methylergonovine, turned out to be a preferential high-affinity agonist at 5-HT2B, and the negative controls were not ^[21]. The counter-example completes the case: Lisuride is an ergoline dopamine agonist chemically similar to the ergot drugs that cause valve disease, but it is a potent 5-HT2B antagonist, and after an estimated 360,000 patient years no case of valvulopathy has been reported with it ^[23].

The rule that came out of that work is now standard: screen every serotonergic drug and its active metabolites for 5-HT2B agonism before it goes near people ^[21]. The general lesson carries beyond serotonin. The harm came from a receptor nobody was aiming at, in an organ nobody was thinking about, from the metabolite rather than the parent drug, and it took years of exposure to appear.

Serotonin syndrome is the acute counterpart and comes from the opposite problem: too much agonism at once, mostly at 5-HT1A and 5-HT2A, usually because two serotonergic drugs were combined. It presents as a triad of mental status change, autonomic instability and neuromuscular hyperactivity, and the neuromuscular signs, particularly clonus that is worse in the legs than the arms, are what distinguish it from the conditions it mimics. Onset is fast, typically within hours of a dose change, which is itself diagnostic. Treatment is withdrawal of the offending drugs plus supportive care, with Cyproheptadine used as a 5-HT2A antagonist ^[24]. The combinations that cause it are predictable from the pharmacology, which makes it one of the more avoidable serious drug reactions.

Selectivity also explains the ordinary side effects nobody dies of. Sexual dysfunction on SSRIs is attributed largely to 5-HT2A and 5-HT2C stimulation; weight gain on several antipsychotics tracks 5-HT2C blockade alongside histamine H1; constipation on setrons is 5-HT3 in the gut; the nausea that starts an SSRI is 5-HT3 as well, which is why it fades as those receptors adapt. In each case the unwanted effect is the wanted mechanism arriving somewhere else.

Serotonin receptors adapt to sustained stimulation, and two very different adaptations dominate the clinical picture: rapid tolerance at 5-HT2A, and slow desensitisation of the 5-HT1A autoreceptor. See tolerance and dependence for the general framework.

5-HT2A downregulates fast, and that is why psychedelic tolerance is so steep. In rats given LSD daily at a dose that reliably produces behavioural tolerance, cortical 5-HT2 binding fell measurably, while binding at 5-HT1A, 5-HT1B, beta-adrenergic, alpha-adrenergic and D2 sites, and at the serotonin uptake site, did not move at all. The fall appeared within three days rather than after a single dose, was still present 48 hours after the last dose and gone by 96 hours; psilocybin produced the same effect, the non-hallucinogenic analogue bromo-LSD did not, and mescaline did not at the doses used ^[17]. The receptor loss tracks behavioural tolerance closely enough that the two are usually treated as one phenomenon, though the authors noted that receptor number alone does not account for every cross-tolerance result.

And then a genuine anomaly. 5-HT2A is downregulated by chronic treatment with its antagonists as well as its agonists, which is backwards for a receptor and does not happen to most GPCRs. The leading explanation is that antagonists at this receptor drive internalisation rather than preventing it, so both directions of drug remove receptors from the surface ^[18]. It is worth knowing because it undercuts the intuitive rule that blocking a receptor makes a system more sensitive to it; at 5-HT2A that rule does not reliably hold.

The 5-HT1A autoreceptor explains the antidepressant delay. An SSRI raises serotonin in the raphe immediately, the somatodendritic 5-HT1A autoreceptors sense it and slow the neurons, and net transmission at the target regions barely changes. Over two to three weeks those autoreceptors desensitise, firing recovers, and only then does serotonin actually rise where it was wanted ^[19]. The hypothesis was tested directly by adding Pindolol, a beta-blocker that also blocks 5-HT1A, to an SSRI from day one; in the original open study, improvement arrived in days rather than weeks ^[20]. Later controlled trials were mixed, and pindolol augmentation never became standard practice, but the experiment remains the clearest demonstration that the delay is an autoreceptor problem rather than a slow-building structural one.

Discontinuation is a separate adaptation. Stopping an SSRI abruptly, particularly a short half-life one, produces dizziness, electric-shock sensations, irritability and flu-like symptoms over days to weeks. It reflects receptor and transporter systems that adjusted to the drug and now face its absence; it is neither addiction nor relapse, though it is regularly mistaken for both.

One adaptation that never arrives. The 5-HT2B receptor does not desensitise usefully to chronic agonism; the valve keeps responding, which is precisely why the damage accumulates rather than plateauing ^[21]. Tolerance is not automatically protective, and assuming it is has been a costly error in this family.

Two receptors remain effectively unstudied. 5-ht1E has no selective ligand and does not exist in rats or mice, which closed off the standard experimental route entirely. 5-ht5A has been characterised biochemically but its function in an intact animal is still largely inferred from knockout behaviour ^[25]. Both keep their lower-case names for that reason, and any source that assigns them confident roles is going beyond the evidence ^[1][2].

Whether a 5-HT2A agonist has to be hallucinogenic to be an antidepressant. This is the most active question in the field. Non-hallucinogenic 5-HT2A agonists have been designed and show antidepressant-like and plasticity-promoting effects in rodents, and the working hypothesis is that signalling bias, or the location of the receptor inside the cell rather than on its surface, separates the two effects ^[16]. No human trial has yet shown that a non-hallucinogenic analogue produces the clinical benefit. Until one does, the separation is a design goal rather than a finding.

Whether low serotonin causes depression. This deserves stating plainly because the receptor biology is often used to imply it. An umbrella review of the main strands of evidence, covering serotonin metabolites, receptor and transporter imaging, tryptophan depletion studies and genetic association work, found no consistent support for the idea that depression is caused by low serotonin or reduced serotonin activity ^[30]. That is a statement about a causal theory, not about whether serotonergic drugs work; they demonstrably do for some people, and their receptor pharmacology is well described. The two claims have been conflated for decades and are separate.

What the multimodal drugs owe to each of their targets. Vortioxetine has six identified serotonergic actions plus transporter blockade ^[29], and no clinical trial design apportions its effect among them. The same holds for most atypical antipsychotics. Mechanistic explanations for these drugs are reconstructions after the fact and should be read that way.

How much of the gut story is central at all. With roughly 90% of the body's serotonin in the gut and a large share of 5-HT receptors in the enteric nervous system ^[8][9], a serotonergic drug taken by mouth acts on both systems, and separating the two experimentally in humans is difficult. The gut-to-brain contribution to mood effects is plausible, actively studied, and not established.

And a framing caution. Serotonin is routinely called the happiness chemical. Nothing on this page supports that; the same molecule constricts blood vessels, triggers vomiting, moves the bowel, clots blood and sets circadian phase, depending entirely on which of fourteen receptors receives it. For the transmitter-level view see the serotonergic system, and for the same one-to-many design in another family see dopamine receptors.