04. Serotonin: The Thermostat

A blade slips in a kitchen and opens a shallow cut across a fingertip. Within seconds platelets are piling into the gap and gluing themselves together, and every one of them carries serotonin it did not make. Platelets cannot build the molecule at all: they scavenged it from blood plasma using SERT, the serotonin transporter, and packed it into two to eight dense granules apiece using a vesicular transporter called VMAT-1. At the wound they degranulate, and the spilled serotonin lands back on their own surface receptors, 5-HT2A and 5-HT3. Aggregation amplifies. The vessel constricts. The bleeding slows, in a closed chemical loop with no nerve involved anywhere in it. The most famous brain chemical in the world has just worked a shift as a logistics contractor.

Between ninety and ninety-five percent of the serotonin in a human body — a widely repeated range rather than one traceable measurement — is made and stored in the gut, by enterochromaffin cells, fewer than one percent of the cells lining it. In 2017 Nicholas Bellono, James Bayrer, David Julius and colleagues showed in Cell what those cells are: chemosensors wired to sensory nerves. They register stretch, chemical irritants and bacterial byproducts, then release serotonin onto sensory nerve terminals and onto the enteric neurons Michael Gershon popularised in 1998 as the "second brain." They are not built from neural crest, the tissue supplying most neuroendocrine cell types; developmental work places them in gut endoderm. Moving a meal and closing a wound: two jobs, no synapse in either.

Vol. I dated part of this molecule's past. It followed serotonin's synthesizing enzyme, tryptophan hydroxylase, deep past animals into a prokaryotic pterin-dependent hydroxylase family; found the whole amine kit already running in a ciliate with no nervous system; and watched a lobster rise on an injection of serotonin, then watched the argument that followed. What it never dated was the receiving end. A molecule with no receptor is a chemical, not a signal, and the receptors are where this chapter's commits live.

There are fourteen of them in a human, drawn from about eighteen genes and sorted into seven families, 5-HT1 through 5-HT7. Thirteen subtypes across six families are G-protein-coupled receptors — proteins that thread the membrane seven times and pass the message inward through a G protein, slowly. One family is not. 5-HT3 is a ligand-gated ion channel from the Cys-loop superfamily: serotonin binds, a pore opens, cations pour through in milliseconds. An unrelated structural class, bolted onto the same ligand. The first 5-HT7-type receptor ever cloned came out of a fruit fly, and the mammalian version was found by homology to the fly's gene, not the reverse.

Nematodes built their own. C. elegans has MOD-1, described by Rajesh Ranganathan, Stephen Cannon and Robert Horvitz in Nature in 2000: a serotonin-gated chloride channel, about thirty percent identical to GABA and glycine receptors, ionotropic like 5-HT3 but passing anions instead of cations, and absent from vertebrates. Vol. I met the same shape in an octopus's fast dopamine receptor. The molecule is inherited once; a fast way of hearing it is invented more than once, on whatever scaffold a lineage has to hand.

One receptor carries most of the molecule's fame. LSD and psilocybin are agonists at 5-HT2A, and that single receptor anchors their effects on human perception — the pharmacology David Nichols set out in Pharmacological Reviews in 2016. But fame is not function, and the popular name for this molecule fails on the evidence. Deakin and Graeff's account has serotonin processing aversive events, not rewarding ones; and in 2009 Molly Crockett, Luke Clark and Trevor Robbins showed in The Journal of Neuroscience that acutely depleting tryptophan, serotonin's raw material, abolishes punishment-induced slowing of behaviour specifically, leaving reward-driven responding intact. A molecule whose most reproducible instruction is wait is a poor candidate for a happiness chemical. It is closer to a thermostat: it does not choose the temperature, it holds one.

So when was the thermostat installed? For the receptor family there is one number in the literature, and it is thirty years old. In 1994 Stephen Peroutka and T. A. Howell found under twenty-five percent homology between the three major classes of 5-HT G-protein-coupled receptor and ran that divergence against a fixed clock of roughly one percent change per ten million years. Out came an age for the primordial 5-HT receptor of more than 700 to 800 million years. Set that beside this library's floor for the first animals — crown Metazoa at 649.8 to 833.5 million years, from dos Reis and colleagues' Bayesian analysis in Current Biology in 2015 — and the receptor date presses against the old end of that window while sitting, at its young end, older than animals are permitted to be. Either that is a real result about a pre-animal ancestor or it is an artefact: a fixed rate with no fossil calibration and no relaxed-rate model is the instrument Vol. I warned about, and nobody has redone this family the modern way. The absence is the state of the art, not a footnote.

What survives without the clock is firmer: 5-HT1-, 5-HT2- and 5-HT7-type receptors cloned directly out of invertebrates put the family in place before the protostome–deuterostome split — older than 600 million years conventionally, 642 to 761 by clock, 558 to 575 when the rocks are allowed to calibrate.

The Argument at the Bilaterian Line

The boundary below that one is live. In June 2023 Matthew Goulty, Roberto Feuda and colleagues argued in Nature Communications that the whole monoaminergic system — synthesis enzymes, vesicular transporters, receptors — is a bilaterian innovation, absent from cnidarians, placozoans, ctenophores and sponges. In 2026 Luis Yañez-Guerra, Gáspár Jékely and colleagues contested that in Current Biology. They deorphanized five G-protein-coupled receptors from the placozoan Trichoplax adhaerens — deorphanizing being the work of finding a catalogued receptor's natural ligand — and four of the five fall inside the bilaterian melatonin-receptor clade. All five answer to tyramine, tryptamine or phenethylamine — trace amines a placozoan builds in one enzymatic step. Serotonin bound too, but weakly — two to three orders of magnitude worse — and placozoans lack tryptophan hydroxylase and AANAT outright, so they cannot make serotonin or melatonin at all. The honest reading is narrower than either headline. The receptor scaffold predates Bilateria; serotonin-specific signalling still looks bilaterian or later — the order Vol. I found in a sponge carrying a postsynaptic scaffold with no receptors to dock onto it. Jékely, whom Vol. I met establishing that noradrenaline-type signalling coexisted with octopamine in the last bilaterian ancestor, is making the same case again: the parts are older than the job.

Melatonin takes one line: serotonin's night-shift derivative, acetylated by AANAT — an enzyme whose activity swings roughly a hundredfold between day and night — then methylated. Vol. I fixed its origin at about 3.0 to 2.5 billion years ago, an order-of-magnitude geological inference tied to the Great Oxidation Event rather than a clock. The derivative has a date. The apparatus that reads the parent does not.

Which leaves the trap. Fluoxetine was found on 24 July 1972 by David Wong, Ray Fuller, Bryan Molloy and Klaus Schmiegel at Eli Lilly, as compound Lilly 110140, and approved as Prozac on 29 December 1987. Its mechanism is not in dispute: it blocks SERT — the transporter the platelets in the opening scene use — so less serotonin is retrieved from the synapse. That transporter's ancestry is a fold, not a function: its architecture was solved through a bacterial leucine carrier that moves no amine at all. Whether blocking SERT treats depression by correcting a serotonin deficit is a separate claim, and contested. Andrea Cipriani and colleagues found in The Lancet in 2018 that all twenty-one antidepressants studied beat placebo, at a modest pooled effect whose clinical meaning is argued over. Joanna Moncrieff and colleagues found in Molecular Psychiatry in 2022 — across serotonin and metabolite levels, receptor and transporter imaging, depletion and gene-association studies — no consistent evidence that depression is caused by lowered serotonin; the review drew a formal reply in the same journal in 2023. Efficacy and mechanism are separable claims, neither settled here, and mood biology has its own volume in this library. This one dates commits.

The dates worth trusting here are about parts, not jobs. The synthesis enzyme has a lineage and no clock; the receptor family has a clock and no calibration; the transporter has a fold and no dated cargo; and the two jobs this molecule most visibly does in a human body have no phylogenetic date at all. A thin log for the most talked-about molecule in the human head — and the thinness is the finding. The next molecule in this book does not ask an animal to hold steady; it asks it to go and get.

Commit Ledger

Overturned if: a living non-bilaterian runs serotonin-specific signalling functionally — own synthesis, own selective receptor, a behaviour that changes when either is blocked — rather than a promiscuous trace-amine receptor serotonin happens to fit.

Sources

• 1994 · Peroutka & Howell, "The molecular evolution of G protein-coupled receptors: Focus on 5-hydroxytryptamine receptors" — Neuropharmacology
• 2000 · Ranganathan, Cannon & Horvitz, "MOD-1 is a serotonin-gated chloride channel that modulates locomotory behaviour in C. elegans" — Nature
• 2005 · Yamashita et al., "Crystal structure of a bacterial homologue of Na+/Cl−-dependent neurotransmitter transporters" — Nature
• 2005 · Wong et al., "The Discovery of Fluoxetine Hydrochloride (Prozac)" — Nature Reviews Drug Discovery
• 2009 · Crockett, Clark & Robbins, "Reconciling the Role of Serotonin in Behavioral Inhibition and Aversion" — The Journal of Neuroscience
• 2015 · dos Reis et al., "Uncertainty in the Timing of Origin of Animals" — Current Biology
• 2016 · Nichols, "Psychedelics" — Pharmacological Reviews
• 2016 · Qi et al., "Serotonin modulates insect hemocyte phagocytosis via two different serotonin receptors" — eLife
• 2017 · Bellono, Bayrer, Julius et al., "Enterochromaffin Cells Are Gut Chemosensors that Couple to Sensory Neural Pathways" — Cell
• 2018 · Cipriani et al., "Comparative efficacy and acceptability of 21 antidepressant drugs" — The Lancet
• 2020 · Soslau, "Serotonin: a platelet hormone modulating cardiovascular disease" — Journal of Thrombosis and Thrombolysis
• 2022 · Moncrieff, Cooper, Stockmann et al., "The serotonin theory of depression: a systematic umbrella review of the evidence" — Molecular Psychiatry
• 2023 · Bartova, Lanzenberger, Rujescu & Kasper, "Reply to: 'The serotonin theory of depression'" — Molecular Psychiatry
• 2023 · Goulty, Botton-Amiot, Rosato, Sprecher & Feuda, "The monoaminergic system is a bilaterian innovation" — Nature Communications
• 2026 · Yañez-Guerra, Jékely et al., "Prebilaterian origin of monoaminergic signaling" — Current Biology