08. Oxytocin and Vasopressin: The Bonding Fork
A labor ward, late in a long night. A bag of clear fluid hangs beside the bed, and a pump counts it down a line taped to the back of a woman's hand. In it is a chain of nine amino acids called oxytocin, built in a factory rather than a brain and sold as Pitocin. It docks onto OXTR, the oxytocin receptor studding every smooth-muscle cell of the uterine wall, which answers by releasing calcium inside the cell — and calcium is what makes muscle shorten. Half a century on, they are among the most widely administered molecules a human ever meets.
The body runs the same circuit as a loop feeding on its own output. Pressure from the descending fetus against the cervix signals the hypothalamus, which releases oxytocin from the posterior pituitary, which contracts the uterus, which drives the fetus harder against the cervix. Physiologists call it the Ferguson reflex — one of the few positive-feedback loops in a body whose control circuits almost all exist to hold a variable steady. A second loop follows birth: suckling triggers oxytocin, which squeezes the myoepithelial cells around the milk glands, and milk lets down half a minute later. Both stop only when what started them is gone.
The bag exists because of a specific chain of results. In 1906, in a paper mostly about ergot, Henry Dale reported that posterior-pituitary extract contracted a pregnant cat's uterus, and named the effect oxytocic, from the Greek for swift birth. The crude extract reached obstetric use within five years; in 1928 Kamm and colleagues at Parke, Davis split it into two active substances of unknown structure. Then in 1953 Vincent du Vigneaud, at Cornell, determined the nine-residue sequence, built the molecule from scratch, and showed it behaved exactly like the natural one. His 1955 Nobel Prize cites "the first synthesis of a polypeptide hormone." A hormone had been reduced to a sequence and a shape.
Kamm's two substances are this chapter's molecules: oxytocin and vasopressin, nine residues each, differing at two positions. In the human genome they sit almost on top of each other — OXT and AVP, roughly eleven kilobases apart on chromosome 20, transcribed toward each other from opposite strands, tail-to-tail. It is an irresistible picture: the duplication still legible, two files side by side because one was copied from the other.
The picture is backwards. Gwee, Tay, Brenner and Venkatesh sequenced the nonapeptide loci in elephant shark and Japanese lamprey and compared them with the arrangements known elsewhere. In elephant shark, coelacanth, Xenopus, chicken and opossum the two genes are linked tail-to-head, and that, on their argument, is ancestral. The tail-to-tail layout in humans and rodents is a later, lineage-specific inversion of the oxytocin gene alone. What gets pointed to as the fork made visible is a mammalian rearrangement of a fork older than the picture showing it.
The same paper settles the larger question with its best outgroups. Japanese lamprey, a living jawless vertebrate, and amphioxus, a cephalochordate more basal still, each carry exactly one nonapeptide gene, and both genomes were searched for a second. In each, that gene is vasopressin-family, not oxytocin-family. Amphioxus's copy, named [Ile4]vasotocin, is a 167-residue prohormone with the vertebrate four-part architecture — signal peptide, nonapeptide, neurophysin, copeptin — differing from vertebrate vasotocin at one residue. Elephant shark, the earliest-diverging jawed lineage sampled, already carries both. Vasopressin is the original file; oxytocin is the copy.
That fixes where the fork happened without dating it. Gwee's team published no clock estimate, and the honest bracket comes from two splits Vol. I already dated: the fork is younger than the jawless/jawed divergence around 560 million years ago and older than the cartilaginous/bony fish split around 450, since elephant shark already carries both. This is not the event Vol. I dated. There the six-gene family of receptors was traced to two rounds of whole-genome duplication in a 600-to-450-million-year window — Ocampo Daza, Bergqvist and Larhammar arguing for two rounds from fourfold synteny, Theofanopoulou and colleagues reading lamprey sequences as one round plus smaller duplications, the dispute unresolved. It is also a different mechanism: a whole-genome duplication copies the library at once; the ligand fork copies one gene beside itself. The peptides forked locally. The receptors that read them forked globally.
The surviving receptors keep fingerprints of that history. Paré, Bortolini and colleagues found OXTR, AVPR1A and AVPR1B all keeping an ancestral two-exon structure while AVPR2 alone gained a third; and in placental mammals OXTR sits under comparatively strong purifying selection while the three vasopressin-family receptors show higher substitution rates. What they read is not really a nine-residue string, either. Six of the nine close into a ring fastened by a disulfide bond, with a three-residue tail off it, and Waltenspühl, Schöppe, Ehrenmann, Kummer and Plückthun's 2020 crystal structure of the human oxytocin receptor confirmed that the ring conformation is what recognition reads. One copy was held still; three were left to drift.
Vasopressin's own day job is unglamorous and probably older than anything social. It binds AVPR2 on the kidney's collecting-duct cells, inserting aquaporin-2 water channels into the membrane so water is pulled back out of the urine; break either end and the result is diabetes insipidus — liters of dilute urine, unrelenting thirst, a night of waking to drink. The evolutionary reading agrees, since the single ancestral gene in both outgroups is the vasopressin-type one, and where that peptide has been characterized in an invertebrate deuterostome, a sea cucumber, the job is feeding and osmoregulation. That is a working inference, not a measured date: nothing dates when social split from metabolic, only when the genes did.
Vol. I walked the rest of that menagerie — isotocin, mesotocin, vasotocin, nematocin, annetocin, conopressin — from a Panamanian rain pool to a worm burrow, and this chapter does not walk it again. What it left standing is the mammalian half: the pair bond, and the popular molecule built on it.
The fork and the myth
The prairie vole made oxytocin famous. In 1993 Winslow, Hastings, Carter, Harbaugh and Insel reported in Nature that central vasopressin was required for pair bonding in these monogamous rodents, and over thirty years the literature became a textbook case: blockers abolished bonds, agonists induced them, receptor maps separated monogamous voles from promiscuous relatives. Necessity and sufficiency, from converging pharmacology. Then the tools changed. Devanand Manoli's and Nirao Shah's laboratories deleted the receptor gene with CRISPR and published in Neuron in 2023 under a title that concedes nothing — the oxytocin receptor is not required for social attachment in prairie voles. The animals bonded anyway, tested every way the team could devise, and receptor-null mothers delivered and raised pups through weaning.
Two readings stay open. Vasopressin signaling through V1aR may compensate when the receptor is absent from conception, which a blocker dripped onto an adult brain can never test; or the necessity claim was wrong from the start. A 2024 follow-up from the same group, reporting that the receptor also shapes promiscuity and developmental timing, is a preprint awaiting peer review. Both bodies of work were done carefully, and they disagree. This is what an overturned commit looks like while it is still being overturned.
The human literature is weaker, and it is the source of the label this book will not use. Kosfeld, Heinrichs, Zak, Fischbacher and Fehr reported in Nature in 2005 that a nasal spray of oxytocin raised how much money strangers transferred in an economic trust game — the single empirical root of "love hormone" and "cuddle chemical." In 2015 Nave, Camerer and McCullough reviewed six replication attempts running from an increase to no effect to a decrease in trust, and flagged serious variability in the assays themselves; a registered replication in Nature Human Behaviour in 2020 again found no reliable effect. In the best-powered clinical test yet run — Sikich and colleagues in the New England Journal of Medicine in 2021, 277 children and adolescents with autism, twenty-four weeks — intranasal oxytocin was null on its primary endpoint. Oxytocin also raises in-group favoritism, envy and out-group bias in some paradigms, which is not the behavior of a love molecule. The nasal route is no cleaner: roughly 0.001 to 0.005 percent of a dose reaches cerebrospinal fluid.
The ledger below is lopsided, and the shape is the finding. The gene history is sturdiest — two outgroup genomes, a synteny argument, a fork bracketed between two dated splits — while the birth and milk physiology, nearly as solid, carries no deep-time date at all. The behavioral rows are where the evidence thins, and the human row is the thinnest so far: a molecule whose deep past is better established than its present reputation. Peptides that hold one animal to another are not the only short chains a vertebrate inherited — the next family lets an animal ignore its own damage.
Commit Ledger
| Function | First commit | Grade | What would move the date |
|---|---|---|---|
| Ancestral nonapeptide — one vasotocin-type gene | 600–450 Mya; bracketed between two independently dated splits, not clock-dated [Vol. I anchor] | B | Resolution of the one-versus-two whole-genome-duplication dispute |
| OXT/AVP ligand fork — local tandem duplication | ~560–450 Mya; phylogenetic bracketing (jawless/jawed split as floor, elephant shark as ceiling); no clock estimate published | B | A nonapeptide locus from a jawed lineage more basal than elephant shark |
| Receptor family — OXTR plus three AVPRs | 600–450 Mya; two whole-genome duplications inferred from fourfold synteny [Vol. I anchor] | B, mechanism contested | Cyclostome resequencing settling the duplication count |
| Water balance / osmoregulation | No independent date; inferred from ancestral gene content in lamprey and amphioxus | D | A functional assay of the lamprey's own vasotocin gene, never performed |
| Birth contraction (Ferguson reflex) | Mechanism established from 1906; no deep-time date | B for mechanism | Whether the same feedback loop runs in marsupials and monotremes |
| Milk ejection (let-down) | Mechanism from twentieth-century physiology; no deep-time date | B for mechanism | Comparative lactation-hormone data outside placental mammals |
| Pair bonding (voles) | Pharmacology 1993–2005 established necessity and sufficiency; a 2023 knockout contradicts it in the same species | C — two rigorous methods disagree | A V1aR-pathway knockout crossed onto the receptor-null background |
| Human trust / social cognition | Founding result 2005; critical review 2015; registered replication 2020; no robust effect | D | A large pre-registered trial with confirmed central delivery |
Overturned if: a genuine oxytocin-family nonapeptide gene turns up in a lamprey or hagfish genome, pushing the ligand fork below the jawless/jawed split and dissolving the argument that the ancestral molecule was vasopressin-like.
Sources
- 1906 · Dale, "On some physiological actions of ergot" — The Journal of Physiology
- 1955 · Nobel Prize in Chemistry, "Vincent du Vigneaud — Facts" — The Nobel Foundation
- 2009 · Gwee, Tay, Brenner, Venkatesh, "Characterization of the neurohypophysial hormone gene loci in elephant shark and the Japanese lamprey: origin of the vertebrate neurohypophysial hormone genes" — BMC Evolutionary Biology
- 2021 · Ocampo Daza, Bergqvist, Larhammar, "The Evolution of Oxytocin and Vasotocin Receptor Genes in Jawed Vertebrates: A Clear Case for Gene Duplications Through Ancestral Whole-Genome Duplications" — Frontiers in Endocrinology
- 2021 · Theofanopoulou, Gedman, Cahill, Boeckx, Jarvis, "Universal nomenclature for oxytocin–vasotocin ligand and receptor families" — Nature
- 2016 · Paré, Paixão-Côrtes, Tovo-Rodrigues, Vargas-Pinilla, Viscardi, Salzano, Henkes, Bortolini, "Oxytocin and arginine vasopressin receptor evolution: implications for adaptive novelties in placental mammals" — Genetics and Molecular Biology
- 2020 · Waltenspühl, Schöppe, Ehrenmann, Kummer, Plückthun, "Crystal structure of the human oxytocin receptor" — Science Advances
- 1993 · Winslow, Hastings, Carter, Harbaugh, Insel, "A role for central vasopressin in pair bonding in monogamous prairie voles" — Nature
- 2023 · Berendzen et al., "Oxytocin receptor is not required for social attachment in prairie voles" — Neuron
- 2024 · "Oxytocin receptor controls promiscuity and development in prairie voles" — bioRxiv preprint, not peer-reviewed
- 2005 · Kosfeld, Heinrichs, Zak, Fischbacher, Fehr, "Oxytocin increases trust in humans" — Nature
- 2015 · Nave, Camerer, McCullough, "Does Oxytocin Increase Trust in Humans? A Critical Review of Research" — Perspectives on Psychological Science
- 2020 · "A registered replication study on oxytocin and trust" — Nature Human Behaviour
- 2021 · Sikich et al., "Intranasal Oxytocin in Children and Adolescents with Autism Spectrum Disorder" — New England Journal of Medicine
- 2024 · MedlinePlus Genetics, "AVPR2 gene" — National Library of Medicine