Keeping at a border between two whole colonies — a star ascidian whose vessels reach out to a neighbour’s, so the pair either join into one circulation or raise a row of necrotic points and part, the decision resting on whether they share an allele at one histocompatibility locus
R949 closed with a crowd that had lost everything except its cells’ surfaces. This page takes the turn it pointed to, and restores the bodies. On a rock or a hull sit colonies of the star ascidian Botryllus schlosseri: many small zooids embedded in a shared gelatinous tunic, each colony budding new zooids on a weekly cycle as the old generation is resorbed, all of them plumbed into a common vascular system inside the colony. That internal system ends at the colony’s edge in blind, finger-like vessels — ampullae. A colony grows outward. Sooner or later its ampullae meet a neighbour’s, whole and in place, and something has to be decided at the touch.
First, the meeting is vascular before it is anything else. The two colonies do not merge their zooids, and they do not exchange cells across open tissue. What reaches across is the plumbing: ampullae from each side extend through the tunic toward the other colony and make contact tip to tip. Only then does the pair take one of two courses. Compatible colonies let the contact vessels join, the border between their tunics breaks down, and the two vascular systems open into each other — one circulation, two genotypes, a natural chimera. Incompatible colonies do roughly the opposite, at the same tips, and the border hardens into a scarred line neither side crosses.
Second, the rule is one shared allele. Classical breeding work placed the decision at a single, highly polymorphic Mendelian locus, called Fu/HC for fusibility / histocompatibility, its alleles expressed codominantly. The rule those crosses produced is unusually clean: if the two colonies share at least one allele there, they treat each other as self enough to fuse; if they share none, they reject. A gene inside that system was identified in 2013 and named BHF, for Botryllus histocompatibility factor — highly polymorphic, strongly expressed in the vasculature, and upregulated in colonies poised to fuse or reject, with later work describing its protein as a compatibility signal whose recognition holds a rejection programme in check. One caution belongs here: naming BHF is not the same as closing the mechanism. The locus sits in an extended haplotype, other allorecognition responses may ride alongside it, and exactly how recognition is read at the ampulla tip is still worked out piecemeal.
Third, rejection is built, point by point. Where no allele is shared, cytotoxic morula cells crowd into the interacting ampullae, cross into the tunic, and discharge their granules there. Among their contents is phenoloxidase, an enzyme whose activity darkens and kills in a small neighbourhood: the contact border acquires a row of dark, necrotic lesions called points of rejection. Vessels haemorrhage, retract, or are amputated at the border, the colonies change their growth away from each other, and the reaction, once run, is not taken back. Recent work adds a limiter to the same event: material released with the granules forms a scaffold that traps the enzyme and its dark products near the tip, so the killing stays at the border instead of spreading through the tunic that both sides still have to live in. Rejection here is not a wall raised in advance. It is damage dealt precisely, then walled off by its own debris.
Fourth, fusion is not peace — it is an opened border. A shared circulation carries more than blood. Mobile somatic and germline stem cells cross into the partner, take up residence in its niches, and compete with the resident lines. The outcomes are strange enough to have earned their own vocabulary: somatic and germline parasitism, where one genotype’s stem cells progressively supply the other partner’s new zooids, or its gametes, or both. The two competitions are separate — germline takeover follows a repeatable, heritable hierarchy between genotypes, while somatic takeover is less predictable and unlinked to it — so a colony can lose the soma and win the germline. And a fused partner can be eliminated outright by allogeneic resorption, its zooids phagocytosed until little but tunic and vessels remain, while its genotype nevertheless persists inside the survivor, sometimes as the only genotype in the survivor’s gonads. Sharing an allele buys the pair a common bloodstream. It buys neither of them a guarantee about whose body, or whose offspring, that bloodstream will end up serving.
Last, that asymmetry is the point of the whole apparatus. Set this beside R949 and the contrast is exact. The sponge’s scattered cells recognised their own one contact at a time and rebuilt a single body; nobody in the dish was asked to live with a second genotype inside the result. Botryllus asks the harder version: two complete, growing colonies, each already a chimera of budded generations, deciding at a moving border whether “own” stretches one allele’s worth to cover the neighbour. If it does, the border dissolves and the argument moves inside, stem cell against stem cell, for the soma in one contest and the germline in another. If it does not, the border becomes the record: a line of dark points where two circulations touched, tested each other, and refused.
A walk from a meeting to its two endings. Five stages: two colonies in place on a shared surface; ampullae reaching tip to tip at the border; the allele comparison at the histocompatibility locus; the branch — vessels join, or morula cells discharge and points of rejection form; and what follows each branch, a chimera whose stem cells compete, or two colonies growing apart behind a scarred border. Schematic, compressed from the published accounts.
Walk it in order.
Schematic only: the stages compress work across colonies, populations and laboratory assays, so no single meeting runs this exact account. Labelled cautions: do not say fusion is decided by the whole genome — the classical rule is one shared allele at one locus, against otherwise unrelated genotypes. Do not present BHF as the whole mechanism of the Fu/HC system — it is the identified gene within an extended haplotype, and the reading mechanism is still being worked out. Do not call rejection a wall built in advance — it is a cytotoxic reaction at the ampulla tips whose necrotic points become the border. Do not call a fused chimera harmonious, or fusion a benefit either partner can bank — resorption and germline or somatic parasitism are documented outcomes, and no clear advantage to chimerism has been established. And do not merge the two stem-cell contests: germline competition is hierarchical and repeatable, somatic competition is not its echo.