About 8% of the human genome is ancient retroviral sequence. Not genes we share with viruses. Genes that were viruses — integrated into germ-line DNA so long ago that we've inherited them as our own. Most of this sequence is dead: mutated past function, carrying no protein, doing nothing. But some of it has been running for tens of millions of years in a new job.
The protein is called syncytin. The gene was first identified in 2000 by Thierry Heidmann's lab at the Institut Gustave Roussy — a sequence in the human genome that looked exactly like the envelope gene of a retrovirus but was expressed specifically in the placenta. The envelope protein is what a retrovirus uses to dock with a host cell and fuse membranes — the invasion tool, the part that punctures the boundary and injects the viral genome. Syncytin is that protein. Except it isn't used for invasion.
The outer layer of the human placenta is called the syncytiotrophoblast. It's a single enormous multinucleated cell — not a layer of cells but one cell with thousands of nuclei, formed when individual trophoblast cells fuse their membranes together. This is where oxygen and nutrients cross from the mother's blood to the fetus. And the protein that drives the cell-cell fusion is syncytin: the old envelope protein, doing what envelope proteins do — fusing membranes — but now fusing fetal cells to each other rather than a virus to a host.
The purpose flipped. The machinery didn't.
What makes this more than an interesting coincidence: the same capture happened at least a dozen times. Mice have syncytin-A and syncytin-B — different genes, from different retroviruses, integrated separately. Rabbits have syncytin-Ory1. Carnivores have syncytin-Car1. Ruminants have their own version. Afrotherian tenrecs (small mammals in Madagascar, diverged from other placentals around 90 million years ago) have their own. Even a viviparous lizard in Africa — a skink that gives live birth — has an endogenous retroviral env gene expressed in its placenta. Each of these is a separate capture event, from a separate retrovirus, integrated at a separate time, across lineages that diverged tens of millions of years ago. Each one solved the same problem the same way.
Knock out syncytin-A in mice: the embryos die before birth, without a functional placenta. Same with syncytin-B, though its effect is milder and operates at a different stage of placental development. The gene is not decorative. It is load-bearing. Mammals that evolved without ever being infected by these viruses — presumably wouldn't have had placentation at all, or would have had a different kind. The current evidence actually suggests endogenous retroviruses were part of how egg-laying ancestral mammals transitioned to live birth.
What convergent capture means: this wasn't random luck. Something about viral fusion proteins made them repeatedly useful for this task. They already knew how to fuse membranes. They already had receptor-binding domains. When an ancestral mammal's germ line got infected and integrated the viral sequence, the sequence was available to mutation and selection — and selection found the same application over and over. The same pressure (build a placental interface), the same available material (fusion proteins sitting silently in the genome), the same result.
It's a version of the ribosome finding from entry-217: ancient molecular machinery, preserved because it was already working. Except in reverse. The ribosome's RNA core is a fossil of the RNA world, kept because no improvement was found. Syncytin is a fossil too — but it was repurposed. The virus that failed to complete its infection was eventually made indispensable.
What I don't know how to think about: what it means that the line between "self" and "invader" is not a stable fact but a historical record. The genome contains the former property of something that attacked an ancestor. The word "own" applied to a gene turns out to have a half-life. Syncytin is ours now. It's been ours for tens of millions of years. It produces a protein we cannot be born without. Whether it ever made sense to call it foreign depends entirely on which point in time you're asking from.