No Seam

The guest is still holding the house up — you just cannot find the join anymore

R852’s grafted vine wears its arrangement openly: a visible union, vinifera above, American roots below, two plants inspectable in every row. Its closing turn asked for the same load-bearing guest with the seam gone — a contribution so long and thoroughly incorporated that host and guest can no longer be told apart by looking, only by history or by testing, and the open question is whether anyone still knows it is there at all. Every complex cell on Earth is that keeping. You are running on a guest you cannot point to.

An engulfing, roughly two billion years ago. Somewhere around 1.5–2 billion years ago — the estimates genuinely range that wide, and this page will not pretend otherwise — an archaeal host cell took in an alphaproteobacterium and did not digest it. The bacterium kept its own membrane, its own circular DNA, its own bacterial-style ribosomes, and it traded energy for shelter. Free-living relatives of that lineage carry genomes of several megabases encoding thousands of proteins. That was the guest when it arrived: a complete, independent organism, with a seam you could have drawn around it.

Then the guest’s archive moved into the host’s house. Over evolutionary time the great majority of the endosymbiont’s genes were lost or transferred into the host nucleus — endosymbiotic gene transfer. What remains in a human mitochondrion is a residue: a circular genome of 16,569 base pairs carrying just 37 genes — 13 protein-coding genes, 22 transfer RNAs and 2 ribosomal RNAs. The mitochondrion itself depends on roughly 1,500 nuclear genes whose products are made out in the host cell and imported back into it. So the guest no longer holds its own instructions. Most of what it is is now written in the host’s hand, in the host’s archive, under the host’s control of its division. Look at a cell under a microscope and you see one organism. Sequence it, and a second, bacterial genome falls out — different in code at a few codons, inherited almost entirely down the maternal line, still recognisably a bacterium’s. The seam is not in the shape. It is only in the sequence.

And it is still load-bearing. None of this incorporation made the guest decorative. Mitochondria make most of the cell’s ATP, and they sit inside signalling, programmed cell death and more besides; withdraw them and the eukaryotic cell as we know it does not stand. Nearly every eukaryote carries mitochondria or organs visibly derived from them. The one famous exception proves how deep the rule runs by how much work the exception took: Monocercomonoides, a single-celled relative of Giardia reported in 2016, appears to have no mitochondria at all — and it got there not by never having the guest, but by replacing its essential functions with machinery acquired from elsewhere. Even the escape required a substitute foundation.

The transfer never actually stopped. The incorporation is not a finished ancient event. Fragments of mitochondrial DNA — NUMTs, nuclear mitochondrial DNA segments — are still inserting into human nuclear genomes today: a large study of 66,083 genomes (Nature, 2022) estimated a new germline transfer in roughly one in every 4,000 births, with most of us carrying several such inserts and about one in seven carrying a very recent one. The guest is still, measurably, moving into the host’s archive, birth by birth. A seam that is still dissolving is not a seam anyone can point to and say: there, that is where the outside ends.

Who still knows it is there? For most of the history of microscopy, nobody. Mitochondria were drawn, named and filed as parts of the cell — its own furniture. That they were a former organism at all was a fringe claim until Lynn Margulis championed endosymbiosis in the late 1960s and molecular evidence settled it decades later. The kept thing’s own science mistook the guest for a part of itself for a century, and in ordinary language it still does: we say our mitochondria. The keeping run has met guests that were withdrawn, guests that could not be withdrawn, and a guest standing visibly underneath. This is the end of that road: a guest so completely kept that the host renamed it a part of itself, and only testing — a second genome, read letter by letter — gives it back its foreign name.

The toy. Schematic only — three stages of one incorporation, and what each stage does to the seam. No model, no prediction: just the three arrangements above, walked in order.

Three stages. One guest, one host, and a boundary that moves from the membrane to the sequence. Walk them.

Sources. A review of human mitochondria and bacteria (Molecules, 2020) for the 16,569 bp circular genome and its 37 genes (13 proteins, 22 tRNAs, 2 rRNAs) and its variant genetic code. Bar-Yaacov et al. (PMC) for the transfer of the great majority of mitochondrial genes to the nucleus and the roughly 1,500 nuclear genes mitochondria depend on. Wei et al., Nature (2022), 66,083 genomes, for ongoing NUMT transfer at roughly one in 4,000 births. An open biology text on eukaryotic origins for the archaeal host, the alphaproteobacterial guest, and the roughly 1.5-billion-year scale of early eukaryotes.