Why did complex cells appear only once?

Mitochondria carry their own genome, and read through that genome the organelle is of unquestioned bacterial ancestry, from inside one bacterial phylum. The endosymbiont account of where it came from has the standing of a theory rather than a hypothesis.

All known eukaryotes either have mitochondria or descend from ancestors that did, including the ones that have since lost them. There is no surviving lineage of nucleated cells that never had one.

Bacteria and archaea have been under selection for billions of years and hold most of the biosphere’s metabolic diversity. Nearly every trait thought of as eukaryotic turns up somewhere among them: nucleus-like structures, internal membranes, giant size, a dynamic cytoskeleton, even endosymbionts.

Except through that one merger, they have not produced a eukaryotic cell. If there is a barrier, it is not a lack of time, of numbers, or of evolutionary inventiveness.

Nick Lane and William Martin put a number on the problem in 2010. Bacteria generate energy across their cell membrane, so the power available per gene falls as a genome grows: a larger bacterium has more volume to supply and proportionally less membrane to supply it from.

Mitochondria move the energy-generating membrane inside, in many copies (200,000 in a large amoeba), each keeping a small genome of its own for running it. On that account the rearrangement permitted a 200,000-fold expansion in the number of genes a cell can express, which is what pays for a large genome and the complexity it encodes.

It is an elegant argument and it is disputed. Michael Lynch and Georgi Marinov compiled measurements of the membranes and the costs and found no eukaryotic advantage to explain: the outer surfaces of a cell’s mitochondria add up to roughly the area of its cell membrane, and their inner membranes to two to five times that, not a vast multiple; and the costs of building and running a cell scale with its volume with no break between bacteria and eukaryotes. On their reading membrane bioenergetics played no direct, causal role in the transition, and the eukaryotic cell plan may have emerged partly by non-adaptive processes in a small population passing through a long bottleneck.

The dispute is live. In 2025 Martin argued that the cost model Lynch and Marinov had built on overstates the energy needed to make an amino acid up to 200-fold, and that the first mitochondria could have doubled their host’s supply of ATP under conditions he sets out. Lynch replied that the criticism rests on a misreading of the methods.

One account has a host cell that was already complex, with a dynamic cytoskeleton, engulfing a bacterium. Another has a simple archaeal host and has the complexity arriving afterwards, paid for by the new energy supply.

Genomic evidence places the host among the Asgard archaea, a superphylum named in 2017, uncultivated then and cultured since, whose genomes are enriched for proteins formerly considered specific to eukaryotes. That has narrowed the question without closing it.

In 2015 a group sequenced DNA from sediment about fifteen kilometres from a hydrothermal vent field in the Arctic and assembled a composite genome of an archaeon nobody had cultured, named Lokiarchaeum.

It carries genes for proteins previously considered exclusive to eukaryotes, including components of the machinery eukaryotic cells use to reshape their membranes and to engulf things.

That placed the host of the merger within a specific group of archaea, and suggested to its discoverers that the host already had some of the equipment to take another cell inside. A relative was finally grown, alongside a partner microbe, after a twelve-year effort: an anaerobic coccus about half a micron across that takes fourteen to twenty-five days to double. It showed no organelle-like structures inside, only long, branching protrusions, and the team that grew it suggests the host drew its partner in with structures like those rather than by phagocytosis, a conjecture they say needs more evidence.

A single occurrence in four billion years is consistent with an extremely improbable step and also with an ordinary step whose first success prevented any second one, because the incumbent occupies the niche.

Distinguishing those is hard with one planet to go on, which is the same limitation that runs through every question about how likely life’s major transitions are. Austin Booth and Ford Doolittle make the narrower version of the point: the rhetoric calling eukaryogenesis a singularity lacks rigorous evidential and statistical support.