In my last post I wrote about the enormous bacterium Epulopiscium, and how it addresses the limitations of Brownian motion. Seeing this, Alex Small sent me a fascinating paper from a few years ago that I hadn’t seen before: “The energetics of genome complexity,” (Nick Lane and William Martin, Nature 467: 929 (2010)) [ungated PDF]. (Thanks, Alex!) The paper explores a deep question: why aren’t prokaryotes very complex? They seem never to have figured out how to access the morphological wonders of multicellularity, for example, while their eukaryotic cousins have, repeatedly. There isn’t some particular trait, like large size, that’s absent in all prokaryotes, so the lack of complexity must have a deeper reason. As the authors write:
Virtually every ‘eukaryotic’ trait is also found in prokaryotes, including nucleus-like structures, recombination, linear chromosomes, internal membranes, multiple replicons, giant size, extreme polyploidy, dynamic cytoskeleton, predation, parasitism, introns and exons, intercellular signalling (quorum sensing), endocytosis-like processes and even endosymbionts. Bacteria made a start up virtually every avenue of eukaryotic complexity, but then stopped short. Why?
(Each of the traits has one or more references cited, which I’ve deleted in the quote.)
The authors propose insights into this mystery that, I note below, once again lead us to puzzles of shape, size, and scaling. As an added bonus, the story also involves membranes. Lane and Martin point out two key differences between prokaryotes and eukaryotes: (1) eukaryotes have vastly larger genomes, by a factor of ~1000; and (2) eukaryotes use much more power per cell.
(1) is well known. Bacterial genomes are around 1-10 megabases long. Eukaryotic genomes span an enormous range, from about 10 to 10^6 megabases, with no obvious correlation between size and complexity or anything else (see The Creation of the Birds). (2) is less well known, and is related to (1). The vastly larger amount of DNA in eukaryotes in itself doesn’t cost much energy (to duplicate, for example), which is why there’s little pressure to reduce genome size. However, expression of the proteins encoded by the genome takes lots of energy. The power consumption of eukaryotes is a few thousand pW per cell, on average; for prokaryotes, it’s about 0.5 pW per cell. Expressing a lot of genes enables a complex array of functions in eukaryotes. Where does the required power come from? Mitochondria.
Mitochondria generate the ATP used by eukaryotic cells as a source of chemical energy, and this generation involves the establishment of very large electrical potentials across the mitochondrial membrane. This much I knew, but what I was not aware of is that it is hypothesized that physical proximity between the mitochondrial DNA and the membrane is crucial to mitochondrial function. Apparently, mitochondrial gene expression is regulated by the oxidation state of the encoded proteins, and the oxidation state presumably rapidly decays to “normal” as a function of distance from the membrane.
Put more simply: to have lots of energy, a cell needs a lot of energy sources, and these need to be located near membranes. Mitochondria satisfy these needs.
Some prokaryotes, as we saw with Epulopiscium, might have lots of DNA in their large cellular volumes, but since “ATP synthesis scales with plasma membrane surface area but protein synthesis scales with cell volume,” this won’t lead to efficient energy use. Lane and Martin note that Thiomargarita namibiensis has thousands of copies of its genome localized to a “shell” just inside its plasma membrane, but this still leaves it with the same scaling problem: the demands of the cell volume outweigh the supply from its surface. Only eukaryotes, it seems, have dealt with the issue by recruiting an endosymbiotic organelle that provides both ATP and lots of membrane area.
The problem, and its resolution, remind me of lungs. Small animals don’t need lungs: their surface area is sufficient for diffusion of gases across surfaces to supply the needs of the animal’s bulk. A human is about 1000x larger in length than an ant, and so, with a similar anatomy, would have 10^6 times as much surface area but 10^9 as much volume — clearly a problem. The solution: lungs, which provide an enormous surface area for gas exchange — human lungs are about 100 square meters in extent!
This raises the question, though, of why prokaryotes don’t try the same trick, making folds and corrugations, poking inside or outside, that vastly expand their surface area. Localizing ATP-related genes to the vicinity of the membranes, perhaps with many copies of plasmids rather than the whole genome, should achieve the same ends as the eukaryotic adoption of mitochondria. Why aren’t there any star-shaped bacteria? Perhaps there’s some mechanical reason — maybe non-convex shapes or high curvatures are hard for bacteria to maintain. (Though viruses don’t seem to have problems with controlling small-scale geometries.) Or perhaps somewhere out there there’s a pincushion-shaped prokaryote, just waiting to be found…
Update (Dec. 4): Both Matt Jemielita and Tristan Ursell point out that there are star-shaped bacteria: Stella and prosthecate bacteria (link1, link2, link3), and both direct me to papers by Kevin Young, which I should probably read!
The Eighteenth Elephant 
My guess: if transport is limited by diffusion from a distance source to the general vicinity of the bacterium, you may not gain much by adding membrane surface area (starfish-like bacterium). In the lungs, air is moved by convection except over the last little bit, so you do gain by adding membrane surface area. It may boil down to the Peclet number.
Good point — this makes sense for transport of gases or dissolved molecules. For simply having a large surface over which to have a membrane potential, like the mitochondria do, it’s not clear to me that diffusion would matter.