Cells are Sloppy

This is the final post in my series on evolution. Previously we encountered LUCA, the last universal common ancestor of all living things today, and discovered rather a lot about the origins of life.^⌘ Now it’s the turn of LECA, the last eukaryotic common ancestor. Before we plunge in though, I’ll quickly review what we’ve learnt in the past 17 posts. If you’ve diligently phagocytosed everything I’ve written, you may wish to skip the next section. Conversely, if you’ve never encountered the archaea, you’re in for a slightly bumpy ride.

At the start^⌘ we met the apparent mystery of how complex Nature is. We revisited Lies to Children, and discovered the wisdom of moving on to better models of how things work. We then found how natural selection^⌘ can best be considered a process that will predictably happen in specific circumstances: imperfect copies are made in an environment where this variation matters—a Grim Reaper selectively deletes those versions that are less suited to the environment. We found clearly observable natural selection happening in almost real time, and naturally segued back to 1859, where Charles Darwin^⌘ pointed out how this works.

We subsequently (by request) spent considerable time exploring On the Origin of Species, and discovered, contrary to common, current opinion, deep insights that are often still relevant. We saw how Darwin was almost uniquely capable at self-criticism, and how this has preserved the robustness of his views over one-and-a-half centuries. He was a good scientist.

In exploring a herbivorous spider,^⌘ we found how he emphasised interconnectedness. We also spotted one of Darwin’s few mistakes—his emphasis that Nature doesn’t make big jumps. She can and therefore she will: we saw this with whole genome duplication events.

To understand how easy it is not to adequately understand a theory, we briefly digressed into scurvy,^⌘ where we discovered the importance of digging right down into the details of why something works, and how this digging counteracts bad science.

We then tried to explain the ridiculous precision of bees.^⌘ Darwin had already worked out that it’s not just physical attributes that are susceptible to natural selection, but behaviour and instinct too. As an aside, we touched on the fascinating topic of bee genetics, specifically haplodiploidy, and this brought us to the contentious topic of kin selection. By popular vote, we however left this for later in favour of an exploration of slavery.^⌘

Hybrids came next,^⌘ notably ducks. We realised that the “tree of life” can be messed up a bit by randy drakes indulging in a bit of introgression, resulting in something that is more network-like than tree-like. The concept of ‘species’ is blurry. We found a more than superficial resemblance between Nature and Nanny Ogg, who will try anything once. Nature even tried the Amazon molly.

Then there was sex.^⌘ Lots and lots of sex. Meiosis, oogamy, isogamy, a bewildering array of hermaphrodites, a multitude of different ways of determining sex,^⌘ and even organisms with thousands of sexes.^⌘ And, of course, those damn frogs.

We then got really down and dirty, with parasites.^⌘ How they work, and how they happen. Endoparasites, ectoparasites, parasitoids, parasitic castrators, directly and trophically transmitted parasites, micropredators, and vectors. We saw how, with parasites, nothing is simple—a land snail can eat some cow poo, walling off parasitic miracidia in its gut and passing out a slime ball. An ant then eats the ball, producing metacercariae that migrate to its brain. Goaded by parasites, the ant goes off on its own every evening and climbs to the top of a blade of grass, staying there with its mandible clamped to the blade till dawn, when it rejoins the colony. Eventually a cow eats the ant; an adult fluke forms in its liver; and the life cycle continues. Parasites may be as arbitrary as it gets.

When we moved on to the ‘Modern Synthesis’,^⌘ we discovered that despite the exciting juxtaposition of Mendel’s experiments, natural selection, and a solid dose of maths, things didn’t quite fit. Something was missing.

So we looked at mutation,^⌘ and discovered that the genome is anything but a library of well-organised genes, that genes are far from tuned to a single purpose, and that there is no ‘efficient design’ among cells and organisms.

We re-emphasised that species are pretty blurry, and that the ‘tree of life’ is often more of a web. We encountered hemiplasy, and the ‘tree of one percent’. We discovered spandrels^⌘ and at the same time saw how brilliant sociobiologists can miss the plot (something we later spent a whole post on^⌘) and smart-ish people like Dawkins can make a religion out of plain old-fashioned atheism. In contrast, pretty much everything we discovered about Nature shows that she’s sloppy, and happy to jolly things along.

We took a dip into Lamarckism^⌘ and confronted how limited epigenetics is. Finally, we discovered Sewall Wright,^⌘ who brought us back to our senses. We realised what we’d missed—and then realised that we knew it all along! What was this? Let’s re-discover it a third time. We’ll start by looking at LECA.

Remarkably, although LECA likely lived about two billion years ago, we can now make very reasonable deductions about how it worked. The key concept here is orthology: the study of genes that diverged from a common ancestor. There’s a huge, public database of these relationships with the cute name eggNOG, containing thousands of representative organisms.

It turns out LECA was really rather complex. If we look at all the families of protein-encoding genes we see in modern organisms, it had about half of these. A very recent (June 2026) paper by Cox et al. teases out some of the details.

LECA would have had at least one nucleus, linear chromosomes & centromeres, all sorts of internal membranes (‘endomembranes’: an endoplasmic reticulum, Golgi apparatus, vesicle trafficking, and a nuclear membrane), a cell skeleton that used actin and tubulin, centrioles, one or more cilia, lysosomes, peroxisomes, and of course mitochondria, which are very special indeed.

A lot of the ‘newness’ in LECA was likely related to sophisticated processing of DNA and RNA, especially now that the organism had a nucleus. Cox et al worked out that many genes of previously unknown function are likely for cell movement and the cytoskeleton.

The human genome has also lost a lot of the original LECA genes. This isn’t that surprising, given the obvious examples of amino acids that we simply can’t synthesise, and other essential dietary components like vitamin C. In addition, we struggle to remove toxic substances like glyoxylate, which LECA could have handled with ease. Sometimes, Nature just throws out things that could have been very useful.

As the image right at the start of this post suggests, LECA likely arose from an unusual arrangement. Most of the main bits came from an archaean, but other key components came from bacteria! Some of the latter was simply by lateral gene transfer, but a major change stands out: mitochondria. These almost certainly represent the legacy of fusion of that archaean and a bacterium1 that could burn up molecular oxygen, turning the products of sugar digestion into carbon dioxide, water, and huge amounts of energy, stored briefly as ATP. This is such a vital and interesting process, I think we should dig into it a bit.

Mitochondria are amazing. They also provide a nice illustration of how arbitrary Nature is.

We first met adenine as one of the four (A,U,G,C) bases that, with ribose and phosphate go to make up RNA.^⌘ Now we meet it in the form of adenosine triphosphate (ATP). The extra two phosphates are important, particularly the third one, as its high-energy phosphate bond carries a lot of grunt. It’s the energy currency of the cell, and like any good credit card, is valid in pretty much any intracellular store. The same magnesium ion that is important for polymerisation of bases into RNA also binds ATP avidly. Sometimes Nature is good at reuse.

If you burn sugar, relatively small amounts of ATP can be produced in the absence of oxygen (by ‘glycolysis’, producing pyruvate), but respiration in mitochondria produces about 15 times as much ATP per sugar molecule. The mitochondrion takes pyruvate and turns it into water and carbon dioxide, creating a hydrogen ion gradient in the process. Those H⁺ still need to be used. The key enzyme is now ATP synthase. This is a tiny molecular machine like a ratchet that can use a hydrogen ion gradient to make ATP.

The finer details are important. ATP synthase sits across a cellular membrane. It has two pieces, F_O and F₁.2 A tiny little molecular axle rotates within F₁! F_O drives this rotation. With rotation in one direction, ATP is synthesised; turn the other way, and ATP is broken down, transporting H⁺.

Look at the picture above. On the left in magenta is F_O. On the right in garish green, we have F₁, with a brown axle between the two. A cyan stator holds the two apart. As this is difficult to visualise, here’s a really cool animation.

Were we to run ATP synthase backwards, it might be used as a pump. Break down ATP, turn the axle, and we can move H⁺ against a gradient! And we encounter just this mechanism in the form of an ancient enzyme: V-ATPase. Eukaryotes use this for many things: pumping acid into vesicles, and driving the exchange of other molecules, such as bicarbonate in kidney cells, Ca²⁺ ions in osteoclasts that make bone, and even neurotransmitter transport into synaptic vesicles in nerve cells. Sperm wouldn’t be able to fertilise egg cells without V-ATPases.

We can look back further into the past. There’s an archaeal ‘A-type’ ATPase that is related to the V-type, but can also be used to make ATP. And this in turn is related to the F-type version that is used by bacteria (and mitochondria, and chloroplasts) to synthesise ATP from a proton gradient.

There’s been a lot of debate about which came first, but now we seem to be leaning more and more towards ATP synthesis early on. In a 2023 Nature article, Tara Mahendrarajah and colleagues reconstruct the history of ATP synthases. They estimate the divergence of F-, and A/V-types at or even before LUCA, more than 4.3 billion years ago (Ga). LECA by their estimates is placed at 1.93 – 1.84 Ga.3

Mulkidjanian and colleagues argue that the rotary dynamo of the F and A/V-type ATPases started life even earlier as a way to transport molecules across membranes—specifically, RNA and proteins. And before this, the hexamer may have acted as an RNA helicase, powered by ATP to unwind RNA.

Everywhere you look, you seem to see these little motors. Bacterial flagella are mostly constructed outside of the cell. The components need to be exported, driven by ATP. There is now evidence that the Type III flagellar protein export mechanism is homologous to our familiar ATPases.

With all of this innovation based on rotary motion, you’d think that other rotating nanomachines would have the same origin too, and that ‘flagella-like’ mechanisms would all be pretty similar. Not so! Nature re-invented the rotor at least 3 times. Even though extrusion of the bacterial flagellum seems to be based on our familiar ATP-based rotor, the rotary movement of the flagellum itself uses a completely different mechanism!4 And then we have the archaellum, yet another distinct rotary motor.

A more modern view of the eukaryotic cell, shaped by drift-dominated evolution, is a messy junk-filled entity, full of Rube-Goldberg contraptions that were hobbled together by non-adaptive forces—Alexander Palazzo and Nevraj Kejiou.

This exploration of rotating motors consolidates an idea that has been building over many posts. On the one hand, Nature seems to parsimoniously and inventively re-use components. A rotary nucleic acid unwinder becomes a trans-membrane transporter of RNA or protein, and then the same mechanism is used both for ATP manufacture and ion pumping. But then all sorts of baroque re-inventions pop up—a proliferation of rotary motors.5 Similarly, we’ve seen eyes invented again and again, and baroque divergences like parasites, and parasites within parasites.

If you take a creationist viewpoint, then the cell is designed “like a watch” by a magnificent watchmaker. Over the past 17 posts, I’m fairly sure you found it increasingly apparent that this is just silly—but if you’ve been diligent in your reading, you will also have realised that the Modern Synthesis fell into the same trap! Read Dawkins, and he largely substitutes the blind force of natural selection for some sort of deity. But he attaches similar idealistic properties that just aren’t there!

The rotary motors are just a microcosm of the whole. As Wright pointed out, there is a massive, multi-dimensional network of interactions between arbitrary genes. This is both ugly and (on occasion) profoundly beautiful—because why else should it be any other way?

It’s human for us to anthropomorphise Nature, and there’s little harm in the metaphor itself. But then we must realise that Nature is naturally arbitrary, and capricious and sloppy—she will walk down any and every possible alley that is susceptible to exploration. She is massively subject to the random drift of chance and circumstance.

Most mutations are of course non-advantageous, and even deleterious mutations can become fixed in a population that is small enough or unlucky enough. Later, they may then—again, pretty arbitrarily—be put to good use as circumstances change. Nature draws pictures of saints on spandrels, simply because they are there. In principle and in practice, pretty much any gene product can be hijacked for any purpose where it fits at the time.

We could end there. But there’s still that one niggling, unresolved problem. Perhaps archaea can help us?

In a 2015 paper in Nature, Anja Spang and her colleagues surveyed deep sea sediments from near hydrothermal vents in the Arctic Mid-Ocean Ridge, notably from near Loki’s Castle, 3283 metres below sea level. They pulled out 92% of the genome of a remarkable archaean they called Lokiarchaeota, and made this bold claim:

“when eukaryotes were included in our phylogenetic analyses, they were confidently positioned within the Lokiarchaeota”.

In other words, our closest early relatives likely were archaea. The drawing at the start of this post is rooted in this idea. When they looked for eukaryotic signature proteins (ESPs), a significant part of the predicted proteome (175 proteins) was most similar to eukaryotic proteins! They found evidence of an actin cytoskeleton, and Ras GTPases were there: both previously thought to be pretty much confined to eukaryotes. There was also internal machinery that resembles eukaryotic endosomes (ESCRT).

Naturally, they were met with early skepticism but, continuing the Norse mythology theme, in 2017 they found Asgard. Katarzyna Zaremba-Niedzwiedzka and colleagues sampled diverse aquatic sediments around the world6 and found genes of multiple organisms related to Lokiarchaeota, fitting into the ‘Asgard’ clade. These are Heimdall-, Odin- and Thorarchaeota. Again, ESPs are rampant.

The remote possibility that somehow these genes were all artefacts was finally put to bed by ten years’ of hard work by Hiroyuki Imachi and a large number of colleagues. Working with deep-sea sediments sampled from the Omine Ridge, Nankai Trough south of Osaka, they actually grew an archaeon they called “Candidatus Prometheoarchaeum syntrophicum strain MK-D1”, pictured above. This is incredibly tedious work: “Grow in a methane-fed anaerobic bioreactor. Wait a year. Check. Repeat.”

It’s full of ESPs, but the key insight from these studies is even more special. We are now pretty sure that LECA represents the alliance of an ancient archaean with aerobic bacteria that eventually became mitochondria.7 Later on, early plants acquired chloroplasts through a similar mechanism.

But what was that mechanism? There is still a widespread belief that this was an aborted hostile event. The archaean phagocytosed the proto-mitochondrion, intent on digestion, and somehow this didn’t happen. We are now coming to the realisation that this may be wrong.

“P. syntrophicum” lives cheek by jowl with bacteria (notably, Methanogenium). In contrast to the above rapacious behaviour, P. syntrophicum suggests another way: working together, better and better. We commonly see Nature as red in tooth and claw—and with eukaryotes, she often is. Archaea however seem to have a more laid-back approach. Perhaps we can learn from them that not all organisms are continually competing, and that some learned to work together two billion years ago.

We’re back to Darwin’s ‘entangled bank’^⌘—the idea that everything depends on everything else, and we all co-evolve together, changing our environment and being changed by it. That’s the bit we all miss, again and again. Let’s not ever forget it.

Which seems a fine place to end my essays on evolution.

My 2c, Dr Jo.

^{⌘ This symbol is used to indicate posts where I’ve discussed the flagged topic in more detail.}

Next Sunday, I’ll start a new series that may now seem a bit anachronistic: ➵ How to program well.