Â鶹ÒùÔº

How do cells know what they should become as the body develops? Biological development depends crucially on spatial patterns: the lines that eventually give rise to segments, organs, or markings like stripes and spots. Yet despite the variation in form, shape and structure in the animal kingdom, the mechanisms that generate these body plans are surprisingly similar across species.

For decades, scientists have known that webs of genes controlling other genes, known as "gene regulatory networks," play a central role in shaping these patterns. But how these networks evolve to create new patterns is less understood.

This question is being explored by Zena Hadjivasiliou, Group Leader of the Crick's Mathematical and Â鶹ÒùÔºical Biology Laboratory. Her team are aiming to unpick how the same genes and molecules produce life in all its diverse forms.

"We have an incomplete picture of how these patterning mechanisms evolve," Hadjivasiliou says. "When a new type of marking appears in a species, such as a new eye spot in a butterfly wing, it doesn't come from scratch. It emerges from subtle rewiring of the underlying genetic architecture. But what kind of mutations make this possible? And are there rules that govern which new patterns are possible, and which remain out of reach?"

Evolving gene networks: A well-trodden path?

The lab's latest research, today in PRX Life, explores whether specific types of mutations in patterning networks accelerate the evolution of new patterns, and if any of these changes yield predictable evolutionary outcomes.

Hadjivasiliou explains that the idea of predicting evolution has been pondered by biologists and philosophers alike over many years.

"There's a classic thought experiment from the paleontologist and evolutionary biologist Stephen Jay Gould: if we were able to replay the 'tape of life,' would we end up with the same diversity in the animal kingdom that we see today?" says Hadjivasiliou. "For Gould, the answer is no—seemingly inconsequential and random events in history pushed us down an evolutionary path which was therefore unlikely to ever happen again."

Changes that happen together stay together

Answering these questions experimentally is nearly impossible. Real embryos are complex, mutations are hard to track, and evolution plays out over immense timescales. To begin to tackle this challenge, Harry Booth, mathematician and computer scientist in the lab and first author of the study, built a virtual version of the process: a computer simulation that models how small networks of genes evolve under natural selection.

"Our system models gene networks that instruct cells to commit to a specific fate, depending on where they are in a tissue," says Booth. "Small changes within these networks can drive large differences in the spatial patterns they produce."

He continues, "By running the simulation not just once, but more than 100,000 times, we were able to build a statistical picture of how evolution tends to proceed in the search for new patterns. Once you reach this kind of scale, all the random features of evolution start to disappear and what you are left with represents general properties of these types of processes."

Using the computer simulation allowed Booth and Hadjivasiliou to uncover rules about the types of mutations that matter most. Adjusting an existing boundary—for example, shifting where a stripe begins or ends—needed only small tweaks to the strengths of existing gene interactions. But creating new boundaries was far more difficult. It usually demanded multiple changes at once: for example, adding entirely new regulatory links and flipping a gene's role from activation to inhibition simultaneously.

Booth summarizes, "In short, fine-tuning is easy, but genuine innovation takes a bigger shake-up of the network."

History makes its mark

And what about Gould's 'tape of life'? Booth took a novel approach to this question. Using the massive dataset of evolutionary trajectories, he trained a machine learning model that could forecast which network design a trajectory was heading towards, finding that, surprisingly, aspects of evolution could be predicted.

"History does make its mark," he concludes. "We found that certain mutations radically shift the predicted evolutionary outcome. This suggests that the mutation introduces a fork in the road early in the journey of evolution, which, despite evolution's inherent randomness, reliably redirects evolution to a specific destination."

This finding lends evidence to the hypothesis that chance events in history shape the future of a species' evolution. As Hadjivasiliou puts it, "the model makes headway in explaining how small, seemingly unimportant events in evolutionary history can have lasting impacts on the development and diversity of animal forms."

And what's next? Booth explains that they're hoping to pick up signals of the principles they identified in their virtual evolution experiment in genetic data from organisms such as fruit flies.

He concludes, "We've laid the groundwork for some really interesting experiments, now that we know a bit more about the types of evolutionary processes required to evolve these networks and how they can produce new patterns."