Get Function First, Become Evolutionarily Relevant Later

How studying the master regulator of X chromosome inactivation across primates reveals that regulatory novelty can emerge before natural selection even notices

When we think about evolution, we often imagine dramatic changes: a fish growing legs, a dinosaur sprouting feathers. But some of the most consequential evolutionary innovations happen at a scale invisible to the naked eye, in the molecular machinery that decides when and where genes get switched on.

The research I have conducted during my PhD, now published in Science Advances, explores how this regulatory machinery evolves. What we found challenges a common assumption in biology: that if a DNA sequence has a function, evolution should have preserved it. Instead, we discovered functional regulatory elements that appear to be almost like evolutionary accidents, at least for now.

Why evolution tinkers with regulation instead of genes?

Your genome is like a blueprint containing instructions for building and operating a human body. But having the plan isn’t enough. First, you need to know when to build the foundations versus the bathroom. Second, you also need to know which and how many building parts you’ll need. That’s gene regulation: the system that controls which genes are active, when, and how much product they make.

Here’s a key insight from decades of evolutionary biology: species might differ not because they have different genes, but because they use the same genes differently. Modifying how a gene is regulated can be evolutionarily “cheaper” than modifying an existing gene or inventing a new gene from scratch. A small change in when or where a gene turns on can have profound effects on an organism’s development and physiology.

But how do these regulatory changes actually happen? And once they appear, do they matter for the species’ survival? That’s what we set out to understand.

The experimental system: XIST and the X Chromosome

To track regulatory evolution in action, we needed a gene whose regulation really matters and for which we already knew some of the key players.

We chose XIST, a gene central to X chromosome inactivation. In mammals, females have two X chromosomes while males have one X and one Y. To balance gene dosage between sexes, females silence one of their X chromosomes early in development. XIST is the master switch for this process: its RNA coats one X chromosome and triggers its shutdown. If XIST isn’t regulated properly, development goes awry.

Previous work had shown that XIST regulation differs dramatically between mice and humans. Actually, so much so that some key regulators in mice don’t even function in humans. But comparing mice and humans means looking across 80 million years of evolution. A lot can happen in that time. To catch regulatory evolution in smaller steps, we turned to our closer relatives: rhesus macaques (diverged from humans ~35 million years ago) and marmosets (~55 million years ago).

We used embryonic stem cells from each species as an experimental model. Embryonic stem cells represent the early developmental stage when X inactivation is being established.

Mapping the XIST regulatory landscape

Finding gene regulators is detective work. We can’t simply read the DNA sequence and know what regulates what. Instead, we look for clues: chemical marks on proteins that coat the DNA, regions where the DNA is accessible to regulatory machinery, and the three-dimensional folding of chromosomes that brings distant sequences into contact.

Using these approaches, we identified three types of candidate XIST regulators:

Enhancers: short DNA sequences that boost gene activity. We found candidate enhancers near XIST in macaques, but intriguingly, the same DNA sequences in humans showed no signs of regulatory activity.

Long non-coding RNAs: RNA molecules that don’t encode proteins but can regulate other genes. We focused on JPX, already known to regulate XIST in mice and in humans.

Chromosome architecture: DNA doesn’t float randomly in the cell nucleus, it folds into organised domains. We discovered that this folding differs between macaques and humans in the region around XIST.

This last finding had a clear culprit: a transposable element, a piece of “mobile DNA” called HERVK, that inserted itself into this chromosomal location specifically in the lineage leading to macaques, around 29-33 million years ago. Transposons are famous for reshaping genomes, but whether these insertions actually do anything functional is an active area of research.

Testing the function of XIST candidate regulators

Identifying candidate regulators is only half the battle. The real test is perturbation: if you break a candidate regulator, does XIST regulation change?

We used CRISPR technologies to either delete DNA sequences entirely or silence them without cutting. Then we measured the consequences for XIST, both the total amount of RNA produced and, crucially, what happens in individual cells.

The results painted a picture of both conservation and divergence:

JPX is a conserved regulator, but not quite the same. In mice, humans, and marmosets, JPX controls whether XIST turns on at all. Disrupt JPX, and fewer cells activate XIST. But in macaques, JPX only fine-tunes how much XIST RNA is made, it doesn’t control the on/off decision. Same gene, related function, but a meaningful difference in mechanism.

The macaque-specific enhancer teams up with JPX. One of our candidate enhancers does regulate XIST, but only in macaques. The same sequence in humans sits idle. Moreover, in macaques, this enhancer works together with JPX: disrupting both has a stronger effect than disrupting either alone. This partnership doesn’t exist in other primates.

The transposon-induced DNA folding change protects neighbouring genes? When we deleted the HERVK element to give macaque cells a “human-like” chromosome architecture, XIST itself was unaffected. Instead, nearby genes got slightly activated. This suggests the new boundary could act as a shield to prevent the macaque-specific regulators from accidentally switching on conserved genes that shouldn’t respond to them.

The paradox: functional but evolutionarily neutral

The next question we asked is what is the relative importance of the functional changes we described for each species we studied. Bear with me because here’s where things get philosophically interesting.

In evolutionary biology, we often use sequence conservation as a proxy for importance. If a DNA sequence looks the same across many species, we assume natural selection has been protecting it because it does something essential. Rapidly changing DNA sequences are also presumed to be important as it indicates exploration of the sequence space toward a more evolutionary beneficial state. However, most of the mammalian genome (including mice, marmosets, macaques and humans) is evolving at rates that are indistinguishable from random chance. It means that most changes in their genomes do not impact neither the survivability nor the evolution of the species, instead the outcomes of those DNA changes are neutral.

Hence, to test the evolutionary relevance of the regulators we functionally characterized, we analyzed how they are evolving. In other words, is their DNA sequence changing faster or slower than expected by random chance?

The answer, for nearly every element we studied: neutral evolution. The sequences are changing at rates indistinguishable from random chance, with no statistical signature of natural selection acting on them.

This creates a paradox. We demonstrated experimentally that these elements regulate XIST. They have measurable functions. Yet evolution appears not to “care” about them, at least not enough to leave a detectable signature of selection.

Surprisingly, for the enhancer that functions specifically in macaques, its sequence is actually better preserved in humans. In human embryonic stem cells, this sequence doesn’t bear the hallmarks of an active enhancer and when perturbed with CRISPR, none of the neighboring genes, including XIST, were affected. Hence, we hypothesized that, in humans, this sequence might function as an enhancer in a different tissue or at a different developmental stage.

Constructive Neutral Evolution: function first, importance later

How do we reconcile demonstrated function with apparent evolutionary irrelevance?

The answer may lie in a concept called constructive neutral evolution. The traditional view of evolution emphasises adaptation: new features arise because they help organisms survive and reproduce. But constructive neutral evolution proposes an alternative path. Sometimes, molecular complexity accumulates not because it’s beneficial, but simply because it’s not harmful enough to be eliminated.

In this framework, regulatory elements can emerge through random processes, like transposon insertions or mutations, and acquire functions without those functions being essential. They become part of the system, contributing to how genes are regulated, even as they remain invisible to natural selection.

But here’s the key insight: these neutrally evolving elements represent latent adaptive potential. They’re like spare parts in a toolbox. Most of the time, they sit there doing something modest. But if the environment changes, if some new selective pressure emerges, these elements could suddenly become crucial. At that point, natural selection would kick in, and what was once neutral would become constrained.

What we learned

Our study captured regulatory evolution in action across three primate species separated by tens of millions of years. In that time:

• A transposon insertion rewired chromosome architecture in macaques
• An enhancer gained activity in macaques while its human counterpart remained silent
• A conserved regulator (JPX) shifted its mechanism, controlling on/off decisions in some species but only expression levels in others
• New regulatory elements co-emerged with potentially protective boundaries, preventing them from disrupting neighbouring genes

All of this functional novelty is evolving neutrally. It emerged, it works, but it hasn’t (yet?) become essential for survival.

This has implications beyond X chromosome biology. The genome is full of sequences whose functions we’re only beginning to understand. Our results suggest that function alone doesn’t mean evolutionary importance and, conversely, that lack of conservation doesn’t mean lack of function. The relationship between sequence, function, and fitness is more subtle than we often assume.

Perhaps the most interesting implication is about evolutionary potential. The neutrally evolving regulatory elements we discovered aren’t evolutionary dead ends. They’re options. They’re complexity waiting to become essential. In the grand experiment of evolution, sometimes you acquire function first and become relevant later.

Read the full paper: Cazottes et al., “Remodeling of XIST regulatory landscape during primate evolution,” Science Advances (2026)

Many thanks to Manon for her feedbacks on the draft of this post!