A skin cell is like a ball rolling across hilly terrain. When young, it hops easily between shallow dips, but with age the landscape changes: some dips grow deeper, others flatten out, and the ball gets stuck in the same hollow more and more often. That hollow is a stable state of the cell, one in which the genes run an "old" program.
Biologists spent a long time searching for which genes make a cell age. Typically they compared young and old cells and looked at which genes showed the biggest change in activity. They found dozens of such genes, but attempts to perturb them did not rejuvenate the cell. It turned out that the striking changes are not the cause but the consequence: like smoke above a chimney, visible from afar but not controlling the fire in the furnace.
The result was a model in which each gene can be either on or off, like a light bulb. Such models are called Boolean networks. They were invented by Stuart Kauffman back in 1969, long before it became possible to read the activity of thousands of genes in a single cell. He proposed that cell types are precisely the stable states of the network, its "dips."
In one of the skin aging trajectories, a model of 613 genes showed a single deep dip: the old state. When the researchers virtually switched off the BACH2 gene, that dip disappeared and the ball rolled toward the young side. The improvement reached 98.9% of the theoretically possible maximum. Yet none of the 25 genes whose activity changed most with age produced any useful effect.
This result changes the strategy for hunting anti-aging drugs. Previously researchers looked for the "loudest" genes, but now it is clear: you have to look for quiet regulatory nodes that hold the cell in the old dip, even if their own activity barely changes. Such nodes are like the latches on a gate: you cannot see them, but they are the ones that decide whether the passage opens.
For now this is a computational experiment, but it offers concrete targets for testing in cell cultures. If knocking out BACH2 or ASCL2 really does rejuvenate keratinocytes, the same approach could be applied to other tissues and even to cancer, where the stable states of tumor cells are likewise described as dips on a landscape.
Boolean networks for modeling gene regulation were proposed by Stuart Kauffman back in 1969, long before single-cell sequencing existed. His idea that cell types correspond to network attractors only gained experimental support decades later.
The idea of reprogramming a cell from an old state to a young one by intervening in its regulatory network echoes the plot of the film 'The Cloverfield Paradox', where a particle accelerator accidentally alters the fundamental parameters of reality. Here, though, the switch is not random but computed.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.