A team in Silver's HMS lab led by Caroline Ajo-Franklin, now at Lawrence Berkeley National Laboratory, and postdoctoral scientist David Drubin decided to demonstrate that not only could they construct circuits out of genetic material, but they could also develop mathematical models whose predictive abilities match those of any electrical engineering system.
"That's the litmus test," says Drubin, "namely, building a biological device that does precisely what you predicted it would do."
The components of this memory loop were simple: two genes that coded for proteins called transcription factors.
Transcription factors regulate gene activity. Like a hand on a faucet, the transcription factor will grab onto a specific gene and control how much, or how little, of a particular protein the gene should make. The researchers placed two of these newly synthesized, transcription factor-coding genes into a yeast cell, and then exposed the cell to galactose (a kind of sugar). The first gene, which was designed to switch on when exposed to galactose, created a transcription factor that grabbed on to, and thus activated, the second gene.
It was at this point that the feedback loop began.
The second gene also created a transcription factor. But this transcription factor, like a boomerang, swung back around and bound to that same gene from which it had originated, reactivating it. This caused the gene to once again create that very same transcription factor, which once again looped back and reactivated the gene.
In other words, the second gene continually switched itself on via the very transcription factor it created when it was switched on.
The researchers then eliminated the galactose, causing the first synthetic gene, the one that had initiated this whole process, to shut off. Even with this gene gone, the feedback loop continued.
"Essentially what happened is that the cell remembered that it had been exposed to galactose, and continued to pass this memory on to its descendents," says Ajo-Franklin. "So after many cell divisions, the feedback loop remained intact without galactose or any other sort of molecular trigger."
As cool as this experiment is, it's worth remembering just how rudimentary it is. The "memory" was really just a biochemical linger, an echo of galactose lingering on as a stuck transcription factor. (Obviously, every memory is just some sort of biochemical event, but what gives those chemical events their meaning is the larger neural network. There is no network here.) But it's still a pretty cool bit of biological engineering. Lord knows what kind of memory chips our kids will be playing with.