Shipman’s project began when he and his colleagues converted the image of a hand into DNA code, so that different sequences of DNA letters represented the color within each pixel. They manufactured all that DNA in the form of small strands, each of which was tagged to resemble the kind of viral sequences that the CRISPR system would naturally snag. The team delivered these strands to colonies of E. coli, which they grew overnight. They sequenced the part of the microbes' genomes where CRISPR information is stored, and decoded those sequences back into digital data. This allowed them to successfully recover the picture of the outstretched hand.
Encoding the horse GIF was more challenging. Not only did the team have to encode each frame, but also the order of the frames. Fortunately, CRISPR makes that easy. When bacteria grab viral DNA, they always insert new sequences after old ones, so the CRISPR system naturally orders information from newest to oldest. Shipman’s team took advantage of that. They offered their bacteria the DNA strands representing each frame of the GIF, one by one. By sequencing everything afterwards, they could recover the full movie. And if they loaded the bacteria with the DNA strands in the reverse order, they recovered a GIF of a horse running backwards. “We wanted to show that we didn’t need to encode timing information at all,” Shipman says.
Even when the team let the bacteria grow for a week—many generations in their time—the data didn’t degrade. “Those sequences aren’t doing anything, so there’s probably little cost to the having them,” says Shipman. Also, the data is distributed, so no single cell in the colony completely encodes the hand photo or the horse GIF. That prevents each individual bacterium from becoming overloaded with extraneous DNA.
“It’s a nice proof of principle of the potential for a living storage system,” says Dina Zielinski, who’s now at the Curie Institute in Paris. “Traditionally, the DNA in which data is encoded is finite. Once the data is written, it can be copied and read, like when you drag and drop new files onto your USB drive. But with a living system there's the potential to ‘write’ additional information. A living system allows for a more dynamic storage architecture.”
Your imagination could run wild thinking of applications for this. When I asked my colleagues, they came up with ideas like conducting cellular espionage by encoding secret messages in microbes, encoding your name in something else’s DNA like a form of genetic graffiti, and rickrolling future scientists by infusing bacteria with the lyrics to “Never Gonna Give You Up” and burying them in permafrost.
None of this is what Shipman has in mind. “We’re not using this to record movies,” he says. “We’re not going to put Wikipedia into bacteria.” Instead, he came into this as a neuroscientist who found it vexingly hard to study what happens as our brains develop—which genes are activated when, and where, and in which neurons? “It’s hard to reconstruct that because every time you touch the system, you disrupt it,” he says. “I imagined that if we had some way of encoding data in living cells while they’re still alive, and have it stored there, we could make progress.”
That’s the ultimate goal: turn cells into living recorders. “We want cells to go out and record environmental or biological information that we don’t already know.” That might include everything from the level of pollutants in a lake, or the genes that the microbes switch on as they go about their lives. All of this is a long way off, but the first necessary step is to show that cells can indeed store meaningful amounts of information.