Scientists have genetically tweaked
E. coli bacteria to create simple computers capable of solving a classic math puzzle, commonly called the “Burnt Pancake Problem.”
The resulting advance in synthetic biology, according to researchers, hints at the ability of tiny “living computers” to aid in data storage, evolutionary comparisons and even tissue engineering.
The mathematical problem imagines pancakes of varying sizes stacked in random order — each with a burnt side and a golden brown side. The solution requires using the minimum number of manipulations to stack the pancakes according to size, with their burnt sides all facedown. Each manipulation involves flipping one or more pancakes, reversing both their order and orientation.
Scientifically, the flipping process is known as sorting by reversals and is the sort of challenge whose complexity increases dramatically with every added pancake. By the time the stack reaches 11, the problem becomes “extremely hairy,” said Karmella Haynes, a visiting adjunct professor of biology at Davidson College in Davidson, N.C.
“It’s kind of like that computer in ‘The Hitchhiker’s Guide to the Galaxy,’ ” she said, referring to a popular novel by the late Douglas Adams. “It’s been working on a problem so long that by the time it comes up with an answer, everybody forgot the question.”
At that level of complexity, “bacteria could probably outperform a conventional computer at solving the problem,” said Haynes, the lead author of a new study suggesting exactly that.
Bacteria as tiny computers
Since 2000, multiple studies have focused on the largely untapped potential for bacteria, yeast and mammalian cells to be harnessed as tiny and abundant computers. “Ours is unique in that the operation required to solve the problem takes place in a living cell,” Haynes said.
The unique in vivo system takes advantage of the remarkable storage capacity of DNA and the efficiency of molecular self-assembly, Haynes said. DNA replication and bacterial cell division can quickly create millions or even billions of parallel processors. “The more little computers you have working on the problem,” she said, “the greater the likelihood that one is going to pick the right path that will take you to the right solution.”
The project began as an undergraduate entry in the International Genetically Engineered Machines, or iGEM, Competition, a showcase for the emerging field of synthetic biology. The students, under the direction of Davidson College professor A. Malcolm Campbell and in collaboration with Missouri Western State University, chose a mathematical puzzle that might lend itself to a biological solution: the Burnt Pancake Problem.
For the study, published last month in the Journal of Biological Engineering, the collaborators tried to demonstrate the advantages of a bacterial approach by exploiting a well-known enzyme from the Salmonella bacteria that cleaves DNA every time it recognizes a specific sequence of 26 DNA letters. The enzyme, Hin invertase, has the potential to cut a piece of DNA, flip it 180 degrees and put it back together.
“It’s like you have spatulas to handle the pancakes you’re flipping, to flip them over within the stack,” Haynes said.
To construct the eight possible solutions for how to flip two DNA “pancakes,” the group introduced both the Hin invertase enzyme and the DNA sequences it recognizes into E. coli, creating 100 distinct DNA components and intermediates in the process. In nature, Hin invertase can cut and flip a single section of DNA. For the project, the group showed that the enzyme can flip one piece, an adjacent piece or both at the same time. When correctly ordered and oriented, the combined DNA pieces were designed to reconstitute a gene allowing E. coli to grow even in the presence of the antibiotic tetracycline.
Unexpectedly, the researchers found that even some scrambled DNA arrangements confer tetracycline resistance. The wrinkle, Haynes said, underscores the need for researchers working on larger “pancake stacks” to find a reliable calling card — a distinct color change or survival of the bacteria, for instance — only when the right combination has been achieved amid a host of possibilities. Several of the project’s participants are now working toward that goal.
Sorting by reversals
Despite the engineering challenges, Haynes said the bacterial system could theoretically tackle any problem that can be converted into a “sorting by reversals” format, including the evolutionary genomics puzzler of how two species are genetically related. Such a problem would require understanding how DNA chunks can become both re-ordered and re-oriented over the millennia. In trying to gauge the evolutionary distance between mice and humans, for example, “You’re essentially asking, ‘What’s the minimum number of reversals that it takes to transform a mouse genome into a human genome?” she said.
A more sophisticated setup might lend itself to data storage, given the compactness of flipped chunks of DNA. “If you imagine that a forward orientation is a 1 and a backward orientation is a 0,” Haynes said, “then you’ve got binary code” — the basis of computer data-encoding. But because Hin invertase scrambles DNA at random, researchers interested in encoding a meaningful array of informational bits would need to exercise far more control over how the fragments are flipped and in what order.
Ron Weiss, an assistant professor of electrical engineering and molecular biology at Princeton University, said the new study provides a “nice demonstration that there’s some capability to instruct cells to carry out computational tasks. And I think this certainly brings a new aspect to what’s been demonstrated before.” Weiss, who wasn’t involved with the research, said the effort is another sign that the field is progressing, despite its relative infancy. “We really are at the beginning, at the vacuum tube stage or something like that, if you use the comparison with computer electronics,” he said.
Understanding how biologically produced parts fit together into a larger whole will be key in determining how quickly the field matures. “Instead of two to five components, you’d have 20 or 50. That’s really going to be perhaps the biggest challenge as we move forward.” Advances in computational modeling and DNA synthesis, however, have provided hope that even sizeable synthetic arrangements aren’t out of reach.
Coaxing bacteria to behave
Three years ago, Weiss’ group coaxed bacteria to make complex shapes and fluorescent color patterns based on instructions from other bacteria, a study that yielded striking images of a bull’s-eye, heart and flower and the tantalizing possibility of creating “smart” devices able to detect hazardous compounds. The team has since introduced a similar pattern-forming capability into mammalian cells, spurring the cells to create distinct tissues instead of colors and shapes.
“Not only can we produce a pretty picture, but they can be valuable,” he said of the biological instructions, though he stressed that science has a long way to go before a similar toolkit could be called upon for whole-scale tissue engineering.
Despite the complexity, Weiss said the underlying mechanism can be thought of as a simple toggle switch: with one signal, cells may become muscle, while another spurs them to differentiate into bone. Adding signals, like the DNA “pancakes” in Haynes’s research, augments the number of choices for what the cells ultimately become.
Separately, Weiss’s team has coaxed mammalian cells to manufacture a communication system normally found only in bacteria and known as quorum sensing. Within the scheme, a secreted molecule allows a bacterial population to sense its own size and adjust its numbers accordingly.
“We now have the mammalian cells that can send and receive these bacteria quorum-sensing signals,” Weiss said, a success that has encouraged his team to work toward a more sensitive setup that one day might be used for recruiting cell populations to transform into, say, bone or neurons.
What about safety concerns?
“For applications that have to do with environment or human health, you really have to build in safety features,” Weiss said.
One biological safety valve under development would cause the whole system to abort if the cell population passes a certain number. Theoretically, he said, scientists also should be able to detect errors in biological networks the same way in which errors can be pinpointed on computer chips, and researchers are working toward developing such quality-control assurances.
Within the broader field, he said, synthetic biologists have adopted the engineering mindset of integrating safety features into their designs, perhaps well aware that their work could be for naught if they cannot assure the public that progress won’t come at the sake of prudence.