Controlling Synthetic Bacteria

But with genetic modification comes the concern that scientists will create new and uncontrollable species that outcompete or share their genes with wild-type organisms, permanently altering the environment or endangering people’s health.

Earlier this year, two research teams led by Yale bioengineer Farren Isaacs and Harvard geneticist George Church showed that they could genetically modify Escherichia coli to incorporate synthetic amino acids into essential proteins. When the bacteria are not fed the amino acids, they cannot produce these essential proteins, and so they die. This strategy yields bacteria that are very unlikely to survive without support from scientists but requires intensive engineering of the bacterial genome. (See “GMO ‘Kill Switches,’” The Scientist, January 2015.)

In contrast, Collins and his colleagues set out to create kill switches that could work in a more diverse range of microbes. “Our circuit-based safeguards can be conveniently transferred to different bacterial strains without modifying the target cell’s genome,” he wrote in an email.

First, Collins and his colleagues generated a kill switch called “Deadman,” named for a locomotive braking system in which the train will only run if the engineer is affirmatively holding down a pedal. In the microbial version of Deadman, a researcher must feed bacteria a substance called anhydrotetracycline at all times, or else the microbes will express a toxin and self-destruct.

The researchers generated a genetic circuit containing genes for the proteins LacI and TetR, a toxin that is only expressed in the absence of LacI, and a protease that degrades LacI. Under normal circumstances, TetR is preferentially expressed over LacI. TetR expression also triggers expression of the protease, which degrades any LacI that has been expressed. Without LacI, cells express the toxin and die. But when the cells are fed anhydrotetracycline, TetR is inhibited and LacI is expressed. LacI represses the toxin and keeps the cells alive.

Other versions of the Deadman circuit can be designed to degrade essential proteins in the absence of anhydrotetracycline, said Collins.

A second kill switch, “Passcode,” similarly requires that researchers maintain a specific environment for cells lest they express a toxin. Passcode requires a combination of input molecules for cells to survive. The system relies on hybrid transcription factors, each with one component that recognizes a specific DNA sequence, and one component that is sensitive to specific small molecules, such as galactose or cellobiose. One hybrid transcription factor, factor C, turns off expression of a toxin. Two other hybrid transcription factors, factors A and B, suppress expression of factor C. But specific small molecules can keep them from interacting with C. Another small molecule could prevent C from repressing the toxin. Therefore, to keep the cells alive, researchers must provide them with two small molecules that keep factors A and B in check, and make sure not to give them a third small molecule that will interfere with C.

Scientists designing Passcode kill-switches could make hybrid transcription factors respond to whatever combination of small molecules they desired, said Collins. “The strength of our kill switches lies in their flexibility and their ability to detect complex environmental signals for biocontainment.” He noted that companies hoping to keep others from using their cells could keep the recipe for their feed a secret.

“The great advantage is that you can effectively scale this and create different combinations of environments that contain different cocktails of these small molecules, thereby allowing you to effectively create a suite of cells that are going to be viable in different environments,” said Isaacs.

But Church warned that Collins’s circuit-based approach might not as effectively contain bacteria as an amino acid-based method, like one his group developed, since the cells are not fundamentally dependent on foreign biology to survive.

“If you need to have the ability to really scale your containment across a number of different species, then I could see the Passcode kill switches would be incredibly valuable,” said Isaacs. “If you are very concerned about escape frequencies and your degree of biocontainment, maybe you’d opt for something where the organism has been recoded and it relies on a synthetic amino acid.”

Still, Stan said the new paper is a demonstration that creating easy-to-insert kill switches based on genetic circuits is feasible. “I think what they wanted to show in the paper is basically that using some existing genetic circuitry . . .  you can obtain biosafety for the here and now.”

C.T.Y. Chan et al., “‘Deadman’ and ‘Passcode’ microbial kill switches for bacterial containment,” Nature Chemical Biology, doi:10.1038/nchembio.1979, 2015.