Human-designed cells are becoming more of a reality every day. But engineering a microbe—for example, that mobilizes metal better than its natural counterparts, cleans water or hunts down and kills a tumor—is still a challenging task. Cells are extremely complicated, and the high uncertainty surrounding them slows down the design process.

According to Adam Arkin at the University of California, Berkeley, at the start of cellular engineering one needs to build standardized sets or parts that can control a particular function in a cell. Ideally one would have many of these control devices that respond to different inputs and that can be used together in a combinatorial fashion.

Researchers in Arkin's laboratory are interested in both the evolutionary and human-manufactured design of cells. “We have one part of the lab that tries to reverse-engineer how cells work and another where we try to exploit what we learn for engineering new function in cells,” he explains.

His group has recently become interested in the engineering of transcriptional controllers for bacteria. Bacteria control the expression of some of their genes through the use of 5′ transcriptional regulation units called cis-regulatory leader peptides. In these systems, first described by coauthor Charles Yanofsky, translation of the leader peptide by the ribosome determines whether the RNA polymerase will efficiently transcribe downstream genes. Arkin and postdoctoral fellow, Chang Liu, identified cis-regulatory leader-peptide elements as a new way to achieve transcriptional regulation units. By engineering this mechanism, the group created transcriptional switches that respond to genetically encoded unnatural amino acids.

Unnatural amino acids are commonly used for protein bioengineering, but the group decided to adapt them as a signaling framework. They hypothesized that by introducing blank codons that do not encode natural amino acids into a leader-peptide sequence, they could prevent its proper translation and therefore affect transcription of downstream genes. If they then introduced the necessary machinery to load these blank codons with unnatural amino acids into the cell, they could make the translation of the leader peptide dependent on the presence of these small molecules. As a result, the group built two classes of genetic switches: transcriptional 'off' switches in which the addition of specific unnatural amino acids inhibit transcription of desired downstream genes and transcriptional 'on' switches that act the opposite way.

One of the exciting capabilities of this system is its modularity and the possibility of making combinatorial logics with it. “It works as if we are building railroad switches; you can add an arbitrary number of individual switches that respond to a given unnatural amino acid, each of which controls whether the train continues moving in one direction or not, and the movement of the train integrates the decisions made at each switch,” explains Liu. A second advantage is that this system relies on components engineered for expanded genetic codes and as such, the number of unnatural amino acid–induced switches that could be made is large and quickly expanding.

In this work the group used the system to control the expression of GFP in Escherichia coli, but this system's potential for more sophisticated applications, such as engineering therapeutic bacteria, is an ongoing effort that Arkin's group is pursuing with the help of collaborators. Notably, other organisms such as yeast or humans also use cis-regulatory leader-peptides to control gene expression. Although several adaptations will be needed to translate this tool to eukaryotes, the group is actively at work on this as well.

Once again, this type of research teaches us how the knowledge obtained from basic science studies—here aimed at understanding how microbes control transcription to survive in the world—can be imaginatively put into practice for the building of tools that will one day be of practical value to humans.