Brewing with Synthetic Biology
A new approach offers a more efficient way to design biological “circuits.”
Synthetic biology rests on the hope that biological “parts” like DNA and proteins can be engineered and assembled just like a machine or computer circuit, but the field still has some way to go before this is the case. As much as biologists know about the structure and function of biological molecules, their behavior when interacting with one another is still unpredictable.
A new approach detailed in this week’s issue of the journal Nature Biotechnology offers a more systematic approach to constructing biological “circuits”–one that makes it easier to predict how they will behave before they are synthesized. The researchers used the technique to engineer yeast for brewing different kinds of beer more precisely, but the approach could also be used in the production of biofuels and therapeutic drugs and for other applications.
Genomics has given scientists an inventory of genes and their associated proteins, but a significant portion of the genome is regulated by other genes in interacting networks. For example, genes can turn each other on or off, and genetic sequences called promoters can control whether certain genes get switched on or get repressed. Synthetic biologists use information about these interactions to construct synthetic gene networks (collections of genes that interact directly or indirectly through their RNA and protein products) that can be introduced into microbes or mammalian cells to perform a desired task, such as creating microbes that produce a chemical by-product that could be used as fuel.
James Collins, a synthetic biologist at Boston University and senior author of the new paper, notes that the engineers who construct computer circuits have hundreds of components to choose from and can map out a schema of the circuit ahead of time, knowing exactly what its components will do. When synthetic biologists construct a genetic circuit, however, it’s a process of trial and error that “can take months or years of tweaking,” he says. Furthermore, these biologists often rely on just a few parts that must be altered to achieve the desired effect.
To make the process more efficient and predictable, Collins’s group has developed an approach that relies on constructing libraries of component parts combined with mathematical modeling to predict the behavior of these parts in a gene network before it is built.
The scientists focused on gene promoters, which are important for fine-tuning the activation of genes in a network. They screened a large list of variants of two promoters to create a library of variants categorized according to the strength of their activity. The researchers then used this functional information to create a computer model of the promoters’ activity, which allowed the different variants to be incorporated into a gene network with a predictable and quantifiable result.
In one demonstration, the researchers used a genetic network to precisely control a process in yeast that is crucial for brewing beer, called flocculation. The timing and strength of flocculation determines whether a beer is cloudy and tannic or clear and smooth, and initiating the process often relies on chemical additives. To genetically engineer flocculation control into yeast, the researchers constructed a toggle switch, a type of synthetic network that makes use of two mutually opposing promoters that can toggle between one state and another. If one promoter is weak and the other is strong, the stronger one will eventually overwhelm the weak one, flipping the switch. The length of time that it takes to flip the switch can be tweaked by changing the relative strengths of the two promoters, essentially creating a genetic timer. The researchers engineered this network into yeast so that when the toggle switches on, the yeast is made to flocculate.
Collins says that simply knowing the properties of components like promoters is not always enough to predict their behavior in complex networks like a genetic timer. His team had to assemble and characterize a single network from a set of parts in order to create a model that could be generalized for other parts in the library. Although this approach does require more work up front to build and characterize a library of components and experimentally test a network, Collins says that it is much more efficient than trying to retrofit a network by trial and error after it is built.
Jeff Hasty, a synthetic biologist at the University of California, San Diego, says that the study addresses a critical “component problem” in synthetic biology: there are only a limited number of components available and limited ways to regulate their activity. Although other groups have begun to construct libraries of genetic parts, Hasty says that the computer modeling shown in this study is also important for predicting how the parts will behave. While electrical engineers can make use of fundamental laws to predict the behavior of circuits, biologists are far from having those kinds of axioms. They must instead develop models based on empirical evidence from experimentation, Hasty says, in order for synthetic networks to “have some hope of behaving the way we predict they’ll behave.”