Genetic Logic Gates and Flipping DNA

The ability for cells to control gene expression in response to changing conditions is crucial for their survival. People spend their entire careers teasing apart these mechanisms, many of which are surprisingly complicated.

In synthetic biology, controlling gene expression is important in order to meet goals like generating medical devices or producing commodity chemicals. The most popular approach to tackling the biological complexity we see in natural systems is to dispense with it entirely. A small field within synthetic biology is working to make simple and reliable logic gates within living cells to act as versatile control elements, streamlined and digital alternatives to the complexity we see in nature.

Logic gates come in a variety of flavors. Computer scientists use them to convert two binary inputs into a single binary output. Outputs from two logic gates can then be wired as inputs into a third logic gate, and so on. These logic gates are absolutely crucial for the data processing that makes our digital world possible.

Most approaches to genetic logic gates have relied on direct control of transcription, and my own contribution to this field used control of translation¹. The challenge is converting inputs that vary somewhat in concentration into discrete-0 or 1-outputs. Usually there is a correlation between signal concentration and output gene expression, while a true logic gate gives a binary output: 1 or 0, all or nothing.

A recent paper² uses a creative approach to this problem. Instead of using inducer molecules to directly drive expression of the output gene, they use the inducers to drive expression of recombinases that then proceed to make irreversible, physical changes to DNA.

DNA is inherently "binary" in a couple ways: a given section of DNA is present or absent, and it has either a forward or reverse orientation. You can assign a 1 or 0 to the measures presence/absence and forward orientation/reverse orientation by making those conditions control expression of a reporter gene (GFP, green fluorescent protein, is the favorite choice). There are no intermediate states for a given DNA strand, since there is nothing between present and absent, and no orientation between forward and reverse since a segment of DNA is anchored in place on either end by chemical bonds. Fortunately, recombinases can control both DNA orientation and presence/absence in a fairly straightforward way.

Salmonella uses this trick to control the expression of adjacent flagellum genes by flipping the promoter that turns the genes on and off³. The system used by these bacteria was the basis of the first in vivo computer⁴, created by the labs where I got my start in research. By flipping and/or deleting transcriptional terminators using a similar system, the authors of this paper² were able to create 5 of the 6 logic gates.

To simplify this a bit, I drew an example focusing only on the AND gate. Like in electronics, the AND gate produces an output of 1 only if both inputs are also 1 (so, 1 AND 1 produces 1). In all other cases, it produces an output of 0. While electronic AND gates use current as both input and output, the genetic AND gate uses two small inducer molecules as inputs, which in turn cause expression of the recombinases that directly activate the logic gate. The output is expression of GFP, causing the E. coli cells to glow green.

At the top of my diagram is the AND gate in its default state: both inputs are 0, as there is no input yet! RNA polymerase (purple blob) is unable to move past the two transcriptional terminators ("STOP") that together compose the AND gate. When both recombinases are expressed (input: 1,1), the two forward-pointing terminators are flipped independently into reverse orientation. In this new orientation the terminators are unable to block RNA polymerase, and transcription proceeds all the way through the GFP gene. The transcript is then translated into GFP protein.

The authors of the paper designed related systems of flipping and deletion of directional transcriptional terminators to create the remaining logic gates: OR, XOR, XNOR, and NOR. Then they tested how predictably they behaved, and how "digital" they were. It's important to note that, although a given DNA strand is completely digital (present or absent, forward or reverse), each cell has many plasmids carrying the logic gates, and a large population of cells must be sampled in order to perform the experiment. In addition, expression of the recombinases is under control of inducers, and so the rate of recombination depends somewhat on the concentration of inducer.

Encouraging for the field, when the inputs for the logic gates were at their extremes (high concentration or none), the output pattern looked digital. When both inducers were at high concentration, almost all of the cells with AND gates showed a spike in fluorescence. When only one or neither inducer was present, nearly all the cells were at the low, baseline level of fluorescence. This more or less held true for all of the gates, though some switched state more completely than others under the appropriate conditions. This is good news. For applications where you want to create permanent changes in DNA, just add a lot of inducer and the job is done reliably.

Other applications might involve inducers that vary in concentration, such as where the logic gate is responsible for responding to a change in cell state. They tested how sharply the gate transitioned from off to on (or the reverse) across intermediate concentrations of inducer. Subjectively, I would say that the switching was all around pretty good. The error rate (the number of cells giving the inappropriate output for their logic gate given the inputs) varied from gate to gate, and inducer to inducer, but was typically less than 20%. Adding more refined control elements might reduce that even further.

Synthetic biology is currently faced with an important tactical question: Should we force biology to behave like electronics, or should we recycle the same mechanisms biology already uses, but in new ways? We are nowhere close to eliminating biological complexity entirely; even near-digital biological devices operate within the messy world of the cell. Whether we should embrace that or capture more ground for the digital remains to be seen, but it is becoming increasingly clear that initial critics of synthetic biology were wrong: biology is not too complex to rationally design.

Want to learn more? Check out this YouTube video narrated by senior author Drew Endy.

References:

1. Sawyer, E. M. Bacterial logic devices reveal unexpected behavior of frameshift suppressor tRNAs. Interdisciplinary BioCentral 4, 10 (2012). [Open Access]

2. Bonnet, J. et al. Amplifying genetic logic gates. Science 340, 599-603 (2013).

3. Johnson, R. C. & Bruist, M. F. Intermediates in Hin-mediated DNA inversion: a role for Fis and the recombinational enhancer in the strand exchange reaction. EMBO J. 8, 1581-1590 (1989). [Open Access]

4. Haynes, K. A. et al. Engineering bacteria to solve the Burnt Pancake Problem. J. Biol. Eng. 2, 8 (2008). [Open Access]