Amplifying Genetic Logic Gates
This web page was produced as an assignment for an undergraduate course at Davidson College.
In this paper, Bonnet et al., present a design for a new type of biological logic gates. These logic gates utilize a genetic device termed a transcriptor to allow cells to perform simple Boolean logic computations. The transcriptor consists of a DNA segment containing GFP, preceded by a transcriptional terminator which is flanked by one or two sets of recombination sites. These recombination sites, when activated by bacterial integrases, allow the transcriptional terminator to be excised or inverted and reinserted into the DNA segment. The terminator is asymmetric, and when inverted it prevents the downstream production of GFP by causing the RNA polymerase to dissociate from the DNA. Thus the transcriptor has two states – the regular, un-inverted state where GFP production is blocked; and the integrase-activated inverted state where GFP production is allowed.
The genes for production of the bacterial integrases are inserted into a cell on a separate plasmid. The two genes, TP901-1 and Bxb1, are induced in a concentration-dependent manner by the application of arabinose (ara) and anhydrous tetracycline (aTc), respectively. Addition of exogenous ara and/or aTc leads to production of the integrases, which in turn leads to the inversion of a transcriptional terminator, which can allow or disallow the production of GFP. By combining recombination sites and transcriptional terminators in different arrangements, Bonnet et al., have created AND, NAND, OR, NOR, XOR, and XNOR logic gates, with ara and aTc serving as the two inputs, and relative fluorescence (GFP production) as the output.
These gates have a number of useful properties which distinguish them from those which have been previously-described. For example, the same regulatory molecules (the integrases in combination with their inducers) can be used with any of the gates, which means that all of the gates can be utilized at the same time in a single logic layer. The logic gates are shown to be highly digital and exhibit significantly amplified fold-changes in GFP production compared to control constructs. In addition, the gates are constructed using pre-existing methods and both the integrases and their inducers have proven to be viable in a variety of organisms ranging from prokaryotes to animals, meaning that these gates have the potential to be used in many organisms and in combination with other synthetic devices.
This paper surprised me in the sense that I would have thought logic gates this simple would have been developed sooner. I’ve seen other papers describing genetic logic gates and comparatively, the design of Bonnet et al.’s, gates seems remarkably simple and elegant. Instead of a complicated, interconnecting mess of regulatory molecules and gates serving as inputs for other gates, Bonnet et al., have produced all of the basic 2-input Boolean logic gates using the same simple device architecture. The gate architecture consists of, at most, two sets of recombination sites and attached transcriptional terminators, all controlled by two exogenous input molecules. Small rearrangements in these elements produce each of the different types of gate. I found the concept of the gates to be very intuitive, and I thought it was neat that such subtle changes could differentiate the gates. What’s more, because all of the gates utilize similar architecture and are controlled by the same inputs, it’s easy to see how multiple gates could be integrated simultaneously into a cell to produce a single, more complex logic layer. It seemed to me like these gates could form a good basis for producing more complex genetic circuits in the future. Other data in the paper suggest that these gates operate reliably, as well; they appear to be sufficiently digital and to produce significantly amplified changes in downstream gene production.
I thought the paper was concise and relatively easy to understand, as a whole. However there does seem to be a lot of important data mentioned in the text that is only contained in supplementary information, which made things a little harder to follow. The figures, though intimidating at first, did a good job of helping me visualize the gate architecture and how it worked. The figures also provided good statistical data about the reliability and efficacy of the gates. I did think that the visualization of data in figure 4 was a bit convoluted and hard to understand, though.
Figure 1. Mechanism through which logic gates operate and example XOR architecture. Figure from Bonnet, et al., 2013. Permission granted.
Panel A shows that the control signals, ara and aTc, lead to the production of integrase, which acts directly on the DNA of the logic gate. Action of the integrase on the logic gate can either allow or prohibit the flow of RNA polymerase along the DNA, thus allowing or prohibiting the production of output.
Panel B shows that recombination sites flanking transcriptional terminators allow for the inversion of the transcriptional terminator. The terminator is asymmetric and does not function to stop transcription when inverted. This means that a transcriptional terminator can be located along the DNA in either an un-inverted, transcription-blocking state, or an inverted, transcription-allowing state.
Panel C shows that, alternatively, recombination sites can allow for the excision of a transcriptional terminator from the logic gate DNA. Excision of a transcriptional terminator allows for downstream transcription.
Panel D shows the design of the XOR logic gate that Bonnet et al., have created. Two nested sets of recombination sites flank a transcriptional terminator, and each set is activated by one of the integrases. If no integrases are present, the terminator blocks downstream production of output. The presence of only one integrase inverts the transcriptional terminator about one pair of recombination sites, allowing downstream production of output. The presence of both integrases inverts the terminator twice, once about each set of recombination sites, causing no net change and leaving transcription of output blocked. Thus, output is only produced in the presence of one of the two integrases, consistent with XOR gate design.
Figure 2. Designs and output intensity of each logic gate. Includes device architecture, predicted and actual fluorescence intensity, and percentage of cells in “on” or “off” state under given input conditions. Figure from Bonnet, et al., 2013. Permission granted.
At the left of the figure we see the name of each gate and its corresponding truth table. The truth table is read horizontally across a row, with 0s indicating absence of a molecule and 1s indicating presence. For example, the AND gate requires presence of both inputs for production of output, so 1s must be present in both input boxes for a 1 to be present in the output box.
The “Predicted” column shows a heatmap representing the predicted percentage of cells exhibiting an “On” state (1 in the output box of the truth table) at a given concentration of two inputs (ara and aTc).
The “Population Measurement” column shows the actual output of each logic gate, given in amount of GFP produced (measured via fluorescence output), at different concentrations of the inducers.
The “Single Cell” column shows a plot of the GFP output of each individual cell for a particular combination of inputs. The input conditions are given by the truth table to the far left of each plot. The red line indicates the divide between cells in the “Off” state (not producing significant amounts of output) and cells in the “On” state (producing significant amounts of output). The bar graph to the right of each plot shows the percentage of cells in the “On” state (cells to the right of the red line).
Figure 3. Digitization and error rates for each of the logic gates. Figure from Bonnet, et al., 2013. Permission granted.
Panel A shows the response digitization of the XOR gate cells in response to aTc. The red contours indicate the fluorescence output of cells with the logic gate, whereas the blue contours represent control cells that only have an inducer-sensitive promoter preceding GFP. The XOR gate switches between “Off” and “On” states between .2 and 2 ng/mL of aTc, which is shown by the jump in the red contours between the .2 and 2 ng/mL panel. This discrete jump indicates that the XOR gate has a digital response to input. A slow increase in output as input concentration was increased would indicate an analog response. The red contours are also higher than the blue contours, indicating that the XOR gate cells produce amplified amounts of GFP relative to the control cells. The red and blue contours are not very far apart, however, suggesting that the XOR gate does not exhibit greatly amplified output in response to aTc as compared to control cells.
Panel B is similar to panel A, but it charts the output of the XOR gate cells and control cells against increasing concentrations of ara. The XOR cell response to ara is also very digital, with a discrete jump between the “Off” and “On” states between 10-4 and 0.5-3 % ara, weight/volume. The distance between the red and blue contours in farther than in panel A, suggesting that the XOR gate cells exhibit more highly-amplified GFP production, relative to controls, in response to ara than in response to aTc.
Panel C shows the digitization error rate of the controls and each of the logic gates in response to aTc. Cells were given an intermediate concentration of aTc and fluorescence output was measured. The black line indicates a digitization threshold that marks the difference between “Off” and “On,” and the number indicates the percentage of cells that did not present the expected state. The AND chart, for example, shows that 4% of the cells that were supposed to be suggesting that these gates are fairly reliable in that they present the state they are expected to be in when given an intermediate amount of aTc. The NAND and NOR gates exhibit digitization error rates greater than the control, suggesting that these gates are less reliable than the others.
Panel D is just like panel C, but shows responses to ara rather than aTc. None of the gates exhibit a digitization error rather greater than that of the control. This suggests that all of the gates are fairly reliable in that they present the state they are expected to be in when given an intermediate amount of ara. The NAND gate has the highest error rate (as in panel C), which could be due to its significantly different architecture that makes use of a constitutive promoter, unlike the other gates.
Figure 4. Amplification of output of select gates in response to inducer, relative to controls. Figure from Bonnet, et al., 2013. Permission Granted
Panel A shows the output fold change on the y-axis over increasing fold changes in control signal, ara,on the x-axis. The dotted line indicates a slope of 1, and represents the control cells where output is GFP is equal to input of inducer. The other lines show the response of AND, OR, and XOR gates to application of ara. The logic gates reach greater levels of output at intermediate concentrations of ara, showing that they all produce an amplification of output relative to controls. The colored boxes indicate the greatest amplification achieved by each logic gate.
Panel B shows the same amplification action by each logic gate, but in response to aTc.
Panel C shows much the same thing as panels A and B, but graphs response to ara by inverted amplifier gates NAND, NOR, and XOR. Here the output fold change is significantly less than the control, as expected, indicating that the gates amplify output response in the negative direction as well. The logic gates are capable of producing better down-regulation of response than controls.
Panel D shows the same amplified negative response, but in response to aTc.
Bonnet, J., Yin, P., Ortiz, M.E., Subsoontorn, P., and Endy, D. (2013). Amplifying Genetic Logic Gates. Science, 1232758. Published Online.
Erich Baker's Homepage
Biology Home Page
Email Questions or Comments to email@example.com or firstname.lastname@example.org
© Copyright 2013 Department of Biology, Davidson College, Davidson, NC 28035