Team:William and Mary/Synthetic Enhancer
One of the fundamental constraints of an electrical engineering style way of thinking about genetic circuitry is that electrical circuits often include digital, often binary, states of expression. This kind of input-output relationship showing two discrete “on” and “off” states is represented by the illustrative gene expression curve in Figure 1. As one can see, although the binary interpretation is appropriate at the low and high extremes of inducer concentration, there is an intermediate region where the continuous, or analog nature of gene expression becomes important. A natural way to expand on the advantages of the digital-circuit paradigm while incorporating the non-binary characteristics of gene expression is to develop genes and circuits which can exhibit discrete, multi-step response profiles. These would allow circuits to exhibit not just “off” and “on” states, but a number of intermediate, discrete states to better facilitate both precision of circuit behavior and readout as well as the expansion of the possibilities in the computation that can be performed with genetic circuitry. Often people will use the low and high expression regions to represent the analogous expression levels to binary 0 and 1, respectively. However, Amit et. al. developed a method by which intermediate levels of gene expression can be obtained in a stable, non-transient manner. To do so, they created a synthetic enhancer suites, schematically shown in Figure 2, consisting of enhancer binding sites, an assembled σ54 promoter, and small cassettes between the enhancer and the promoter, which can house DNA binding proteins in different combinations. It is the looping of the DNA from the enhancer to the promoter that causes an interaction that allows for transcription of the output. Binding of repressor proteins, such as TetR, to the addition cassettes between the enhancer and promoter can affect the flexibility of DNA looping and thus make it thermodynamically more difficult for the interaction to take place and thus suppresses transcriptional initiation. This synthetic enhancer suite can not only increase the complexity and sensitivity of a circuit, but also allow for a multimodal response with addition of discrete sets of defined enhancer binding protein binding sites, TetR, that interact independently from the enhancer and promoter. To test our ability to shift an arbitrary circuit, we first attempted to shift our previous example circuit, a pTet GFP and constitutive expressed TetR. We obtained a plasmid containing 85 TetO repeats off of Addgene from Finney-Manchester et al. (2013) (“Harnessing mutagenic homologous recombination for targeted mutagenesis in vivo by TaGTEAM). We moved the segment containing the repeats to the Biobrick Backbone, and then transformed a reporter circuit containing pTet GFP and TetR (Bba_K2066053) on the high copy plasmid 1A3 either with or without the repeat array on the high copy 1C3 backbone. We induced both circuits with varying concentrations of aTC and then measured fluorescence using flow cytometry, which allowed us to get single cell level resolution (Figure 5)
We also created a mathematical model of a TetO binding array, and found that our data closely mirrored the expected results (Figure 6). While the 85x array will undoubtedly be useful to many teams in the future, one of the strengths of decoy binding arrays is their modularity. Since the magnitude of the shift is determined by the number of binding sites, it is possible to vary the magnitude of the shift by using the same binding array on different copy number plasmids. While this allows for more possible numbers of total binding sites, it is inherently limited by the fact that there are only a limited number of plasmid origins. To get around this problem we decided to submit to the registry a suite of Iterative Capped Assembly (ICA) parts that can be used to assemble a TetO or LacO binding array of any size. These parts were designed based upon the ICA method of assembling repeat sequences from Briggs, et al. 2012 (“Iterative capped assembly: rapid and scalable synthesis of repeat-module DNA such as TAL effectors from individual monomers”). Additionally, for the tetO ICA parts we submitted 3 different versions, with either 8,16 or 64 base pair spacers between the tetO monomers, this is because it has been suggested by Amit et al. 2012 (“Building Enhancers from the Ground Up: A Synthetic Biology Approach” that anticooperativity plays a role in the effectivity of DNA protein binding. Anticoopritivity means that if two repressor binding sites are very close to each other, a repressor binding to one can spatially hinder, or even completely prevent a repressor from binding to the other. We thought that enabling another level of tuning to allow for even finer levels of shift magnitude by altering the level of anticooperativity would be a useful tool to include within the binding array section of the toolbox. Finally, to help future teams get started, we created a mathematical model to determine the strength of titration needed to shift their transfer function, as well as the number of repeats required to achieve that titration for a given tetO or LacO array.
Synthetic Enhancer
Background
Figure 1 - Illustration of induction curve for generic gene expressionMechanism
Figure 5: Population level FACs data comparing the relative fluorescence of a pTet GFP and TetR reporter with and without a tetO binding array. While the data is noisy, it is clear that the inflection point of the circuit with the binding array has shifted to the left as expected. Additionally, both circuits experienced a decrease in fluorescence at higher aTC concentrations, which was thought to be a result of ATC causing toxicity at high levels. However, during the course of our modeling we found that in fact this is the expected result of a repressor system.
Figure 6: Plot of predicted absolute fluorescence and observed experimental fluorescence for systems with and without 85x binding array. Reduced fluorescence at high [aTC] is an expected result from our model, as is a reduction in absolute fluorescence. It is likely that an 85x binding array on a high copy backbone shifts the function too far. Model predictions indicate that a smaller number around 5 would be more appropriate. (See modeling page.)
