Team:Peking R/Project/Application/AG

Fine-tuning Logic Gate Using Soft-coding of Genetic Program

Screening for well performing devices usually requires laborious, time-consuming refinement cycles, especially in the case of information processing devices. Utilizing the platform introduced previously, we aim at making tuning devices fast, affordable and more predictable. To demonstrate the versatility and validity of the platform, it was firstly applied to modulating a genetic AND gate. We chose the AND gate as a proof of concept primarily for two reasons:
(1) It presents biologically significant functions in information processing, e.g., integration of environmental information to generate appropriate responses in microorganisms.
(2) Engineering and screening a synthetic AND gate module involves laborious characterization when using conventional methods, which would best contrast with the advantages of our platform’s methodology.

The AND gate we utilized was designed by Anderson and his colleagues and consequently standardly redesigned by PKU_Beijing 09 team (BBa_K228258, Fig. 1A).

Figure 1
A. The standardly redesigned AND gate of PKU_Beijing 09 team.
B. The topology and mechanism of AND gate. Parts showing in box constitute the core processing module.

As shown in Fig. 1B, pBAD promoter serves as input 1. It is activated by arabinose, leading to the expression of SupD, an amber suppressor tRNA. The input 2 is pSal promoter activated by salicylate via binding to NahR. It regulates the expression of a T7 polymerase coding sequence with two amber codons reside in the position 8 and 14 (T7ptag), resulting in premature translational termination. When both T7ptag and SupD tRNA genes are expressed, functional T7 polymerase would be synthesized and consequently active output, T7 promoter. A green fluorescent protein (GFP) generator regulated by T7 promoter serves as a reporter.

It has been demonstrated that translation strength of T7ptag is a major determinant of AND gate quality. Mutagenesis of RBS leads to various performance, thus we modeled the AND gate using transfer function, as further illustrated by phase diagrams. The phase diagram displays the system’s output properties in response to all possible input combinations (Fig 2). In the case of AND gate, two inputs and one output are involved. Inputs are the expression strength of pSal and pBAD induced by different concentration of the two inducers, salicylate and arabinose, respectively, and the output is the fluorescence intensity of reporter GFP. To construct such a phase diagram, the input promoter strengths are measured by GFP fluorescence in response to varying inducer concentrations. We also normalized output fluorescence intensity values to the maximum possible fluorescence determined by model parameters. Assuming that the output is measured at the system’s steady state, we derived the output function using ordinary differential equations (ODEs):

Where F is an arbitrary, unitless measure for output fluorescence and Fmax is the highest possible level of output determined by parameters in ODEs describing the process of gene expression and regulation. I1 and I2 are the inputs in units of fluorescence (au). a is a parameter reciprocally proportional to the translation strength of the T7ptag coding region, and b merely results from a scaling factor to unify units of measurement.
(The detailed construction of each ODE describing transcription and translation processes and integrating input values into the output function can be found in Appendices.)

The phase diagrams were obtained by plotting the output function as a color map over the 2-dimensional space with axes corresponding to the two inputs, I1 and I2. The variables’ space is extended to cover a sufficiently large range of inputs to represent a general AND gate module (Fig. 2(a)). In reality, the two inputs only vary within a limited range that is a sub-region of the I1-I2 space. For any particular AND gate module, variations in the translation strength of the T7ptag gene can be reflected in changes in the range of the input I2, resulting in shifting of rectangular region in the phase diagram that corresponds to the range of inputs.

The module in which translation strength of T7ptag gene is strong would be a poor AND gate because within its range of inputs, the “ON” state (high output) can be induced by addition of arabinose (I2) alone even when I1 is low. This is probably due to leakage of the Psal promoter, i.e., the promoter has a relatively high basal activation level.

The translational strength of T7ptag is dominantly regulated by RBS strength when the promoter strength is the same. Several AND gates with different RBSs, i.e., different T7ptag translation strengths in this condition, are characterized, of which experimental data fit the model prediction well.

(a)

(b)
Figure 2
(a)The output properties in response to all possible input combinations. AND gate performance is influenced by translation strength of T7ptag. Color bar on the right indicates the output strength corresponding to each color. Rectangular regions represent the largest possible range of the two inputs in experimental conditions (right to left): AND gate with progressively lower translation rates (more positive △G) of T7ptag gene (white, yellow, green and pink).
(b)Experimental result of the AND gate behaviour in response to changes in translation rate. Output is plotted as a function of the two inducer concentrations, [arabinose] and [salicylate]. Its values have been normalized to the maximal fluorescence measured, facilitating comparison with the theoretical phase diagrams. (Left)An AND gate module with an RBS sequence which has a high translation rate(△G=-7.07kJ/mol). (Right)AND gate module with a weaker RBS(decreased translation rate, △G=-0.49kJ/mol). It can be seen that changing translation rate result in different AND gate performance.

In order to optimize the performance of AND gate, the first step is to determine the optimal translation strength of the T7ptag gene using our RNA controller toolkit. By placing a ligand-responsive RNA controller element upstream of the coding sequence, we obtained an AND gate modulator whose T7ptag gene translation rate varies in response to ligand concentration (Fig. 3A). By optimizing the strength of translation, we are able to make up for the leakage in transcription and a translation rate that endows the AND gate with satisfactory performance (Fig 3B).

Figure 3 Optimization of AND gate performance using RNA controller(TPP ribozyme).
(a) Output fluorescence of the AND gate device without addition of TPP ligand(corresponding to a △G of -5.78kJ/mol). (b)Output fluorescence of the AND gate device with addition of maximal concentration of TPP ligand(1μM，corresponding to a △G of -3.38kJ/mol). Vertical and horizontal axes indicate logarithm of the concentrations of arabinose and salicylate respectively. Apparently, addition of TPP ligand(which attenuates translation strength) improved the AND gate performance by decreasing the area of region for “ON” state. The two output color plots are mapped to their corresponding positions in the full phase diagram in Figure 2, showing that they display fair agreement with modeling results(white and yellow rectangular respectively).

The translation rates are expressed in terms of the parameter a, whose values can be determined using the a value corresponding to the original sequence without the RNA controller (a can be converted to a more direct indicator of translation strength - △G using a quantitative relationship derived in the Appendix). A dose(ligand concentration) response curve used for finding the a value corresponding to a given TPP concentration is given in Figure 4.

 
Figure 4 Correlation of parameter a with ligand (TPP) concentration.
It is shown that a increases(i.e., translation rate decreases) as TPP concentration increases, in agreement with the function of TPP ribozyme(inhibiting translation via ligand-responsive cleavage). Notably, the dose response curve displayed saturation at TPP concentrations above 0.3μM.

The final step is to replace the RNA controller together with the RBS sequence downstream of the controller with an automatically designed RBS sequence that meets the translation strength configuration determined by our RBS calculator, thus to ‘fix’ the strength in genetic program, producing desired performance.

As illustrated above, minor differences in RBS strength may largely influence the performance of genetic program, consistent with the fact that tuning genetic circuits requires laborious and refinement cycles, especially in the case of complex systems. However, when applying our platform to the modulation of the AND gate, the whole process does not require construction of mutation library while still allowing for high-throughput screening for an optimal translation strength. Additionally, our platform enables its user to quantitatively analyze the correlation between system behavior and translation strength of core genetic component, thus to predictably manipulate the performance of genetic devices. In brief, soft-coding of genetic program makes synthetic biology fast, affordable and more predictable.

(To validate the reliability of parameter a as the single variable used in the model, we plotted experimental measurements of the AND gate output under each input combination against varying a values (reflecting ligand concentration), and theoretical curves were fitted to the experimental results to test the robustness of parameters under different input combinations. For more detailed description of this part, see section 3 in the Appendix.)

Reference:

Anderson, J. C. et al. (2007) Environmental signal integration by a modular AND gate. Nature Molecular Systems Biology 3; Article number 133; doi:10.1038/msb4100173

[TOP]