Rational design of modular circuits for gene transcription: A test of the bottom-up approach

Abstract

Background: Most of synthetic circuits developed so far have been designed by an ad hoc approach, using a small number of components (i.e. LacI, TetR) and a trial and error strategy. We are at the point where an increasing number of modular, inter-changeable and well-characterized components is needed to expand the construction of synthetic devices and to allow a rational approach to the design.

Results: We used interchangeable modular biological parts to create a set of novel synthetic devices for controlling gene transcription, and we developed a mathematical model of the modular circuits. Model parameters were identified by experimental measurements from a subset of modular combinations. The model revealed an unexpected feature of the lactose repressor system, i.e. a residual binding affinity for the operator site by induced lactose repressor molecules. Once this residual affinity was taken into account, the model properly reproduced the experimental data from the training set. The parameters identified in the training set allowed the prediction of the behavior of networks not included in the identification procedure.

Conclusions: This study provides new quantitative evidences that the use of independent and well-characterized biological parts and mathematical modeling, what is called a bottom-up approach to the construction of gene networks, can allow the design of new and different devices re-using the same modular parts.

Figure 1

Gene circuits. Repressor (A) and Reporter (B) plasmids. Boxes highlight the parts that could be used as independent elements to tune the circuit outcome in a modular framework. As a proof of principle of the modular architecture, we created nine different gene circuits, combining three different operator sequences, O₁, O₂ and O_s, in two plasmids. (C) Gene circuit in cells transformed with both the plasmids. (D) Sequences of the three operator sites. Bold characters are used for the central G-C base pair in the naturally occurring operator sites O₁ and O₂. The left and right half operator sites in O₁ and O₂ are pseudo-symmetric, lower-case characters highlight the deviations from perfect symmetry. The base pairs that are different between the operators O₁ and O₂ are underlined in the O₂ sequence. The artificial operator site O_s was defined removing the central base pair G-C and creating a perfectly symmetric sequence based on the left half of the operator O₁. The affinity to lactose repressor of the three operator sites O₁, O₂ and O_s spans a range of over two orders of magnitude [27,28], which allows the analysis of the system behavior under wide changes of the parameters values.

Figure 2

Fluorescence levels. The normalized fluorescence, equation 2, in arbitrary units, is shown for cells transformed with: Reporter plasmid without the operator sequence (n = 6); Reporter plasmids with the O₁ (n = 5), O₂ (n = 8), and O_s (n = 10) operator sequence; Repressor plasmids with any of the three operator sequences and Reporter plasmid lacking the lactose operator sequence (n = 10); O₂O₂ induced with 1 mM IPTG (n = 13); O₁O₁ induced with 1 mM (n = 9) and 2 mM (n = 5) IPTG; O_sO_s induced with 1 mM (n = 4) and 8 mM (n = 3) IPTG. In absence of Repressor plasmids, the normalized fluorescence is the same if the Reporter plasmid does not include the operator sequence, or if it includes any of the three operator sequences tested. Co-transformation with a Repressor plasmid and a Reporter plasmid without the operator site causes a decrease in the maximum fluorescence, even if no LacI activity is exerted. The same fluorescence was observed in the O₂O₂ gene circuit at saturating concentrations of IPTG. On the other hand, the maximum fluorescence in the O₁O₁ and O_sO_s gene circuits is statistically different from the maximum value.

Figure 3

Percentage fluorescence in absence of IPTG (A), and dose-response curves for the gene-circuits O₂O₂ (B), O₁O₁ (C), and O_sO_s (D). In panel (A) grey bars with standard deviations are used for the experimental data while black squares are used for the simulated responses. In panels (C) and (D), a continuous line is used for the simulated dose-response curves in presence of a residual affinity for the operator sites by induced LacI molecules, while a dotted line is used for the simulated dose-response curves when this residual affinity is turned off. The y-axis is in logarithmic scale in panel (A). The x-axis is in logarithmic scale in panels (B) to (D). All the measurements were repeated at least five times.

Figure 4

Dose-response curves for gene circuits with different operator sequences on the Reporter and the Repressor plasmids. A continuous line is used for the simulated dose-response curves in presence of a residual affinity for the operator sites by induced LacI molecules, while a dotted line is used for the simulated dose-response curves when this residual affinity is turned off. All the measurements were repeated at least five times.