Programming an in vitro DNA oscillator using a molecular networking strategy

Abstract

Living organisms perform and control complex behaviours by using webs of chemical reactions organized in precise networks. This powerful system concept, which is at the very core of biology, has recently become a new foundation for bioengineering. Remarkably, however, it is still extremely difficult to rationally create such network architectures in artificial, non-living and well-controlled settings. We introduce here a method for such a purpose, on the basis of standard DNA biochemistry. This approach is demonstrated by assembling de novo an efficient chemical oscillator: we encode the wiring of the corresponding network in the sequence of small DNA templates and obtain the predicted dynamics. Our results show that the rational cascading of standard elements opens the possibility to implement complex behaviours in vitro. Because of the simple and well-controlled environment, the corresponding chemical network is easily amenable to quantitative mathematical analysis. These synthetic systems may thus accelerate our understanding of the underlying principles of biological dynamic modules.

Figure 1

Experimental assembly. All the reactions shown were performed at 38.5°C, initiated with 0.1 nM α and monitored (ex. 490 nM; em. 510 nM) using EvaGreen-induced fluorescence. (A) One-node positive-feedback loop (autocatalytic module). In the presence of Bst Polymerase (80 U ml⁻¹) and nicking enzyme Nt.bstNBI (200 U ml⁻¹), template T₁ (60 nM) performs an exponential amplification of its input α. The fluorescence reaches a plateau when the template gets saturated with α. The low subsequent increase is due to the accumulation of single-stranded α, weakly fluorescent in these conditions. In the presence of exonuclease RecJ_f (30 U ml⁻¹), the reaction reaches a flat steady state instead. (B) Inhibited amplification. Increasing amounts of inhibitor (from 0 to 1 eq. of T₁) decrease the amplification rate of the previous system (−RecJ_f). (C) Oscillator. Production of Inh is connected to the presence of α as in Figure 1F. This three-templates (T₁ and T₃: 30 nM; T₂: 5 nM) three-enzymes (Bst, Nt.BstNBI, RecJ_f) system produces sustained fluorescent oscillations with a period of 100 min, in good agreement with the predicted evolution of the total concentration of base pairs (D).

Figure 2

Measurement and simulation. (A) Schematic representation of the reactions included in the kinetic model. For clarity, the oligomers α, β and Inh have been omitted when not bound to a template. (B) Fluorescence signal (top panel, dashed line) and discrete concentration measurements of total individual oligomers α, β and Inh taken between t=235 and 520 min (Supplementary information S3). The plots show the mean of three replicates ±s.d. The solid curves are the best simultaneous agreement (‘optimized model') for fluorescence signal and oligomer concentrations, respectively (Supplementary information S5).

Figure 3

Phase diagram in the [T₁]–[T₃] 2D space. The red curve shows the computed border, within which stable oscillations are expected. Each frame shows the fluorescence (a.u.) from t=0 to 1800 min, at [T₁] and [T₃] corresponding to the position of the star in the matrix. Blue line: experiments; green line: genuine predictions made with the set of parameters of the ‘optimized model' (a constant value was added to avoid overlap of the two lines). Green, blue and red frames correspond to sustained, damped or non-oscillating experiments, respectively.