We propose a negative feedback architecture that regulates activity of artificial genes, or "genelets", to meet their output downstream demand, achieving robustness with respect to uncertain open-loop output production rates. In particular, we consider the case where the outputs of two genelets interact to form a single assembled product. We show with analysis and experiments that negative autoregulation matches the production and demand of the outputs: the magnitude of the regulatory signal is proportional to the "error" between the circuit output concentration and its actual demand. This two-device system is experimentally implemented using in vitro transcriptional networks, where reactions are systematically designed by optimizing nucleic acid sequences with publicly available software packages. We build a predictive ordinary differential equation (ODE) model that captures the dynamics of the system and can be used to numerically assess the scalability of this architecture to larger sets of interconnected genes. Finally, with numerical simulations we contrast our negative autoregulation scheme with a cross-activation architecture, which is less scalable and results in slower response times.

This paper focuses on the regulation of RNA production for in vitro synthetic gene networks. We consider the problem of matching the production rate of several RNA species concurrently transcribed to form RNA complexes. We design regulatory feedback loops based on two schemes: negative auto-regulation (which can minimize the potentially harmful amount of molecules not used to form products) and cross-activation (which can maximize the overall output flux). We analyze numerically the performance of these schemes: in both cases, transcription rate matching can be achieved through proper feedback constants; negative feedback is faster and maintains stability. A possible experimental implementation of a three and four genes negative feedback architecture is also numerically studied.