Yannick Rondelez (University of Tokyo, Japan)

  Version imprimable de cet article RSS
29 février 2016 11:30 » 12:30 — Bibliothèque PCT - F3.04

In Vitro Models of Cellular Computing

Living cells use webs of chemical reactions organized in precise networks to perform informational tasks. This concept is at the very core of systems biology, but also share links with electronic or computer science. The dynamical and information-processing properties of living cells, i.e. their ability to make decisions, sense their environment, maintain their integrity, memorize bits of information, interact, coordinate, etc. is indeed encapsulated in the topology and dynamics of their molecular networks. I will discuss new methods, based on basic DNA biochemistry, to build artificial analogues of cellular circuits in test tubes, using a simple enzymatic decoding machinery. It uses modular elementary interactions (activation, inhibition, degradation), which can be connected in tightly regulated networks of desired topology. The system is kept out-of-equilibrium using catalytic resolution of kinetic bottlenecks, but could also in principle be embedded in an open system. This approach was initially demonstrated by building de novo and in vitro a robust chemical oscillator : we implemented positive- and a delayed-negative feedback loops, encoded in the sequence of small DNA templates, and obtained the predicted oscillatory dynamics. More recently, we have extended the approach to encode other types of biological networks, such as those involved in complex ecosystems. Also, because of the simple and well-controlled environment, the chemical network is easily amenable to quantitative mathematical analysis so that the approach can be partially automatized. Other features of living systems are being integrated. For example many networks, most notably morphogenic subsystems, use diffusion as a key functional tool, and I will present initial results in this direction. Linearity/non-linearity of the individual interactions is also an essential component of the function, and we have worked to make it programmable as well. Our results show that the rational cascading of standard elements opens the possibility to implement complex behaviours in vitro. These synthetic systems may thus accelerate our understanding of the underlying principle of biological dynamic networks and provide building blocks for the construction of integrated emergent behaviours at the molecular scale.

10 Rue Vauquelin, 75005 Paris