A cell-free system is an in vitro platform that reconstitutes essential biological processes such as transcription and translation using cellular machineries supplemented with a defined reaction mixture (buffer) and a DNA or RNA template.  Two main types of cell-free systems are defined depending on the origin of these machineries: lysate-based systems (CFS) and those composed of purified components. Compared to living cells, cell-free systems offer major advantages, including their suitability for high-throughput experimentation, openness allowing precise control over biochemical conditions (pH, metabolite concentrations, etc.), and the decoupling of expression from growth. These features have made CFS powerful tools for protein production (particularly for toxic or difficult-to-purify proteins) as well as for synthetic circuit prototyping, biosensing, artificial cell construction, and education. They also serve as valuable model systems for studying fundamental biological mechanisms, as shown by Nirenberg and Matthaei’s pioneering work in deciphering the genetic code in CFS in the 1960s.

Gene regulation is a complex process that enables organisms to adapt to fluctuating environments. It is typically investigated in vivo using direct RNA sequencing (DRS), one of the most informative transcriptomic approaches currently available. Adapting DRS to CFS could overcome key in vivo limitations, such as the restricted control of intracellular conditions and the requirement that the organism of interest is cultivable. It highlights the relevance of CFS as tools for studying living systems and raises the question of their potential for deciphering gene regulation.

The aim of this thesis was to develop a proof of concept for cell-free transcriptomics. This involved two main steps: (1) optimizing a CFS to produce sufficient RNA for DRS analyses, and (2) expressing the genome of a model organism to assess the biological relevance of in vitro transcriptional data. We focused on optimizing the buffer composition of a standard CFS based on Escherichia coli lysate containing the T7 RNA polymerase (T7 RNAP). To this end, we implemented an automated active learning cycle using a dual fluorescent reporter plasmid to simultaneously quantify ARNm and protein production across a wide range of buffer compositions. Analysis of 653 CFS reactions identified an optimized buffer producing 20-fold more ARNm than the reference buffer. Expressing the T7 phage genome in this optimized system enabled us to compare transcriptional profiles obtained in vitro with those measured in vivo and in an in vitro minimal system with purified T7 RNAP. Whereas the minimal system only reflects transcription efficiency by T7 RNAP, CFS reveals a balance between production and degradation, along with maturation of T7 RNAs likely resulting from RNase III activity.  A mechanism not recapitulated in CFS was the 3′-end RNA degradation observed in vivo, likely due to the inefficiency of degradation in the lysate. Together, production, degradation and maturation form a gradient of biological complexity allowing the dissection of key transcriptional processes. Preliminary experiments have also explored extending this approach to bacterial genomes by developing CFS with the endogenous RNA polymerase of the organism of interest.

Taken together, this work establishes the foundation for in vitro transcriptomics applicable to both phage and bacterial genomes using endogenous RNA polymerases, providing a new high-throughput framework for studying bacterial gene regulation. Cell-free transcriptomics could enable the exploration of transcriptional landscapes of non-cultivable bacteria, which remain poorly characterized due to limited access to their RNA.