Pré-Publication, Document De Travail Année : 2025

Cell-free genome-wide transcriptomics through machine learning optimization

Résumé

Despite advances in transcriptomics, understanding of genome regulation remains limited by the complex interactions within living cells. To address this, we performed cell-free transcriptomics by developing a platform using an active learning workflow to explore over 1,000,000 buffer conditions. This enabled us to identify a buffer that increased mRNA yield by 20-fold, enabling cell-free transcriptomics. By employing increasingly complex conditions, our approach untangles the regulatory layers controlling genome expression. Main

Lysate-based cell-free systems (CFS) are valuable platform for studying the living and elucidating molecular mechanisms 1-3 . Transcription and translation are restored in bacterial lysates by supplementing with a buffer containing ATP regeneration substrates, crowding agents, salts, and essential precursors such as nucleotides and amino acids. As such, CFS are simpler than living cells in terms of molecular interactions as they eliminate complex and interfering biological processes such as cell division and membrane-related functions. For example, previous experiments showed that production and metabolic burdens can be untangled using cell-free compared with in vivo measurements 4 . CFS provide an open platform ideally suited for protein production, highthroughput prototyping of genetic circuits and parts, biosensing biomanufacturing and gene expression analysis and modelling 5-7 . Genome expression in living cells is routinely analyzed using RNA-seq 8 ; however, low transcript yields in CFS have so far limited the application of this technique in vitro 9 . Here, we developed a general methodological pipeline that integrates active learning loops with high-throughput experimentation to optimize mRNA production in Escherichia coli BL21(DE3) CFS. This approach enabled transcript abundances sufficient for RNA-seq-based, genome-wide transcriptomic analysis. As proof of concept, we applied the pipeline to enhance RNA synthesis by T7 RNA polymerase (T7 RNAP) and performed in vitro transcriptomic profiling of the phage T7 genome.

Fichier principal
Vignette du fichier
biorxiv.full.pdf (806.71 Ko) Télécharger le fichier

Dates et versions

hal-05170246 , version 1 (18-07-2025)

Licence

Identifiants

Citer

Léa Wagner, An Hoang, Olivier Rue, Olivier Delumeau, Valentin Loux, et al.. Cell-free genome-wide transcriptomics through machine learning optimization. 2025. ⟨hal-05170246⟩
2346 Consultations
261 Téléchargements

Altmetric

Partager

  • More