COMPREHENSIVE TRANSLATIONAL PROFILING AND STE AI UNCOVER RAPID CONTROL OF PROTEIN BIOSYNTHESIS DURING CELL STRESS

Le June 12 2024 at 10h00 Séminaire

Nikolay E. Shirokikh¹, Attila Horvath¹, Yoshika Janapala¹, Katrina Woodward¹, Shafi Mahmud¹, Alice Cleynen², Elizabeth E. Gardiner³, Ross D. Hannan^{1, 4-7}, Eduardo Eyras^1,8, Thomas Preiss^1,9

¹ Division of Genome Sciences and Cancer, The John Curtin School of Medical Research, and The Shine-Dalgarno Centre for RNA Innovation, The Australian National University; Canberra, ACT 2601, Australia

² CNRS, Université de Montpellier, Montpellier, France​

³ Division of Genome Sciences and Cancer, The John Curtin School of Medical Research, and The National Platelet Research and Referral Centre, The Australian National University; Canberra, ACT 2601, Australia

⁴ Department of Biochemistry and Molecular Biology, University of Melbourne; Parkville 3010, Australia

⁵ Peter MacCallum Cancer Centre; Melbourne 3000, Australia

⁶ Department of Biochemistry and Molecular Biology, Monash University; Clayton 3800, Australia

⁷ School of Biomedical Sciences, University of Queensland; St Lucia 4067, Australia

⁸ Division of Genome Sciences and Cancer, The John Curtin School of Medical Research, and The Centre for Computational Biomedical Sciences, The Australian National University; Canberra, ACT 2601, Australia

⁹ Victor Chang Cardiac Research Institute; Darlinghurst, NSW 2010, Australia

Translational control is important in all life but remains a challenge to accurately quantify. When ribosomes translate messenger (m)RNA into proteins, they attach to the mRNA in series, forming poly(ribo)somes, in which the ribosomes can co-localise. Here we propose and, using rapid crosslinking-based enhanced translation complex profile sequencing (eTCP-seq), confirm co-localisation of ribosomes on mRNA resulting from diffusional dynamics.

We further demonstrate that the co-localised ribosomes (such as disomes) can be of a different origin: some are related to translation elongation delays, others are reflective of the polysome spatial arrangement or descend from stochastic molecular events. Together with the other types of translational complexes, co-localised ribosomes contain new rich information about the translational state of mRNA in vivo.

Employing unbiased machine learning to the eTCP-seq data in a novel AI-based Stochastic Translation Efficiency (STE) pipeline, we demonstrate accurate prediction of the absolute translation output from footprints and detect strong (>50-fold) and highly specific translational regulation on certain mRNAs in just under 10 minutes of glucose starvation response in yeast.

STE is the first application of AI to decipher high-throughput footprinting data. STE ranks mRNAs by the ‘power’ of their translation in a single experiment or between conditions. Importantly, STE does not utilise bias-inducing normalisation to the RNA abundance or signals of a different type and relies on self-normalised data. STE AI thus has new applications in translationally dissecting cell states in disease pathophysiology and drug development, and will facilitate the design of next-generation synthetic biology constructs and mRNA-based therapeutics.