Sebastian Castillo-Hair, Stephen Fedak, Ban Wang, Johannes Linder, Kyle Havens et al.
mRNA therapeutics are revolutionizing the pharmaceutical industry, but methods to optimize the primary sequence for increased expression are still lacking. Here, we design 5’UTRs for efficient mRNA translation using deep learning. We perform polysome profiling of fully or partially randomized 5’UTR libraries in three cell types and find that UTR performance is highly correlated across cell types. We train models on our datasets and use them to guide the design of high-performing 5’UTRs using gradient descent and generative neural networks. We experimentally test designed 5’UTRs with mRNA encoding megaTALTM gene editing enzymes for two different gene targets and in two different cell lines. We find that the designed 5’UTRs support strong gene editing activity. Editing efficiency is correlated between cell types and gene targets, although the best performing UTR was specific to one cargo and cell type. Our results highlight the potential of model-based sequence design for mRNA therapeutics.
Sebastian Castillo-Hair, Stephen Fedak, Ban Wang, Johannes Linder, Kyle Havens, Michael Certo, Georg Seelig. Optimizing 5’UTRs for mRNA-delivered gene editing using deep learning. 2024-06-20. https://doi.org/10.1038/s41467-024-49508-2
A pooled in-vitro-transcribed EGFP reporter library with a constant 25-nt 5′ prefix and a fully randomized 50-nt 5′UTR segment was profiled in HEK293T cells. Two biological replicates were fractionated by polysome load, and sequence-level mean ribosome load (MRL) was calculated from UMI counts.
A HEK293T IVT EGFP reporter library with a 50-nt fully randomized 5′UTR and only the T7-compatible 5′ guanine prefix was profiled in one biological replicate. UTR-level MRL values were retained with total UMI coverage after polysome fractionation.
The fixed-end N50 IVT EGFP reporter library was transfected into activated human T cells enriched from healthy-donor PBMCs. Two biological replicates were fractionated by polysome load and summarized as sequence-level MRL values from UMI counts.
A fixed-end N50 IVT EGFP reporter library with a constant 25-nt 5′ prefix was assayed in HepG2 cells. One biological replicate was fractionated by polysome load and summarized as UTR-level MRL from UMI counts.
A HEK293T IVT EGFP reporter library with a 25-nt fully randomized 5′UTR and only the T7-compatible 5′ guanine prefix was profiled in two biological replicates. UTR-level MRL values were combined with replicate-total UMI coverage after polysome fractionation.
Original supplemental and deposited inputs retained for this study. Download files individually or together as a ZIP; nested folders are preserved. Source reuse terms apply, and sequencing reads may be omitted.
Download all 12 files (ZIP)41467_2024_49508_MOESM4_ESM.xlsx41467_2024_49508_MOESM5_ESM.csvGSE114002_family.soft.gzGSE232927_family.soft.gzGSE232927_processed_defined_end_hepg2_r1.csv.gzGSE232927_processed_defined_end_tcell_r1.csv.gzGSE232927_processed_defined_end_tcell_r2.csv.gzGSE232927_processed_random_end_hek293t_N25_r1.csv.gzGSE232927_processed_random_end_hek293t_N25_r2.csv.gzGSE232927_processed_random_end_hek293t_N50_r1.csv.gzGSM3130435_egfp_unmod_1.csv.gzGSM3130436_egfp_unmod_2.csv.gz