High-throughput 5′ UTR engineering for enhanced protein production in non-viral gene therapies

Abstract

Despite significant clinical progress in cell and gene therapies, maximizing protein expression in order to enhance potency remains a major technical challenge. Here, we develop a high-throughput strategy to design, screen, and optimize 5′ UTRs that enhance protein expression from a strong human cytomegalovirus (CMV) promoter. We first identify naturally occurring 5′ UTRs with high translation efficiencies and use this information with in silico genetic algorithms to generate synthetic 5′ UTRs. A total of ~12,000 5′ UTRs are then screened using a recombinase-mediated integration strategy that greatly enhances the sensitivity of high-throughput screens by eliminating copy number and position effects that limit lentiviral approaches. Using this approach, we identify three synthetic 5′ UTRs that outperform commonly used non-viral gene therapy plasmids in expressing protein payloads. In summary, we demonstrate that high-throughput screening of 5′ UTR libraries with recombinase-mediated integration can identify genetic elements that enhance protein expression, which should have numerous applications for engineered cell and gene therapies.

A Synthetic Transcription Platform for Programmable Gene Expression in Mammalian Cells

“…RT-qPCR data also suggest that an SI in the 5' UTR directly contributes to gene transcription in our crisprTF promoter system, consistent with gene expression enhancement by 5' UTR sequences [52].…”

“…Here, the incorporation of an SI immediately downstream of the operator in the 5’ UTR of the target gene [50] increased gene expression at least 200% with nearly all tested constructs, whether episomal or chromosomal, suggesting a high degree of compatibility between the transcriptional control by the crisprTF promoters and the intron-mediated enhancement by SI [51]. RT-qPCR data also suggest that an SI in the 5’ UTR directly contributes to gene transcription in our crisprTF promoter system, consistent with gene expression enhancement by 5’ UTR sequences [52]. However, overly strong target gene expression may impact cell proliferation ( Figure 5E ).…”

Translation rate prediction and regulatory motif discovery with multi-task learning

“…Highly GC enriched motifs like filter 236 and the binding site of RBM4 repress translation, but filter 197 with A and U shows a stronger repressive effect. More directly, the GC ratio has been included as the feature by Cao et al [11]. The performance gap between their RF model and our RF with MTtrans features also implies the need to refine the sequence signature.…”

“…For ribosome profiling data, three datasets collected from HKE293T cell line [26], PC-3 cell line [27] and muscle tissue [28] were selected because they are the benchmark datasets on which the random forest model with the handcraft features [11] was built. The ribosome profiling data were first pre-processed as in [11]. Later, to prevent data leakage, we filter the isoforms with the same sequence 100nt upstream of the start codon and keep the most abundant isoforms.…”

“…Methods to identify the regulatory code of translation control from these high throughput data evolve quickly. Previous methods required domain knowledge to firstly list out potential elements and tested them with the sequencing data afterward [4, 9, 10, 11]. Recently, it has become a new paradigm in the area of regulatory genomics to use deep learning methods to learn the sequence motifs by the model itself [7, 8, 12, 13, 14, 15].…”