Human tRNA genes enable differential nonsense suppression in live cells and a model of frontotemporal dementia

Nonsense mutations generate premature termination codons (PTCs) that are responsible for 11% of human genetic disease alleles. Thus, nonsense suppressor tRNAs have broad therapeutic potential. Humans encode > 600 tRNA genes with many identical or similar copies of each tRNA. We hypothesized that tRNA gene variants will enable differential nonsense suppression and developed a dual fluorescent reporter to measure suppression in live cells. The arginine (CGA) to stop (UGA) mutation is the most common PTC, and an Arg nonsense suppressor tRNAArg (G36A) is found in 0.01% of human genomes in the tRNAArg TCG-6-1 gene. In multiple mammalian cell lines, we found tRNAArgUCA promotes readthrough levels that depend on the sequence of the tRNA gene and the cell type. We tested G36A variants of all six human tRNAArgUCG isodecoders, and only the TCG-6-1 gene was unable to translate nonsense codons. With tRNA sequencing, we showed that a suppressor tRNA derived from the TCG-3-1 gene was expressed 2-fold higher and generated 3-fold more nonsense suppression than a tRNA derived from the TCG-4-1 gene. In a neuroblastoma cell model of frontotemporal dementia (FTD), we observed up to 60% readthrough of the progranulin R493X allele with a human tRNAArg suppressor. The tRNAs outperformed an aminoglycoside nonsense suppression drug (G418) in efficacy, tolerability to the cells, and translation fidelity according to mass spectrometry. Our studies show that nonsense suppressor tRNAs can correct genetic defects that cause disease.