Development and characterization of chemical modifications for siRNA strand selection

Abstract

Small interfering RNA (siRNA) are non-coding, double-stranded RNA that enable efficient gene silencing through RNA interference (RNAi) mechanisms. The primary effector of RNAi is an RNA-guided nuclease, Ago2, which selects one of the two strands as a guide to form the RNA induced silencing complex (RISC). Incorporation of the antisense, rather than the sense, strand is crucial for potency and safety siRNA therapeutics. Thermodynamic stability, phosphorylation status, and nucleotide sequence play a role in canonical strand selection; however, chemical modifications can improve strand selection and activity. To evaluate the impact of novel chemical modifications on strand activity, an RT-qPCR-based target cleavage assay was developed and validated. Dose-response analysis highlighted the consequence of overdosing, as dosages over 8 pM (for our siRNA sequence) resulted in significant sense strand activity without any on-target benefit. This assay was then used to characterize strand activity for siRNAs carrying azobenzene linkers, propargyl linkers, triazole-bound folate linkers, cubane linkers, hydrophobic phosphotriester modifications, or a 5’ phosphorofluoridate. Antisense vs. sense strand profiles revealed that nucleic acid substitutions (linkers) tend to reduce overall potency, likely due to an altered helical conformation and impaired recognition, but can effectively limit sense strand uptake when placed within the central region. These non-nucleotidyl linkers lack key hydrogen bond interactions between the guide RNA and both Ago2 and target mRNA, which are important for orienting the mRNA for cleavage. Surprisingly, placing a folic acid linker at the sense 3’ end was highly effective at limiting 3’ activity. Hydrophobic phosphate triester modifications have been shown to facilitate carrier-free uptake and are well tolerated. Profiles revealed that small phenethyl linkers had a limited impact of strand activity while larger octadecyloxy hydrophobic tails were more deactivating, likely due to steric constraints. Lastly, phosphorofluoridate analogues were also developed for their first introduction to siRNA. They exhibited a limited impact on strand activity. This may be the result of enzymatic hydrolysis and replacement with a natural phosphate, or efficient uptake of the modification within Ago2.