The Base-Editing Enzyme APOBEC3A Catalyzes Cytosine Deamination in RNA with Low Proficiency and High Selectivity

Abstract

Human APOBEC3A (A3A) is a nucleic acid-modifying enzyme that belongs to the cytidine deaminase family. Canonically, A3A catalyzes the deamination of cytosine into uracil in single-stranded DNA, an activity that makes A3A both a critical antiviral defense factor and a useful tool for targeted genome editing. However, mutagenesis by A3A has also been readily detected in both cellular DNA and RNA, activities that have been implicated in cancer. Given the importance of substrate discrimination for the physiological, pathological, and biotechnological activities of A3A, here we explore the mechanistic basis for its preferential targeting of DNA over RNA. Using a chimeric substrate containing a target ribocytidine within an otherwise DNA backbone, we demonstrate that a single hydroxyl at the sugar of the target base acts as a major selectivity determinant for deamination. To assess the contribution of bases neighboring the target cytosine, we show that overall RNA deamination is greatly reduced relative to that of DNA but can be observed when ideal features are present, such as preferred sequence context and secondary structure. A strong dependence on idealized substrate features can also be observed with a mutant of A3A (eA3A, N57G), which has been employed for genome editing due to altered selectivity for DNA over RNA. Altogether, our work reveals a relationship between the overall decreased reactivity of A3A and increased substrate selectivity, and our results hold implications both for characterizing off-target mutagenesis and for engineering optimized DNA deaminases for base-editing technologies.

Structure-Guided Design of a Potent and Specific Inhibitor against the Genomic Mutator APOBEC3A

“…It should be noted that for DB-P , mZ was maintained in a similar 5′-TmZ context to avoid differences in activity due to sequence preference. Given A3A’s strong mutational preference for the target C being located on the 3′-end of hairpin loops, we expected a significant impact on potency. − Placement of the mZ in the interior of the stem eliminated all inhibitory activity, which was expected as A3A does not engage with double-stranded DNA. Replacement of one of the loop mZ in DB-V resulted in a 3.4-fold decrease in inhibitory potency compared to the parent dumbbell DB (Figure b), suggesting that the bivalent dumbbell design enhances inhibition.…”

“…This order of preference differed from that found in DNA substrate structures, where 3 nt loops are favored over longer loops. 33,34 One potential explanation for this contrast in preference may involve differential binding interactions between A3A and 3 or 4 nt loops, where more transient binding with 3 nt loops results in higher turnover in substrates but poorer performance by dumbbell inhibitors. Another possibility lies in that the dumbbell design results in a closed loop, which may constrain the overall hairpin conformation differently compared to an open-ended stem loop found in natural substrates.…”

“…A3A activity, in particular, is driven by substrate features outside of the canonical preference for 5′-TC sequence motifs. Genomic and mRNA tumor data have revealed a strong preference for substrates that take on a hairpin structure where the target cytosine is situated on the 3′-end of the loop, and this preference appears significantly diminished for A3B. − Modulating specific features of the hairpin, such as stem strength, loop size, and cytosine position, is highly correlated to A3A deamination activity and largely shared between preferred ssDNA and RNA substrates. − The mechanistic basis for such selectivity was revealed by the crystal structures of A3A complexed with a nonstructured ssDNA substrate. The active site of A3A engages with the ssDNA bent into a U-shaped conformation, suggesting that hairpin structures could make up for this entropic cost of DNA bending by prepositioning the target base in an optimal configuration (Figure c). , …”

The cytidine deaminase APOBEC3A is required for large ribosomal subunit biogenesis

“…Unexpectedly, we did not observe robust enrichment of the UC dinucleotide preferred by APOBEC3A for nuclear pre-mRNA (intron-enriched) editing. More recent work has highlighted, in addition to the UC motif, the presence of a hairpin secondary structure in APOBEC3A edit sites on DNA ( 87 ) and RNA ( 88 ). This may also contribute to APOBEC3A’s preference in the nuclear RNA sites we have identified.…”

“…It is possible that it is APOBEC3A’s ssDNA editing function that is leading to downstream changes that we observe as RNA edits. APOBEC3A prefers ssDNA over RNA in vitro ( 88 ). Although it is lowly expressed in MCF10A cells, the addition of siRNAs could induce its expression [as suggested by our results in MCF10A cells (Fig.…”