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

Abstract

Nucleic acid structure plays a critical role in governing the selectivity of DNA- and RNA-modifying enzymes. In the case of the APOBEC3 family of cytidine deaminases, these enzymes catalyze the conversion of cytosine (C) to uracil (U) in single-stranded DNA, primarily in the context of innate immunity. DNA deamination can also have pathological consequences, accelerating the evolution of viral genomes or, when the host genome is targeted by either APOBEC3A (A3A) or APOBEC3B (A3B), promoting tumor evolution leading to worse patient prognosis and chemotherapeutic resistance. For A3A, nucleic acid secondary structure has emerged as a critical determinant of substrate targeting, with a predilection for DNA that can form stem loop hairpins. Here, we report the development of a specific nanomolar-level, nucleic acid-based inhibitor of A3A. Our strategy relies on embedding the nucleobase 5-methylzebularine, a mechanism-based inhibitor, into a DNA dumbbell structure, which mimics the ideal substrate secondary structure for A3A. Structure-activity relationship studies using a panel of diverse inhibitors reveal a critical role for the stem and position of the inhibitor moiety in achieving potent inhibition. Moreover, we demonstrate that DNA dumbbell inhibitors, but not nonstructured inhibitors, show specificity against A3A relative to the closely related catalytic domain of A3B. Overall, our work demonstrates the feasibility of leveraging secondary structural preferences in inhibitor design, offering a blueprint for further development of modulators of DNA-modifying enzymes and potential therapeutics to circumvent APOBEC-driven viral and tumor evolution.

Figure 1.

Purposeful and pathological mutagenesis by APOBEC3A (A3A). (a) Schematic representation of A3A-mediated purposeful mutation of viral and cancer genomes. In the setting of viral infection (top), A3A is upregulated and DNA deaminases can target viral genome replication intermediates, restricting replication. Sublethal mutation can result in the evolution of new, more fit variants. Within cancer cells (bottom), dysregulated A3A targets the host genome and deaminates exposed cytosines during events of DNA unwinding. Hypermutation results in extensive genomic damage and apoptosis, while hypomutation contributes to tumor adaptation and progression. (b) Mechanism-based inhibition of A3A can be achieved with 4-desaminocytosine analogs, including 5-methylzebularine (mZ). (c) Co-crystal structure of human A3A (green) in complex with a ssDNA (yellow) substrate (PDB: 5SWW). The target cytosine (pink) is flipped out into the zinc-coordinating (grey) active site as the rest of the ssDNA backbone takes on a U-shaped binding conformation. (d) Representation of a preferred A3A ssDNA hairpin substrate and a mZ-containing oligonucleotide which form the basis for an idealized DNA dumbbell inhibitor design.

Figure 2.

Methylzebularine-containing DNA dumbbells are nanomolar-level inhibitors of A3A. (a) Schematic representation of a fluorescence-based in vitro deamination assay. Co-incubation of a 50-mer, FAM-labelled ssDNA substrate (S, 500 nM) containing a single 5ʹ-TC site with purified A3A (4 nM) results in deamination of cytosine (C) to uracil (U), followed by the production of an abasic site by excess UDG. The product strand then undergoes alkali-catalyzed strand scission to yield a 35-mer (P), which can be resolved from the substrate by FLD-HPLC. (b) Representative HPLC spectra of A3A deamination inhibition by either nonstructured (NS) or DNA dumbbell (DB) inhibitor oligonucleotides. (c) IC₅₀ curves display differences in A3A inhibition by NS, DB or CS mZ oligonucleotides under conditions where [I] = 640 pM - 2 μM. Each data point represents means of three independent replicates with standard deviation shown (absence of error bars indicates error is smaller than width of symbol).

Figure 3.

Structure-activity relationship between DNA dumbbell stem loop features and inhibition of A3A activity. Representative IC₅₀ curves display differences in A3A inhibition based on conditions where [I] = 640 pM - 2 μM. Results with systematic variation in (a) loop size, (b) stem strength, and (c) stem length are shown. Each data point represents mean of three independent replicates with standard deviation shown (absence of error bars indicates error is smaller than width of symbol).

Figure 4.

Structure-activity relationship between nucleobase moiety features and inhibition of A3A activity. Representative IC₅₀ curves display differences in A3A inhibition based on conditions where [I] = 640 pM - 2 μM. Results with systematic variation in (a) identity of base at target position or (b) location of the mZ inhibitor moiety are shown. Each data point represents mean of three independent replicates with standard deviation shown (absence of error bars indicates error is smaller than width of symbol). (c) Schematic illustrations of the features of ideal DNA substrates and ideal DNA dumbbell inhibitors, highlighting key similarities and differences derived from genome-wide analysis, mutations, or systematic SAR studies, respectively.

Figure 5.

Inhibitory DNA dumbbells show specificity towards A3A. (a) Schematic representation of amino acid homology between A3A and A3B C-terminal domain (A3B_CTD). Similar residues are in blue and differences in red, with high overall similarity (91%) noted. Structural elements are labeled and numbered with α-helices in red, β-strands in blue, and intervening loops noted (L). (b) Differential inhibition of A3A and A3B_CTD by nonstructured (NS) or dumbbell (DB) mZ-containing oligos. Lanes 1–2: Oligonucleotide controls with 5ʹ-TC or 5ʹ-TU (without A3A). [MBP-A3A-His] = 10 nM, [MBP-A3B_CTD-His] = 560 nM, [S] = 500 nM, [I] = 2 μM. S: intact 50-mer substrate, P: cleaved 35-mer product. (c) IC₅₀ curves display differences in inhibition of full-length A3A, A3B, and A3F in HEK293T cell lysates by nonstructured (NS) or dumbbell (DB) mZ-containing oligos where [I] = 640 pM - 2 μM. Each data point represents means of three independent replicates with standard deviation shown (absence of error bars indicates error is smaller than width of symbol).