Modeling the Embrace of a Mutator: APOBEC Selection of Nucleic Acid Ligands

Abstract

The 11-member APOBEC (apolipoprotein B mRNA editing catalytic polypeptide-like) family of zinc-dependent cytidine deaminases bind to RNA and single-stranded DNA (ssDNA) and, in specific contexts, modify select (deoxy)cytidines to (deoxy)uridines. In this review, we describe advances made through high-resolution co-crystal structures of APOBECs bound to mono- or oligonucleotides that reveal potential substrate-specific binding sites at the active site and non-sequence-specific nucleic acid binding sites distal to the active site. We also discuss the effect of APOBEC oligomerization on functionality. Future structural studies will need to address how ssDNA binding away from the active site may enhance catalysis and the mechanism by which RNA binding may modulate catalytic activity on ssDNA.

Keywords: APOBEC; RNA and DNA binding; RNA editing; cancer; crystal structure; cytidine deaminase; gene editing; innate and adaptive immunity; mutation; structural biology; zinc-dependent deaminase..

Figure 1

Key Figure: Variation of the Cytidine Deaminase Fold Drives Diversity of APOBEC Function (A) APOBEC3A NMR solution structure (PDB ID 2M65) as a representative APOBEC structure depicting the canonical cytidine deaminase (CD) fold common among all family members is displayed as a ribbon diagram representation of the lowest energy conformer of the ensemble and depicts the conserved structural elements of the canonical CD fold. These include a five-stranded β-sheet flanked by six α-helices. β-strands are shown in gray, while α-helices are shown in aquamarine, and intervening loops are colored green. The catalytic zinc ion is depicted as a purple sphere. Sidechains of the zinc-coordinating residues are depicted in orange while sidechains of the catalytic glutamic acid are depicted in red. The ‘β2-bulge’ is a disruption of the β2-strand observed in some APOBEC structures (A3A, A3B catalytic domain, and both the catalytic and noncatalytic domains of A3G), the function of which is presently unknown. (B) Structural alignment of unliganded single domain APOBEC proteins, AID, A2, A3A, A3C, and A3H and catalytic domains of dual-domain APOBEC proteins, A3B and A3G illustrates the similarities of the canonical APOBEC fold and differences in length and conformation of Loops 1, 3, 5, and 7. These loops are adjacent to the catalytic pocket and are critical for selection, regulation, and direct binding of substrate; structural elements are colored as in (A). (C) Variation of the common CD fold drives the diversity of regulatory mechanisms that ultimately define the molecular function of each APOBEC family member. Variation of Loops 1, 3, 5, and 7, poised just above the deep zinc-centered catalytic pocket, drive selection of DNA or RNA as substrate (or both) and the selection of substrate based upon neighboring nucleotide sequences. These loops, along with other nearby surface features, create unique protein surface channels that select for specific nucleic acid conformation as one mechanism for regulating target selection. Some A3 family members have evolved tandem deaminase domain structures wherein the C terminal domain is catalytically active, the N terminal domain is catalytically inactive, but both domains have crucial regulatory function through both RNA and DNA binding. Many APOBECs oligomerize to form functional dimers or tetramers, while others function as monomers. Complicating matters, oligomerization may occur through either protein–protein interactions or may be nucleic acid dependent. For DNA-deaminating APOBECs, RNA-driven oligomerization of DNA-deaminating APOBECs leads to catalytically inactive complexes, but RNA binding can also drive subcellular and viral localization. Conversely, DNA binding at regions distal to the catalytic site or on the noncatalytic domain of dual-deaminase domain APOBECs has been shown to enhance substrate binding and catalytic turnover. RNP, Ribonucleoprotein.

Figure 2

APOBEC Interactions with Substrate and Regulatory Nucleic Acid Differ Despite a Common Core. (A) The co-crystal structure of A3A (E72A) with linear ssDNA substrate showing 5′-dA_-2dT_-1dC₀dG₊₁ dG₊₂dG₊₃-3′, where dC₀ is substrate cytidine (PDB 5SWW, Table 1). A3A molecular surface is shown in gray, with surface residues of L1, L3, and L7 colored yellow, pink, and blue, respectively. DNA is shown in stick representation with atoms of carbon, phosphate, oxygen, and nitrogen colored green, orange, red, and blue, respectively. The DNA binds a surface groove between L1, L3, and L7 and takes on a tight U-shaped configuration centered on dC₀, which is flipped out and buried in the active site pocket; a dT nucleotide is preferred at position –1 and it is shown buried in a shallow surface pocket of L7 residues. (B) A ribbon diagram representation of A3A illustrating sidechains (blue) of residues involved in critical binding interactions with ssDNA substrate or in maintaining competent binding site configuration. The catalytic zinc ion is shown as a purple sphere. (C) AID from the co-crystal structure with a dCMP ligand (PDB 5WOU, Table 1) is shown as a molecular surface and as a cartoon diagram (D) in the same orientation and illustrates regions of the bifurcated binding surface model for recognition of G4-structured DNA substrates. In both (C) and (D) the dCMP ligand is depicted with sticks and colored as in (A); the catalytic zinc ion is shown in (D) as a purple sphere. ssDNA overhangs 3′ of the G4 core structure are postulated to bind a substrate channel formed by positively charged L1 residues that are conserved in AID from zebra fish to humans. A second ssDNA of the branched nucleic acid structure is predicted to bind the ‘assistant patch’, a positively charged surface patch formed by conserved residues of α6-helix. These conserved negatively charged residues are depicted as blue surfaces (C) and sidechain sticks (D). Negatively charged residues of L7 serve as a separation wedge for the negatively charged ssDNA backbones and are depicted as a red surface (C) and sidechain sticks (D). In (C) dashed lines indicate the predicted path of ssDNA in the substrate channel and assistant patch, which diverge at the separation wedge. As with other APOBECs, L1, L3, L5, and L7 are clustered near the active site, but in contrast to other APOBECs, α6-helix in AID is believed to play a major role in substrate stabilization and conformation selection through conserved, positively charged residues. (E) The co-crystal structure of dimeric human A3H in complex with an 8-mer RNA duplex (PDB 6BOB, Table 1). A3H molecules are shown as cartoons with gray colored α-helices and β-strands and green loops. RNA molecules are depicted as sticks, with atom coloration as in (A). Each A3H molecule makes significant contact with the RNA molecules but does not form protein–protein interactions. Electrostatic interactions between RNA backbone and α6-helix arginines and L1 tyrosine are shown in the insets. (F) The same co-crystal structure from (E) with RNA removed for clarity is shown superposed with the crystal structure of human apoA3H (PDB 5W45) (dark gray loops). The conformation of L1 and 7 are distorted to accommodate RNA duplex compared with apoA3H.

Figure I

A Nonspecific DNA Binding Site on A3F Facilitates Substrate Catalysis at the Active Site. (A) The catalytic domain of A3F (CD2) with a poly-dT₁₀ ssDNA (PDB 5W2 M, Table 1) co-crystallized with eight A3F (CD2) molecules and two strands of DNA in the asymmetric unit. A3F-based crystallographic interfaces are not robust enough to be true oligomeric interfaces, thus only one A3F (CD2) molecule is shown as a ribbon diagram. In this structure, the zinc ion is not coordinated canonically for a cytidine deaminase, but the CD fold is maintained and select secondary structural features are labeled accordingly. For clarity, only half (dT₁ – dT₅) of the poly-dT₁₀ nucleic acid is shown in stick format with coloration the same as in Figure 1. (B) Close-up views of the five conserved lysines (blue sticks) and two conserved tyrosines (purple sticks) of α5-helix, L10, and α6-helix in ribbon diagram (top) and surface representation (bottom). Y333 pi-stacks with the dT₃ base and Y359 stacks with the dT₄ base. Extensive electrostatic interactions occur between lysines (K352, K355, and K358) and the phosphates of the DNA backbone. Both pi-stacking with bases and backbone electrostatic interactions are consistent with non-sequence-specific binding of nucleic acids.