Recognition of duplex RNA by the deaminase domain of the RNA editing enzyme ADAR2

Abstract

Adenosine deaminases acting on RNA (ADARs) hydrolytically deaminate adenosines (A) in a wide variety of duplex RNAs and misregulation of editing is correlated with human disease. However, our understanding of reaction selectivity is limited. ADARs are modular enzymes with multiple double-stranded RNA binding domains (dsRBDs) and a catalytic domain. While dsRBD binding is understood, little is known about ADAR catalytic domain/RNA interactions. Here we use a recently discovered RNA substrate that is rapidly deaminated by the isolated human ADAR2 deaminase domain (hADAR2-D) to probe these interactions. We introduced the nucleoside analog 8-azanebularine (8-azaN) into this RNA (and derived constructs) to mechanistically trap the protein-RNA complex without catalytic turnover for EMSA and ribonuclease footprinting analyses. EMSA showed that hADAR2-D requires duplex RNA and is sensitive to 2'-deoxy substitution at nucleotides opposite the editing site, the local sequence and 8-azaN nucleotide positioning on the duplex. Ribonuclease V1 footprinting shows that hADAR2-D protects ∼ 23 nt on the edited strand around the editing site in an asymmetric fashion (∼ 18 nt on the 5' side and ∼ 5 nt on the 3' side). These studies provide a deeper understanding of the ADAR catalytic domain-RNA interaction and new tools for biophysical analysis of ADAR-RNA complexes.

Figure 1.

(A) (top): domain maps for ADAR proteins. hADAR2-D refers to the deaminase domain of human ADAR2. Yellow: deaminase (aa 299–701); red: double-stranded RNA binding (dsRBD); green: Z binding domain (ADAR1 only); brown: R-rich domain (ADAR3 only). (B) (top): deamination reaction catalyzed by ADARs showing high-energy intermediate with tetrahedral carbon at adenine position 6. Bottom: nucleoside analog 8-azanebularine (8-azaN) forms a mimic of deamination intermediate when hydrated (28).

Figure 2.

(A) (top): predicted secondary structure for 55 nucleotides flanking a previously identified ADAR2 site in the S. cerevisiae bdf2 mRNA shown bolded (a) (27). Bottom: primer extension analysis for hADAR2-D deamination products with (a). Line indicates removal of unrelated lanes. (B) (top): secondary structure for 43-nt bdf2-derived substrate (b) with reactive adenosine shown bolded. Bottom: primer extension analysis for hADAR2-D deamination products with (b). (C) Plot of deamination product as a function of time for (b) and hADAR2-D.

Figure 3.

(A) (top): secondary structure for (b) and (c) with X = A or 8-azaN at reactive site. The 39-nt RNA (c) lacks the four nucleotides shown in blue in figure. Bottom: autoradiogram of gel used to resolve bound from free RNA in EMSA with hADAR2-D and (b) where X = 8-azaN. Lanes 1–12: 0, 5, 10, 25, 50, 75, 100, 250, 500, 750, 1000 and 3200-nM ADAR2 added. (B) Plot of fraction (b) bound as a function of hADAR2-D concentration: squares: X = 8-azaN; circles: X = A.

Figure 4.

(A) (top): predicted secondary structure for bdf2-derived RNA (d). Bottom: autoradiogram of gel used to resolve bound from free RNA in EMSA with hADAR2-D and (d) where X = 8-azaN. Lanes 1–11: 0, 5, 10, 25, 50, 75, 100, 250, 500, 1000 and 3200-nM ADAR2 added. (B) (top): predicted secondary structure for GluR B R/G site RNA (e). Bottom: autoradiogram of gel used to resolve bound from free RNA in EMSA with hADAR2-D and (e) where X = 8-azaN. Lanes 1–10: 0, 25, 50, 75, 100, 250, 500, 750, 1000 and 3200-nM ADAR2 added. (C) Plot of fraction RNA bound as a function of hADAR2-D concentration: circles: (d) where X = 8-azaN; diamonds: (e) where X = 8-azaN; squares: (d) where X = A.

Figure 5.

Duplexes f–j with varying complementary strand structures. Lower case gray indicates a 2′-deoxynucleotide. X = 8-azaN.

Figure 6.

(A) Autoradiogram of gel used to resolve bound from free RNA in EMSA with hADAR2-D and (i) where X = 8-azaN. Lanes 1–11: 0, 5, 10, 25, 50, 75, 100, 250, 500, 1000 and 3200-nM ADAR2 added. (B) Autoradiogram of gel used to resolve bound from free RNA in EMSA with hADAR2-D E488Q and (i) where X = 8-azaN. Lanes 1–9: 0, 5, 10, 25, 50, 75, 100, 250 and 500-nM ADAR2 added. (C) Plot of fraction (i) bound as a function of protein concentration: squares: hADAR2-D E488Q; circles: hADAR2-D.

Figure 7.

(A) Secondary structure of the 65-bp 8-azaN-containing structure with N = 8-azaN at reaction site. (B) Autoradiogram of gel used to resolve bound from free RNA in EMSA with hADAR2-D E488Q and the long 8-azaN-containing structure. Samples were equilibrated in 10-mM Tris-HCl, pH 7, 0.2% glycerol, 60-mM KCl, 25-mM NaCl, 10 μM 2-mercaptoethanol, 15-μM EDTA, 100 μg/ml BSA, 0.3-mM MgCl₂ and 1.0-μg/ml yeast tRNA for 30 min at 30°C. Lanes 1–7: 0, 25, 50, 75, 100, 150 and 200-nM ADAR2 added. (C) Nuclease V1 footprint of the long 8-azaN-containing RNA. Phosphor autoradiogram of a 12% denaturing polyacrylamide gel separating the 5′-end-labeled RNA cleavage products, nucleotides protected from V1 cleavage are identified with a bracket. Reaction conditions as labeled: lane 1, T1; lane 2, T1 dephosphorylated; lane 3, untreated RNA; lane 4, no nuclease V1; lanes 5–12: 0, 10, 20, 25, 50, 75, 150 and 200-nM ADAR2 added. Each line indicates the removal of a lane containing an over-digested sample.