The dsRBP and inactive editor ADR-1 utilizes dsRNA binding to regulate A-to-I RNA editing across the C. elegans transcriptome

Abstract

Inadequate adenosine-to-inosine editing of noncoding regions occurs in disease but is often uncorrelated with ADAR levels, underscoring the need to study deaminase-independent control of editing. C. elegans have two ADAR proteins, ADR-2 and the theoretically catalytically inactive ADR-1. Using high-throughput RNA sequencing of wild-type and adr mutant worms, we expand the repertoire of C. elegans edited transcripts over 5-fold and confirm that ADR-2 is the only active deaminase in vivo. Despite lacking deaminase function, ADR-1 affects editing of over 60 adenosines within the 3' UTRs of 16 different mRNAs. Furthermore, ADR-1 interacts directly with ADR-2 substrates, even in the absence of ADR-2, and mutations within its double-stranded RNA (dsRNA) binding domains abolish both binding and editing regulation. We conclude that ADR-1 acts as a major regulator of editing by binding ADR-2 substrates in vivo. These results raise the possibility that other dsRNA binding proteins, including the inactive human ADARs, regulate RNA editing through deaminase-independent mechanisms.

Figure 1. ADR-1 alters editing at specific adenosines in multiple mRNAs

(A and B) Editing levels at individual nucleotides within the 3′ UTRs were measured for 3 biological replicates. Error bars represent standard error of the mean (SEM). Significant changes (p≤ 0.05) in editing levels between (A) wild-type and adr-1(−) or (B) adr-1(−) and FLAG-ADR-1 are marked with an asterisk.

Figure 2. ADR-1 binds ADR-2 substrates in vivo

(A) Lysates from the indicated worm lines were subjected to FLAG IP and treatment with Proteinase K (Prot. K). A portion of the untreated lysate (IP−, Prot. K−), IP (IP+, Prot. K−) and beads after Prot. K treatment (IP+, Prot. K−) were subjected to immunoblotting for the FLAG epitope. (B) cDNA levels for the indicated endogenous mRNAs were measured using qRT-PCR. Values from the IP samples of FLAG-ADR-1 in adr-1(−) and the negative control adr-1(−) were divided by their respective input levels. Error bars represent SEM for three biological replicates. (C) Lysates from the indicated worm lines were subjected to immunoprecipitation with magnetic FLAG resin. A portion of the input lysate and IPs were subjected to immunoblotting for the FLAG epitope. (D) cDNA levels for the indicated endogenous mRNAs were measured using qRT-PCR. Ratios of the cDNAs present in the IP samples of the indicated strains were divided by their respective input levels and normalized to the negative control adr-1(−) to give a fold enrichment. Error bars represent SEM for three biological replicates.

Figure 3. Mutation of the KKxxK Motif within the dsRBDs of ADR-1 abolishes dsRNA binding and editing regulation

(A) Schematic of ADR-1 protein with dsRBDs (grey ovals) and deaminase domain (patterned rectangle). Lysine (K) residues mutated are indicated above each dsRBD. (B) FLAG Immunoblotting of lysates and IPs of the indicated strains. (C) Ratio of the cDNA present in the IP samples divided by the input cDNA levels for the indicated strains were divided by the IP:input ratio of the adr-1(−) worms. Error bars represent SEM for three biological replicates. (D) Calculated percent editing in the indicated strains for the endogenous mRNAs of C35E7.6, lam-2 and pop-1. Error bars represent SEM of 3 biological replicates. Significant changes (p≤ 0.05) in editing levels between FLAG-ADR-1 and FLAG-ADR-1 ds1+2 mutant are marked with an asterisk.

Figure 4. Impact of dsRNA binding by ADR-1 on the editing transcriptome

(A) Bioinformatics strategy depicting the major steps for processing RNA-seq data into A-to-I sites for each strain. (B) Distribution of identified RNA editing sites within annotated transcriptome regions. (C) Nucleotide preferences for the 270 candidate editing sites were calculated compared to a randomized control. Enriched and depleted nucleotides are shown above and below the axis, respectively. The level of conservation is represented by letter height. Logos were generated using a t-test with P<.005 and no Bonferroni correction. (D) Scatter plots of percent editing of quantified sites that overlap in the wildtype (CEN2) and FLAG-ADR-1 datasets. The r² fit to the y=x line (black diagonal). The margin (dotted line) between no-change and differentially-edited sites equals 12 units of change in the edit % (one standard deviation). (E) Editing levels for 13 sites from the RNA-seq data where editing levels between adr-1(−) and FLAG-ADR-1 and between FLAG-ADR-1 and FLAG-ADR-1 ds1+2 mutant were greater than 12% (Table S3). Adenosines that had no observed editing are marked with a zero above the x-axis. (F and G) Immunoblotting analysis of FLAG IPs from the indicated strains. IPs were performed as previously stated except worms were not subjected to UV-crosslinking and only light salt washes were employed.