| Description |
Adenosine deaminases that act on RNA (ADARs) deaminate adenosines in doublestranded RNA (dsRNA) to produce inosines. The extent an adenosine is edited depends on the sequence context of the target adenosine. Human ADAR2 (hADAR2) has a 5' nearest neighbor preference of U>A>C>G and a 3' preference of G>C>U«A, but it is not known which amino acids mediate these preferences. Previous studies show that preferences are derived mainly from the catalytic domain. Thus, we adapted a previously reported screen in yeast to identify mutations in the hADAR2 catalytic domain that allow editing of an adenosine in context of a disfavored triplet, GAC. A favored triplet, UAG, was used as the positive control. Hairpin substrates containing disfavored GAC and favored UAG were based on the R/G editing site of GRIA2 (glutamate receptor, ionotropic, AMPA 2) pre-mRNA, a well-studied endogenous substrate for hADAR2. Four mutants that edited GAC more than WT hADAR2 (E488Q, V493T, N597K and N613K) and one mutant that did not edit GAC (T490A) were further characterized by determining their binding affinity, catalytic rate, base-flipping and preferences to understand the effect of these mutations on ADAR reactivity. Gel-shift assays showed two mutants, N597K and N613K, had ~2-fold higher binding affinity compared to WT hADAR2, suggesting these mutants may have been selected in the screen due to tighter binding. Other mutants E488Q, T490A and V493T, which are on a highly conserved loop close to the active site, showed similar binding affinity as WT hADAR2 for both UAG and GAC, indicating discrimination was not derived from differences in binding affinity. We also determined catalytic rates, and probed base-flipping by substituting the target adenosine with the fluorescent base analog 2-aminopurine (2-AP). Remarkably, with both UAG and GAC substrates, mutants with similar binding affinity showed a correlation between catalytic rate and base-flipping, as indicated by a change in 2-AP fluorescence intensity (FI). Our data provide the first information on the residues important for preferences, and point to a conserved loop as key. Unexpectedly, our data suggest that hADAR2's preferences are derived from differences in base-flipping, rather than direct recognition of the neighboring base. |
