We report the crystal structure of the catalytic domain of human ADAR2, an RNA editing enzyme, at 1.7 angstrom resolution. The structure reveals a zinc ion in the active site and suggests how the substrate adenosine is recognized. Unexpectedly, inositol hexakisphosphate (IP 6 ) is buried within the enzyme core, contributing to the protein fold. Although there are no reports that adenosine deaminases that act on RNA (ADARs) require a cofactor, we show that IP 6 is required for activity. Amino acids that coordinate IP 6 in the crystal structure are conserved in some adenosine deaminases that act on transfer RNA (tRNA) (ADATs), related enzymes that edit tRNA. Indeed, IP 6 is also essential for in vivo and in vitro deamination of adenosine 37 of tRNA ala by ADAT1.One form of RNA editing is catalyzed by adenosine deaminases that act on RNA (ADARs), a family of enzymes that deaminate adenosine to form inosine in double-stranded RNA (dsRNA) (Fig. 1A) (1). ADARs are important for proper neuronal function (2-4) and also are implicated in the regulation of RNA interference (RNAi) (5-7). Inosine is recognized as guanosine by most cellular proteins and the translation machinery, and it pairs most stably with cytidine. Therefore, editing of RNA can alter a codon, create splice sites, and change its structure. The latter occurs when an AU base pair is changed to an IU mismatch and may be important for the effects of ADARs on the RNAi pathway.ADARs from all organisms have a common domain structure consisting of one to three dsRNA binding motifs (dsRBMs) near the N terminus, followed by a conserved C-terminal catalytic domain (1,8). Human ADAR2 (hADAR2) contains two dsRBMs, and its best characterized substrates are the pre-mRNAs of glutamate and serotonin receptors (9,10). Editing of codons within these RNAs leads to altered amino acids and generates receptors with altered function. hADAR2 also edits its own message to create a new splice site (11). Purified hADAR2 deaminates substrates in vitro (12) in the absence of any added cofactors, and deletions of Nterminal sequences, including dsRBM1, result in an active protein that accurately edits an RNA substrate (13). In addition, we found that a protein consisting of only the catalytic deaminase domain of hADAR2 (hADAR2-D, residues 299 to 701) ( fig. S1A) was active in vitro, although it deaminates RNA less efficiently than full-length hADAR2 ( fig. S1B).

Metabolism is highly interconnected and also has profound effects on other cellular processes. However, the interactions between metabolites and proteins that mediate this connectivity are frequently low affinity and difficult to discover, hampering our understanding of this important area of cellular biochemistry. Therefore, we developed the MIDAS platform, which can identify protein-metabolite interactions with great sensitivity. We analyzed 33 enzymes from central carbon metabolism and identified 830 protein-metabolite interactions that were mostly novel, but also included known regulators, substrates, products and their analogs. We validated previously unknown interactions, including two atomic-resolution structures of novel protein-metabolite complexes. We also found that both ATP and long-chain fatty acyl-CoAs inhibit lactate dehydrogenase A (LDHA), but not LDHB, at physiological concentrations in vitro. Treating cells with long-chain fatty acids caused a loss of pyruvate/lactate interconversion, but only in cells reliant on LDHA. We propose that these regulatory mechanisms are part of the metabolic connectivity that enables survival in an ever-changing nutrient environment, and that MIDAS enables a broader and deeper understanding of that network.

In budding yeast, the Met15 enzyme has long been assumed to be the sole homocysteine synthase, facilitating de novo synthesis of sulfur-containing organic compounds (organosulfurs) from inorganic precursors. Here we show that an alternative homocysteine synthase encoded by the previously uncharacterized gene YLL058W supports growth of mutants lacking MET15 in the absence of exogenous organosulfurs. This growth is observed specifically when cells are deposited in an automated fashion to seed colonies, but not with traditional cell propagation techniques such as thick patches of cells or liquid cultures. We show that the lack of growth in these contexts, which has historically justified the status of MET15 as a classic auxotrophic marker, is largely due to toxic levels of hydrogen sulfide accumulation rather than an inability to perform de novo homocysteine biosynthesis. These data have broad implications for investigations of sulfur starvation/metabolism, including studies of aging and emerging cancer therapeutics.

No abstract