N 5 -carboxyaminoimidazole ribonucleotide synthetase (N 5 -CAIR synthetase) converts 5-aminoimidazole ribonucleotide (AIR), MgATP, and bicarbonate into N 5 -CAIR, MgADP, and P i . The enzyme is required for de novo purine biosynthesis in microbes yet is not found in humans suggesting that it represents an ideal and unexplored target for antimicrobial drug design. Here we report the x-ray structures of N 5 -CAIR synthetase from Escherichia coli with either MgATP or MgADP/P i bound in the active site cleft. These structures, determined to 1.6-Å resolution, provide detailed information regarding the active site geometry before and after ATP hydrolysis. In both structures, two magnesium ions are observed. Each of these is octahedrally coordinated, and the carboxylate side chain of Glu 238 bridges them. For the structure of the MgADP/P i complex, crystals were grown in the presence of AIR and MgATP. No electron density was observed for AIR, and the electron density corresponding to the nucleotide clearly revealed the presence of ADP and P i rather than ATP. The bound P i shifts by approximately 3 Å relative to the γ-phosphoryl group of ATP and forms electrostatic interactions with the side chains of Arg 242 and His 244. Since the reaction mechanism of N 5 -CAIR synthetase is believed to proceed via a carboxyphosphate intermediate, we propose that the location of the inorganic phosphate represents the binding site for stabilization of this reactive species. Using the information derived from the two structures reported here, coupled with molecular modeling, we propose a catalytic mechanism for N 5 -CAIR synthetase.The original pathway proposed for de novo purine biosynthesis by Buchanan in the late 1950s involved the 10-step conversion of phosphoribosyl pyrophosphate into the common branchpoint intermediate inosine monophosphate (1). This view of purine biosynthesis existed well into the mid-1990s until it was discovered that the 10-step pathway was not universal across all organisms (2-4). Indeed, de novo purine biosynthesis is different in microbes versus humans, and the difference centers around the synthesis of the key intermediate, 4-carboxy-5-aminoimidazole ribonucleotide (CAIR). In bacteria and yeast, CAIR is produced from 5-aminoimidazole ribonucleotide (AIR) by the action of two enzymes rather than one as in higher eukaryotes (Scheme 1). Clearly, de novo purine biosynthesis has evolutionarily diverged among lower and higher organisms. This divergent nature of a primary metabolic pathway has significant ramifications for the design of novel antimicrobial agents. Genetic studies have shown that both N 5 -CAIR synthetase and N 5 -CAIR mutase are required for microbial growth (6-9). Deletion of either enzyme results in a purine auxotroph that is unable to propagate in human or mouse serum and is not viable in animal models predictive of disease (6,(10)(11)(12). These results support the conclusion that inhibitors of these enzymes could be potential antibacterial and antifungal agents.The first structural ...