The X-ray crystal structure of adenosylcobinamide kinase/adenosylcobinamide phosphate guanylyltransferase (CobU) from Salmonella typhimurium bound to GMP has been determined by molecular replacement to 2.2 A resolution. CobU is a bifunctional enzyme, which catalyzes the phosphorylation of the 1-amino-O-2-propanol side chain of the adenosylcobinamide ring and subsequently functions as a guanylyltransferase to form adenosylcobinamide.GDP. The transferase activity involves a covalent enzyme-guanylyl intermediate that is most likely a phosphoramidate linkage to His(46). Previous studies have shown that the enzyme is a homotrimer and adopts a pinwheel shape. Each subunit consists of a single domain of six parallel beta-strands and one antiparallel strand flanked on either side by a total of five alpha-helices and one helical turn. Interestingly, His(46) in the apoenzyme is located a considerable distance from the kinase active site or P-loop motif and is solvent-exposed [Thompson, T. B., et al. (1998) Biochemistry 37, 7686-7695]. To examine the structural relationship of the two active sites, CobU was cocrystallized with GTP and pyrophosphate. Crystals belong to space group P2(1)2(1)2(1) with the following unit cell dimensions: a = 58. 4 A, b = 87.8 A, and c = 101.6 A. The structure shows electron density for the hydrolysis product GMP rather than the expected covalent guanylyl intermediate which appears to have been hydrolyzed in the crystal lattice. Even so, CobU exhibits a substantial conformational rearrangement. The helix axis containing His(46), the site of guanylylation, rotates 30 degrees and translates 11 A relative to the apo structure and is accompanied by compensatory unwinding and rewinding at the helix ends to allow the induction of a guanosine binding pocket between beta-strand 2 and alpha-helix 2. This conformational change brings the C(alpha) of His(46) approximately 10 A closer to the P-loop motif such that a phosphate ion located in the P-loop is only 6 A from the alpha-phosphate of GMP. This suggests that the P-loop motif may be used to coordinate the terminal phosphates in both the transferase and kinase reactions and implies that the active sites for both reactions overlap.