The binary complex of diacetylchloramphenicol and chloramphenicol acetyltransferase (CAT) has been studied by a combination of isotope-edited H-1 NMR spectroscopy and site-directed mutagenesis. One-dimensional HMQC spectra of the complex between 1,3-[2-C-13]diacetylchloramphenicol and the type III natural variant of CAT revealed the two methyl H-1 signals arising from each C-13-labeled carbon atom in the acetyl groups of the bound ligand. Slow hydrolysis of the 3-acetyl group by the enzyme precluded further analysis of this binary complex. It was possible to slow down the rate of hydrolysis by use of the catalytically defective S148A mutant of CAT(III) (Lewendon et al., 1990); in the complex of diacetylchloramphenicol with S148A CAT(III), the chemical shifts of the acetyl groups of the bound ligand were the same as in the wild-type complex. The acetyl signals were individually assigned by repeating the experiment using 1-[2-C-13], 3-[2-C-12]diacetylchloramphenicol, where only one signal from the bound ligand was observed. A two-dimensional H-1,H-1 NOESY experiment, with C-13(omega-2) half-filter, on the 1,3-[2-C-13]diacetylchloramphenicol/S148A CAT(III) complex showed a number of intermolecular NOEs from each methyl group in the ligand to residues in the chloramphenicol binding site. The 3-acetyl group showed strong NOEs to two aromatic signals which were selected for assignment. The possibility that the NOEs originated from the aromatic protons of diacetylchloramphenicol itself was eliminated by assignment of the signals from enzyme-bound diacetylchloramphenicol and chloramphenicol using perdeuterated CAT(III). Examination of the X-ray crystal structure of the chloramphenicol/CAT(III) binary complex indicated four plausible candidate aromatic residues: Y25,F33,F103, and F158. Each was separately substituted with isoleucine, and the two-dimensional C-13(omega-2) half-filter NOESY experiment was repeated on each mutant enzyme/diacetylchloramphenicol binary complex. The two NOEs were present in each case, except for the Y25I mutant, which yielded no cross peaks. We conclude that Tyr-25 is close to the 3-acetyl group in the 1,3-diacetylchloramphenicol/CAT(III) binary complex; this information allows a more precise formulation of a structural model for the interaction of CAT(III) with its ultimate product (in the forward reaction) or substrate (for the reverse reaction).