Heteronuclear NMR spectroscopy and other experiments indicate that the true substrate of the E1 component of 2-oxo acid dehydrogenase complexes is not lipoic acid but the lipoyl domain of the E2 component. E1 can recognize the lipoyl-lysine residue as such, but reductive acylation ensues only if the domain to which the lipoyl group is attached is additionally recognized by virtue of a mosaic of contacts distributed chiefly over the half of the domain that contains the lipoyl-lysine residue. The lipoyl-lysine residue may not be freely swinging, as supposed hitherto, but may adopt a preferred orientation pointing towards a nearby loop on the surface of the lipoyl domain. This in turn may facilitate the insertion of the lipoyl group into the active site of E1, where reductive acylation is to occur. The results throw new light on the concept of substrate channelling and active-site coupling in these giant multifunctional catalytic machines.