The interaction between lipoamide dehydrogenase (E3) and dihydrolipoyl transacetylase (E2p) from the pyruvate dehydrogenase complex was studied during the reconstitution of monomeric E3 apoenzymes from Azotobacter vinelandii and Pseudomonas fluorescens. The dimeric form of E3 is not only essential for catalysis but also for binding to the E2p core, because the apoenzymes as well as a monomeric holoenzyme from P. fluorescens, which can be stabilized as an intermediate at 0 degree C, do not bind to E2p. Lipoamide dehydrogenase from A. vinelandii contains a C-terminal extension of 15 amino acids with respect to glutathione reductase which is, in contrast to E3, presumably not part of a multienzyme complex. Furthermore, the last 10 amino acid residues of E3 are not visible in the electron density map of the crystal structure and are probably disordered. Therefore, the C-terminal tail of E3 might be an attractive candidate for a binding region. To probe this hypothesis, a set of deletions of this part was prepared by site-directed mutagenesis. Deletion of the last five amino acid residues did not result in significant changes. A further deletion of four amino acid residues resulted in a decrease of lipoamide activity to 5% of wild type, but the binding to E2p was unaffected. Therefore it is concluded that the C-terminus is not directly involved in binding to the E2p core. Deletion of the last 14 amino acids produced an enzyme with a high tendency to dissociate (Kd approximately 2.5 microM). This mutant binds only weakly to E2p. The diaphorase activity was still high. This indicates, together with the decreased Km for NADH, that the structure of the monomer is not appreciably changed by the mutation. Rather the orientation of the monomers with respect to each other is changed. It can be concluded that the binding region of E3 for E2p is constituted from structural parts of both monomers and binding occurs only when dimerization is complete.