Eur. J. Biochem. 265, 1098±1107 (1999) q FEBS 1999 Pyruvate dehydrogenase from Azotobacter vinelandii Properties of the N-terminally truncated enzyme Annechien F. Hengeveld, Siemen E. Schoustra, Adrie H. Westphal and Aart de Kok Department of Biomolecular Sciences, Laboratory of Biochemistry, Wageningen University, the Netherlands The pyruvate dehydrogenase multienzyme complex (PDHC) catalyses the oxidative decarboxylation of pyruvate and the subsequent acetylation of coenzyme A to acetyl-CoA. Previously, limited proteolysis experiments indicated that the N-terminal region of the homodimeric pyruvate dehydrogenase (E1p) from Azotobacter vinelandii could be involved in the binding of E1p to the core protein (E2p) [Hengeveld, A. F., Westphal, A. H. & de Kok, A. (1997) Eur J. Biochem. 250, 260±268]. To further investigate this hypothesis N-terminal deletion mutants of the E1p component of Azotobacter vinelandii pyruvate dehydrogenase complex were constructed and characterized. Up to nine N-terminal amino acids could be removed from E1p without effecting the properties of the enzyme. Truncation of up to 48 amino acids did not effect the expression or folding abilities of the enzyme, but the truncated enzymes could no longer interact with E2p. The 48 amino acid deletion mutant (E1pD48) is catalytically fully functional: it has a Vmax value identical to that of wild-type E1p, it can reductively acetylate the lipoamide group attached to the lipoyl domain of the core enzyme (E2p) and it forms a dimeric molecule. In contrast, the S0.5 for pyruvate is decreased. A heterodimer was constructed containing one subunit of wild-type E1p and one subunit of E1pD48. From the observation that the heterodimer was not able to bind to E2p, it is concluded that both N-terminal domains are needed for the binding of E1p to E2p. The interactions are thought to be mainly of an electrostatic nature involving negatively charged residues on the N-terminal domains of E1p and previously identified positively charged residues on the binding and catalytic domain of E2p. Keywords: pyruvate dehydrogenase; multienzyme complex, mutagenesis; binding; catalysis. The pyruvate dehydrogenase multienzyme complex (PDHC) 20±40 amino acids. This architecture allows a high degree of from gram-negative bacteria consists of multiple copies of three flexibility of the individual domains, required for catalysis. One different enzyme components: pyruvate dehydrogenase (E1p), to three N-terminal lipoyl domains, each carrying a lipoyl- dihydrolipoyl acyltransferase (E2p) and lipoamide dehydro- lysine group, are attached to the E1/E3 binding domain. The genase (E3). The complex catalyses the oxidative decarboxy- C-terminal catalytic domain forms the structural core of the lation of pyruvate and the subsequent acetylation of coenzyme complex. Three-dimensional structures of the different domains A to acetyl-CoA (reviewed in [1±4]). The substrate specific of E2 have been solved either by X-ray crystallography [5,6] or thiamin diphosphate dependent E1p catalyses the decarboxy- by NMR [7±11]. Structures of E3 from several sources have lation of pyruvate and subsequently the reductive acetylation of been solved by X-ray crystallography [12,13]. The well- the lipoamide groups attached to E2. The E2 component then conserved E2p binding domain consists of two parallel helices transfers the acyl group to CoA. Finally, the reduced lipoyl connected by a short strand, a turn and a disordered loop group is reoxidized by the FAD dependent E3 component, [14,15]. The binding mode of Bacillus stearothermophilus E3 which transfers the reduction equivalents to NAD+. to the binding domain was recently solved by X-ray crystallo- The E2 component plays a central role in the complex, graphy [16]. The interactions between E3 and the binding both catalytically and structurally. This component consists domain are dominated by an electrostatic zipper formed by of three to five domains interconnected by flexible linkers of basic residues of the binding domain and acidic residues of one Correspondence to A. de Kok, Department of Biomolecular Sciences, of the subunits of E3. Because E3 is homodimeric the binding Laboratory of Biochemistry, Wageningen University, Dreijenlaan 3, domain can interact with both E3 subunits, but the binding site 6703 HA Wageningen, the Netherlands. Fax: + 31 0 317 484801, is located so close to the twofold symmetry axis that it is Tel.: + 31 0 317 482866, E-mail: [email protected] impossible for two molecules of the binding domain to bind Abbreviations: PDHC, pyruvate dehydrogenase multienzyme complex; simultaneously to one E3 dimer. The E2p binding domain E1p, pyruvate dehydrogenase; E2p, dihydrolipoyl transacetylase; E3, behaves like a Janus-face protein: while E3 interacts with the lipoamide dehydrogenase; E1o, 2-oxoglutarate dehydrogenase; E1b, N-terminal helix, mutagenesis experiments indicate that E1p branched-chain a-keto acid decarboxylase; Cl2Ind, from Azotobacter vinelandii interacts with the C-terminal part 2,6-dichlorophenol-indophenol. [17]. The point mutation R416D in the catalytic domain from Enzymes: pyruvate dehydrogenase (EC 1.2.4.1); dihydrolipoyl A. vinelandii also resulted in highly impaired E1p binding. transacetylase (EC 2.3.1.12); lipoamide dehydrogenase (EC 1.8.1.4); Deletions in the flexible linker region between the binding and 2-oxoglutarate dehydrogenase (EC 1.2.4.2); branched-chain a-keto acid the catalytic domain had no effect on the affinity of E1p for decarboxylase (EC 1.2.1.25). E2p. Thus it can be concluded that the binding site for E1p (Received 15 June 1999, revised 25 August 1999, accepted consists of two regions, one located on the binding domain and 9 September 1999) one on the catalytic domain. q FEBS 1999 Properties of the N-terminally truncated E1p (Eur. J. Biochem. 265) 1099 The E1 component exists both as a homodimer (a2)ora lacZ DM15] was used [25]. The plasmids pUC9 and pUC18 heterotetramer (a2b2) depending on the source and type of were used as cloning vectors [26]. complex. Homodimeric E1 with a molecular mass of approximately 100 kDa is found in all 2-oxoglutarate dehydrogenases (E1o) and in pyruvate dehydrogenases from Construction of N-terminal deletion mutants gram-negative bacteria. Heterotetrameric E1 is found in Standard DNA operations were performed as described [27]. branched-chain-2-oxoacid dehydrogenases (E1b) and in pyru- pAFH001 [24], containing the complete E1p gene, was used as vate dehydrogenases from eukaryotes and gram-positive the starting material for the construction of the endonuclease bacteria. Little sequence similarity is found between the Bal-31 deletion mutants. The general approach for preparing different classes of E1 [18], with exception of the thiamin Bal-31 deletion mutants was derived from [28]. The region diphosphate binding motif, present in all thiamin diphosphate encoding A. vinelandii E1p was excised from pAFH001 using dependent enzymes [19]. Until now no structural information is EcoRI and HindIII and cloned into pUC18 digested with available for either homodimeric or heterotetrameric E1 and HincII. The resulting construct pSES01 was used for the any information on the interaction between E1 and E2 is deletion experiments. A 1829 basepair l-DNA fragment lacking. Previously, cryoelectron microscopy experiments (EcoRI-XmaI) was inserted upstream from the E1p encoding showed that E1o, E1p and E3 are separated from the surface region to prevent removal of bases from the region encoding the of the E2 core by 4±6 nm, and sometimes thin bridges of promoter and ribosome binding region of the vector (pSES02) density are visible in the gap between the core and the bound (Fig. 1). pSES02 (5 mg) was linearized with SmaI and subunits. The bridging density between E1 and E2 seems to incubated with Bal31 nuclease (0.015 U) at 37 8C yielding a represent more mass than between E2 and E3 [20,21]. Studies digestion speed of 12 bp´min21. Aliquots were taken at timed by scanning transmission electron microscopy show that E1 is intervals up to 25 min and the reaction was stopped by addition bound along the edges of the cubic core, while E3 is present on of EGTA, pH 9.0 mp to a final concentration of 70 mm. The the faces of the core [22]. When E1 is bound to the complex, samples were digested by EcoRI to remove the remaining the lipoyl domains are located mostly at the E2±E1 interfaces, l-DNA fragment, blunt-ended, ligated and transformed in while removal of E1 from the complex gives a much larger E. coli TG2 cells. The transformed E. coli TG2 cells were conformational freedom to the lipoyl domains [23]. screened for E1p expression by Western blotting using E1p from A. vinelandii has recently been cloned and antiserum raised against A. vinelandii E1p [29]. Positive expressed in Escherichia coli, which makes it possible to colonies were subsequently analysed by SDS/PAGE. The plas- study the interaction between E1 and E2 in more detail. Limited mids were isolated and the DNA-sequences of the N-terminal proteolysis experiments of A. vinelandii E1p showed that its coding regions were determined using Taq polymerase accord- N-terminal region can easily be cleaved off [24]. The remaining ing to [28]. fragment is still active, but unable to bind to E2p. This suggests that E1p has an N-terminal domain that is involved in the binding to E2p, but not in catalysis. To further investigate the Enzyme expression and purification role of this domain in the binding to E2p N-terminal deletion mutants were made using endonuclease Bal-31. The charac- E. coli TG2 harbouring either the recombinant plasmid terization of these mutants is described in this paper. pAFH001, expressing wild-type E1p, or a deletion mutant were grown and purified as described in [24]. However, E1pD48 elutes at 0.2 m KCl from the Q-Sepharose column EXPERIMENTAL PROCEDURES Materials Restriction endonucleases, Endonuclease Bal-31, T4-DNA-ligase and Taq polymerase were obtained from Bethesda Research Laboratory (BRL).