Purification of the Pyruvate Dehydrogenase Multienzyme Complex of Zymomonas mobilis and Identification and Sequence Analysis of the Corresponding Genes

Abstract

The pyruvate dehydrogenase (PDH) complex of the gram-negative bacterium Zymomonas mobilis was purified to homogeneity. From 250 g of cells, we isolated 1 mg of PDH complex with a specific activity of 12.6 U/mg of protein. Analysis of subunit composition revealed a PDH (E1) consisting of the two subunits E1α (38 kDa) and E1β (56 kDa), a dihydrolipoamide acetyltransferase (E2) of 48 kDa, and a lipoamide dehydrogenase (E3) of 50 kDa. The E2 core of the complex is arranged to form a pentagonal dodecahedron, as shown by electron microscopic images, resembling the quaternary structures of PDH complexes from gram-positive bacteria and eukaryotes. The PDH complex-encoding genes were identified by hybridization experiments and sequence analysis in two separate gene regions in the genome of Z. mobilis. The genes pdhAα (1,065 bp) and pdhAβ (1,389 bp), encoding the E1α and E1β subunits of the E1 component, were located downstream of the gene encoding enolase. The pdhB (1,323 bp) and lpd (1,401 bp) genes, encoding the E2 and E3 components, were identified in an unrelated gene region together with a 450-bp open reading frame (ORF) of unknown function in the order pdhB-ORF2-lpd. Highest similarities of the gene products of the pdhAα, pdhAβ, and pdhB genes were found with the corresponding enzymes of Saccharomyces cerevisiae and other eukaryotes. Like the dihydrolipoamide acetyltransferases of S. cerevisiae and numerous other organisms, the product of the pdhB gene contains a single lipoyl domain. The E1β subunit PDH was found to contain an amino-terminal lipoyl domain, a property which is unique among PDHs.

The gram-negative, fermentative bacterium Zymomonas mobilis catabolizes glucose anaerobically via the Entner-Doudoroff pathway to pyruvate. Up to 98% of the pyruvate is converted to the fermentation end products ethanol and CO₂. Only a small part of the pyruvate is oxidatively decarboxylated by the reaction of the pyruvate dehydrogenase PDH complex to acetyl coenzyme A + CO₂ and NADH (56). Since Z. mobilis lacks the 2-oxoglutarate dehydrogenase complex and other enzymes of the tricarboxylic acid cycle, the PDH complex plays an exclusively anabolic role in this organism (11).

PDH complexes consist of multiple copies of the three enzymes PDH (E1p), dihydrolipoamide acetyltransferase (E2p), and dihydrolipoamide dehydrogenase (E3). The E2p component forms the structural core of the complex with either octahedral symmetry (24-mer), as found in gram-negative bacteria, or icosahedral symmetry (60-mer), as is the case in gram-positive bacteria and eukaryotes studied so far. The E1p and E3 components are attached noncovalently to the E2p core. The E1p component occurs in two forms dependent on the symmetry of the complex. In octahedral complexes, E1p is a homodimer (α₂); in icosahedral complexes, it exists as a heterotetramer (α₂β₂) (37, 54). The structure and reaction mechanism of the complex depend on the highly segmented structure of the E2 chain. From the N terminus, it consists of one to three lipoyl domains, containing the lipoyl lysine residues, a small domain responsible for the E3 and/or E1 binding, and a C-terminal domain, which contains the acetyltransferase active site and aggregates to form the octahedral or icosahedral core of the complex. The domains are separated by flexible linker segments (49) which allow the lipoyl domains to move and facilitate substrate transfer between the active sites of the three component enzymes. The genes encoding the E1p, E2p, and E3 components of the PDH complex from various prokaryotic and eukaryotic sources have been cloned and sequenced. The genes encoding the E1p and E2p components of the PDH complex are clustered in the genomes of all prokaryotes studied so far. In contrast to the substrate-specific E1p and E2p, the lipoamide dehydrogenase (E3) is a common component of all 2-oxo acid dehydrogenase complexes. The E3-encoding gene was found either as part of the pdh gene cluster or, as in Pseudomonas aeruginosa and Azotobacter vinelandii, as part of the odh gene cluster, which encodes the 2-oxoglutarate dehydrogenase complex.

In this paper, we report on the purification and structural organization of the PDH complex of Z. mobilis. Furthermore, we describe the identification and cloning of the PDH complex-encoding genes.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids used in this study are listed in Table 1.

TABLE 1.

Bacterial strains and plasmids used

Strain or plasmid	Relevant characteristics	Reference or source
Bacterial strains
Z. mobilis ATCC 29191	Wild type	66
E. coli
DH5α	supE lacU Δ(lacZ)M15 hsdR recA endA gyrA thi relA	23
JM109	recA supE endA hsdR gyrA relA thi (Δlac-proAB) F′ traD proAB+lacIq Δ(lacZ)M15	72
M15	recA+ uvr+ F− mtl gal ara lac Rifs Strs Nals	Qiagen
S17-1	RP4-2(Tc::Mu)(Km::Tn7) integrated in the chromosome	63
Plasmids
pUC18, pUC19	2.7 kb; bla, lacZα	68
pBluescript SK	2.96 kb; bla, lacPOZ′, T7 and T3 promoters	Stratagene
pQE70	4.2 kb, bla, T5, C-terminal His6 tag fusion vector	Qiagen
pZY557	10.1 kb; cat lacIq	60
pQE709	1.5-kb SphI/BamHI fragment from Z. mobilis harboring the lpd gene cloned into pQE70	This work
pUCE1	6.3-kb EcoRI fragment from Z. mobilis carrying the complete pdhB gene and part of the lpd gene cloned into pUC19	This work
pUCE3	5-kb EcoRV fragment from Z. mobilis carrying the lpd gene cloned into pUC19	This work
pUN552	4.2-kb fragment carrying the pdhB and lpd gene	This work
pSKU41	3.3-kb EcoRI fragment from Z. mobilis carrying the pdhAα gene and the 5′ part of the pdhAβ gene cloned into pBluescript SK	This work
pSKU13	5-kb EcoRV fragment from Z. mobilis carrying the complete pdhAα and pdhAβ genes cloned into pUC18	This work
pSKU80	15-kb HindIII fragment carrying the pdhA gene region cloned into pBluescript SK	This work

Growth of bacteria.

Z. mobilis ATCC 29191 was grown anaerobically at 30°C in a complex medium (VM) containing 50 g of glucose per liter as described previously (10). For large-scale fermentation, Z. mobilis was grown in minimal medium (12). The Escherichia coli strains listed in Table 1 were grown aerobically at 37°C in Luria-Bertani (LB) medium (40). Antibiotics were added at the following concentrations: ampicillin, 100 μg/ml; kanamycin, 25 μg/ml; chloramphenicol, 25 μg/ml for E. coli and 80 μg/ml for Z. mobilis; and nalidixic acid, 40 μg/ml for Z. mobilis.

DNA isolation, manipulation, and sequencing.

Chromosomal DNA of Z. mobilis was isolated by the method of Byun et al. (14). Plasmid DNA from Z. mobilis and E. coli was prepared by the alkaline extraction procedure (6), modified for Z. mobilis by preincubation with lysozyme for 30 min. The pdhB-lpd gene region was sequenced on the basis of deletion derivatives generated by exonuclease III digestion. For introduction of unidirectional deletions, an Erase-a-Base kit (Promega) was used as instructed by the manufacturer. Double-strand DNA sequencing was performed by the dideoxynucleotide chain termination method of Sanger et al. (58), using a T7 Auto Read sequencing kit and an A.L.F. (automated laser fluorescent) DNA sequencer (Pharmacia). The pdhAαβ gene region was sequenced by Eurogentec (Seraing, Belgium). DNA fragments were isolated from agarose gels by use of a QiaEx kit (Qiagen, Hilden, Germany). All other DNA-manipulating techniques were performed by standard protocols (57).

Hybridization and gene isolation techniques.

For construction of size-selected plasmid libraries, chromosomal restriction fragments of the desired sizes were excised from an agarose gel and purified. The fragments were ligated into pUC18/19 or pBluescript SK (Stratagene, Heidelberg, Germany) vectors, respectively. E. coli DH5α was transformed with the ligation products and plated on LB agar. Colony hybridization and Southern hybridization were performed according to the DIG (digoxigenin) application manual from Boehringer, Mannheim, Germany. Oligonucleotide gene probes were 3′ labeled with DIG-dUTP/dATP tail by terminal transferase. Other hybridization probes were labeled by the random priming technique (19). Probe labeling and chemiluminescent detection were performed with a DIG-DNA labeling and detection kit (Boehringer).

Synthesis of oligonucleotides.

Oligonucleotides were synthesized in 0.2-μmol portions from deoxynucleoside phosphoamidites (15) with a Gene Assembler Plus apparatus (Pharmacia-LKB Biotechnology) as instructed by the manufacturer. Release of oligonucleotides from the support and removal of protection groups was achieved by incubation at 65°C overnight in 32% (vol/vol) ammonia. Oligonucleotides were purified by gel filtration on NAP-10 columns (Pharmacia).

DNA amplification by PCR.

Specific synthesis of DNA fragments by PCR (43) was carried out in a DNA thermal cycler (Perkin-Elmer/Cetus). The reaction mixture contained 200 μM deoxynucleoside triphosphates, 10 μl of reaction buffer, and 5 U of Taq polymerase (all reagents from Boehringer) added with either 1 ng of chromosomal DNA of Z. mobilis and 2 nmol of each degenerate primer or 0.1 pmol plasmid DNA and 10 pmol of each specific primer. The amplification program consisted of 30 cycles each of 1 min at 94°C, 2 min at 45 or 55°C, and 1 min at 72°C. PCR products were purified by using a PCR Clean Up kit (Boehringer).

Purification of lipoamide dehydrogenase.

Lipoamide dehydrogenase of Z. mobilis was purified by the method for soluble His₆-tagged proteins (16), using affinity chromatography with an Ni-nitrilotriacetic acid (NTA)-agarose column and an increasing imidazole gradient (0 to 0.5 M) for elution. Production of a C-terminal His₆-tagged protein was achieved by cloning a 1.5-kb SphI/BamHI PCR fragment, encoding the lpd gene, without stop codon which was amplified with the specific primers P1 and P2 (Table 2) into the vector pQE70, resulting in the hybrid plasmid pQE709. A 1-liter culture of the recombinant strain E. coli M15[pREP4;pQE709] was grown for 5 h in the presence of 2 mM isopropylthiogalactopyranoside (IPTG). After growth, cells were disrupted by sonication and the cell extract was applied on a 5-ml Ni-NTA-agarose column, equilibrated with 50 mM sodium phosphate buffer (pH 8.0), with a flow rate of 0.2 ml/min. Chromatography was performed according to the protocol of Qiagen.

TABLE 2.

Oligonucleotides used in this study

Oligonucleotide for PCR	Sequence	Purpose
L1	5′-GA(A/C/T/G)GC(A/C/T/G)GG(A/C/T/G)CC(C/A/T/G)GG	lpd gene probe
	(G/A/T/C)TA(C/T)GT(C/A/T/G)GC(C/A/T/G)GC-3′
L2	5′-TT(G/C/T/A)(G/C)(T/A)(G/T/A/C)GG(G/A/T/C)AT(A/G)CA	lpd gene probe
	(C/A/G/T)CC(G/C/T/A)AC(G/A)TT(G/A/T/C)A(G/A)(A/G)CA-3′
P1	5′-CCGCATGCTCAGGTGCCGCCGATGAA-3′	lpd His tag
P2	5′-CTGGATCCAAAATGCAGTGCGCGCCC-3′	lpd His tag
J1	5′-GC(AGCT)AT(ACT)GAA(CT)T(AGCT)AAAATG-3′	pdhAβ probe
J2	5′-A(AG)(AGCT)GT(AGCT)CCTTCTTCCAT-3′	pdhAβ probe
J3	5′-GC(AGCT)AA(AG)GC(AGCT)AC(AGCT)CA(AG)GA(CT)-3′	pdhAα probe
J4	5′-GCTTT(AG)TG(AGCT)GG(AGC)C(GT)(AG)TT-3	pdhAα probe

Purification of Z. mobilis PDH complex.

The PDH complex was purified from Z. mobilis by a modification of the method described for the purification of A. vinelandii E2p (25). A 50-liter culture of Z. mobilis was anaerobically grown in minimal medium containing 10% glucose for 20 h at 30°C. Cells (250 g [wet weight]) were suspended in 350 ml of 50 mM potassium puffer (pH 7.0) containing 1 mM EDTA and 1 mM PMSF (phenylmethylsulfonyl fluoride) and disrupted in a French press at 9,000 lb/in². After centrifugation for 30 min at 14,000 rpm, nucleic acids were precipitated by addition of 0.1% (wt/vol) protamine sulfate and discarded after centrifugation. A poly(ethyleneglycol) 6000 (PEG)-MgCl₂ precipitation was carried out in two steps. At 6% (wt/vol) PEG, a large amount of protein precipitated whereas the PDH complex remained in solution. Addition of PEG and MgCl₂ to final concentrations of 10% (wt/vol) and 0.75 mM, respectively, resulted in precipitation of the PDH complex. After centrifugation for 30 min at 20,000 rpm, the pellet was resuspended in 150 ml of 20 mM potassium phosphate buffer (pH 7.0) containing 0.1 mM MgCl₂, 0.1 mM thiamine pyrophosphate 25 μM EDTA, and 50 μM PMSF (standard buffer). The solution was applied to a Q-Sepharose column (350 ml) and eluted with a 0 to 600 mM KCl gradient in standard buffer. Active fractions were concentrated by ultrafiltration (Amicon YM100) and applied to a Sephacryl S400 column (2.6 by 100 cm); 50 mM potassium phosphate standard buffer containing 150 mM KCl was used for separation. The first peak fractions were again concentrated by ultrafiltration, analyzed for PDH complex activity, and by subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Purification of Z. mobilis E2p-E3 subcomplex.

A 3-liter culture of recombinant Z. mobilis [pUN552] was grown overnight in VM medium supplemented with 5 μM lipoic acid, 80 μM chloramphenicol, and 1 mM IPTG. Cells were centrifuged and resuspended in 40 ml of 50 mM potassium phosphate buffer (pH 7.0) containing 1 mM EDTA and 1 mM PMSF. The E2p-E3 subcomplex was purified by the same method as described for the PDH complex except that the PEG precipitation steps were omitted.

Enzyme assays.

Enzyme activities were measured at 25°C, and specific activities are expressed in units/milligram of protein; 1 U is the amount of enzyme transforming 1 μmol of substrate/min. Activity of lipoamide dehydrogenase was measured at 340 nm by formation of NADH as described by Westphal and de Kok (70). The K_m value for NAD was determined in standard buffer with various concentrations of NAD between 12.5 μM and 1 mM.

Activity of dihydrolipoamide acetyltransferase was monitored at 240 nm by the formation of acetyl lipoamide as described by Schwarz and Reed (61). The molar extinction coefficient at 240 nm (ɛ₂₄₀) of acetyl lipoamide is 5 × 10³ M⁻¹ cm⁻¹ (70).

PDH activity was monitored at 600 nm by the reduction of dichlorophenolindophenol (Cl₂Ind) instead of ferricyanide as described by Reed and Willms (53). The ɛ₆₀₀ of Cl₂Ind is 16.1 × 10³ M⁻¹ cm⁻¹.

The overall activity of the PDH complex was measured either by the reduction of ferricyanide at 430 nm as described by Snoep et al. (64) or by the formation of NADH at 340 nm as described by Schwarz and Reed (62), depending on the presence or absence of pyruvate decarboxylase. The ɛ₃₄₀ of NADH is 6.22 × 10³ M⁻¹ cm⁻¹; the ɛ₄₃₀ of ferricyanide is 1.03 × 10³ M⁻¹ cm⁻¹. Protein content was determined by the method of Bradford (9).

SDS-PAGE and protein transfer.

SDS-PAGE was performed by the method of Schägger and von Jagow (59). For N-terminal amino acid sequencing, proteins were transferred to a polyvinylidene difluoride membrane (Millipore) by the semidry blot technique and stained with amido black. N-terminal amino acid sequencing was performed by the method of Edman and Begg (18).

Chemicals.

dl-Dihydrolipoamide was prepared by reduction of dl-lipoamide (Sigma Chemie, Deisenhhofen, Germany) with NaBH₄ (52). All other chemicals were obtained from Sigma or Merck AG (Darmstadt, Germany).

Electron microscopy.

For electron microscopy, a carbon-coated film was treated for 5 to 10 s with a solution of the enriched E2 component, containing 100 μg of protein/ml in 20 mM potassium phosphate buffer (pH 7.0). The carbon film with the adsorbed particles was rinsed with H₂O and then treated with a 3% solution of sodium phosphotungstate (pH 7.0) until the carbon film was totally floated (39). The negatively stained probe on the carbon film was then applied to a grid, and the residual fluid was removed by filter paper. The electron micrographs were taken with a Philips EM301 microscope at a primary magnification of ×33,400.

Nucleotide sequence accession number.

The nucleotide sequences reported in this paper have been submitted to the GenBank/EMBL data bank and assigned accession no. X93605 and Y12884.

RESULTS

Isolation and nucleotide sequence analyses of the pdhB and lpd genes.

To isolate the genes encoding the PDH complex of Z. mobilis, two degenerate lpd-specific primers (L1 and L2 [Table 2]) were synthesized on the basis of the consensus amino acid sequence of the highly conserved N terminus of lipoamide dehydrogenases from several species. These primers were used in the PCR to amplify a homologous 129-bp lpd fragment from Z. mobilis chromosomal DNA. An internal 45-bp oligonucleotide of this PCR fragment was used as a gene probe for hybridization analysis. Z. mobilis chromosomal DNA was restricted with various endonucleases and hybridized with the lpd probe. Size-selected plasmid libraries of desired restriction fragments were constructed and screened by colony hybridization. Two hybrid plasmids, pUCE1 and pUCE3, harboring a 6.3-kb EcoRI fragment and a 5-kb EcoRV fragment, respectively, were selected in order to clone the complete pdh gene region. The two fragments overlapped in a region of 1,500 bp in which the hybridization site of the lpd probe was localized.

Nucleotide sequence analysis of the complete 6.3-kb EcoRI fragment and the adjacent region of the 5-kb EcoRV fragment revealed several possible open reading frames (ORFs). The deduced amino acid sequence of two ORFs (1,323 and 1,401 bp) exhibited significant similarities to the E2p and E3 components, respectively, of PDH multienzyme complexes from various sources. In analogy to the related genes of other species, the 1,323-bp ORF was referred to as pdhB and the 1,401-bp ORF was referred to as lpd from Z. mobilis (Fig. 1B). Both genes could be functionally expressed in E. coli. Recombinant E. coli JM109 strains carrying a plasmid with the pdhB or lpd gene under the control of lacZ showed increased E2p and E3 enzyme activities, respectively. Another ORF, ORF2, encoding a protein of 149 amino acids was localized between the structural genes pdhB and lpd. An ORF between the pdhB and lpd genes was also identified in the pdh gene clusters of Alcaligenes eutrophus (pdhA-pdhB-ORF3-pdhL) and Neisseria meningitidis (E1p-E2p-ORF3-E3). However, no similarity could be detected between ORF2 of Z. mobilis and ORF3 of A. eutrophus and N. meningitidis. The amino acid sequence deduced from ORF2 revealed up to 41% amino acid identity (65% similarity) to the P14 gene of E. coli (48). The function of this ORF, however, remains unclear. We could not identify any promoter-like structures upstream of pdhB or upstream of lpd. Therefore, we have no indication of whether the pdhB and lpd genes are transcribed as a single operon or if the lpd gene is expressed from its own promoter. At a distance of 253 bp downstream of lpd, we identified an inverted repeat (ΔG = −12.4 kJ/mol) which might be a rho-independent transcription terminator.

FIG. 1.

Molecular organization of the Z. mobilis pdhAα-pdhAβ gene region (A) and the pdhB-ORF2-lpd gene region (B) encoding the E1p, E2p, and E3 components of the PDH complex. The positions of identified genes are indicated by arrows. Double-headed arrows indicate the DNA fragments isolated from plasmid libraries. Relevant restriction sites: B, BamHI; E, EcoRI; N, NcoI; M, MunI; P, PstI; R, EcoRV.

The lpd gene encodes a protein of 466 amino acids with a calculated mass of 49.8 kDa. The predicted amino acid sequence was closely related to other lipoamide dehydrogenases with up to 54% identity to the E3 component of the acetoin dehydrogenase complex of Klebsiella pneumoniae (46) and 40% identity to the Lpd protein of Pseudomonas fluorescens (5). The sequence contained the characteristic motifs of flavin-containing disulfide oxidoreductases (71) (Fig. 2B). This includes the flavin adenine dinucleotide and NAD binding sites as well as residues Cys 41 and Cys 46, which build the redox-active disulfide bridge involved in electron charge transfer with flavin adenine dinucleotide (31), and the conserved residues His 444 and Glu 449 in the interface domain, which possibly function as the electron donor-acceptor couple as shown by site-directed mutagenesis of the human protein (32).

FIG. 2.

Relevant portions of the nucleotide and deduced amino acid sequences of the PDH complex encoding genes including ORF2. The nucleotide sequences of two separate gene regions, pdhAα-pdhAβ (A) and pdhB-ORF2-lpd (B), are given in the 5′-3′ direction, each starting with the nucleotide 1. Putative ribosome binding sites are shown in boldface; functional domains identified in homologous proteins are underlined and indicated below the amino acid sequences. Dotted lines indicate gaps in the nucleotide sequence. TPP, thiamine pyrophosphate.

The pdhB gene codes for a protein of 441 amino acid residues, corresponding to a protein of 46.8 kDa. The deduced amino acid sequence shows high identity to the sequences of E2 components of eukaryotic PDH complexes, with 42.5% identity to Saccharomyces cerevisiae (45) and 43% identity to Rattus norvegicus (38) and Arabidopsis thaliana (21) but only low sequence identity to the E2 subunits of E. coli (29%) (65) and A. vinelandii (28%) (24) PDH complexes. The overall close relationship of Z. mobilis E2p with dihydrolipoamide acetyltransferases of eukaryotic species is shown in a phylogenetic tree, calculated from progressively aligned sequences (Fig. 3). The amino acid sequence of Z. mobilis E2p shows the characteristic multidomain structure of dihydrolipoamide acetyltransferases (47), containing an amino-terminal lipoyl domain (residues 1 to 84), a subunit binding domain (145 to 190), and a C-terminal domain (215 to 440) (Fig. 2B). We found a number of conserved amino acids in the lipoyl domain and the C-terminal domain, including the active-site motif HXXXDG common to all E2 enzymes, as well as the substrate-specific residues of acetyltransferases K325, Q352, and F369 (55). The E2p of Z. mobilis exhibits some interesting conspicuous features. In contrast to the E2p components of all known gram-negative bacteria, which possess two or three lipoyl domains, the E2p of Z. mobilis contains only a single lipoyl domain. The N terminus of this lipoyl domain contains the characteristic P(S/A)LSPTM sequence, which is a highly conserved motif common to eukaryotic lipoyl domains of E2p and protein X components of PDH complexes (44). The linker sequence connecting the lipoyl domain to the subunit binding domain is unusually long with approximately 60 amino acid residues. The amino acid composition of the linker segment shows a high proportion of charged amino acid residues, such as aspartate, glutamate, glutamine, lysine, and serine, but is deficient in proline.

FIG. 3.

Phylogenetic tree of dihydrolipoamide acetyltransferases from prokaryotic and eukaryotic sources. The branching order and distance score were calculated by the program TREE as described by Feng and Doolittle (20).

In all gram-negative and gram-positive bacteria so far studied, the genes encoding the substrate-specific E1p and E2p components of the PDH complex are clustered such that the gene for E1p is located next to and upstream of the gene for E2p. Since the same gene organization was expected for Z. mobilis E1p, the nucleotide sequence of 4 kb upstream of the pdhB gene was analyzed (data not shown). Surprisingly, no similarity to PDH (E1p component) could be found within this sequence. To exclude cloning effects, we verified by hybridization experiments that the cloned 6.3-kb EcoRI and 5-kb EcoRV fragments represented the original gene organization of the Z. mobilis genome (data not shown). From these data, we conclude that the E1p-encoding gene of Z. mobilis is located on the chromosome in a region separated from the sequenced gene locus of pdhB and lpd genes described here.

Genetic approaches to isolate the E1-encoding gene of Z. mobilis PDH complex by the use of heterologous gene probes from E. coli (aceE) or A. eutrophus (pdhA) failed. PCR with degenerate primers corresponding to conserved regions of E1, a strategy similar to that used for lpd probing, was likewise unsuccessful. Therefore, an approach using the purified enzyme was chosen for cloning of pdhA.

Purification of Z. mobilis PDH complex.

The PDH complex from Z. mobilis was purified to obtain information about the subunit composition and to determine whether E1 was a homodimer (α₂) or a heterotetramer (α₂β₂). Furthermore, N-terminal sequencing of the protein components should serve to generate an E1-specific gene probe by PCR with degenerate primers deduced from the obtained amino acid sequence. The purification procedure (see Materials and Methods) allowed isolation of 1 mg of the PDH complex from 250 g of cell paste, with a yield of 9% (Table 3). The purification factor of about 1,200 reflects that the PDH complex is an anabolic enzyme in Z. mobilis, present at a low activity level in the cell. The specific activity of the purified PDH complex was 12.6 U/mg of protein. The specific activities for the E1p, E2p, and E3 components of the complex were 0.12, 5.6, and 39.8 U/mg of complex protein, respectively. By SDS-PAGE, the PDH complex was found to consist of four polypeptide chains, corresponding to apparent molecular masses of 56, 50, 48, and 38 kDa (Fig. 4). By N-terminal amino acid sequencing, these bands could be assigned to PDH E1β subunit (56 kDa; approximately 15 kDa larger than other E1β subunits), lipoamide dehydrogenase or E3 component (50 kDa), dihydrolipoamide acetyltransferase or E2 component (48 kDa), and PDH E1α subunit (38 kDa). The existence of a heteromeric E1p component in the PDH complex of Z. mobilis was surprising, since this subunit composition is usually found in the PDH complexes of gram-positive bacteria and eukaryotes but not in gram-negative bacteria. PDH complexes of other gram-negative bacteria contain homodimeric E1p components with a molecular mass of approximately 90 kDa per monomer (37). N-terminal sequencing of the E1α and E1β subunits resulted in amino acid sequences of 19 (AKATQDSNRPHKA[D]VT[S]AI) and 20 (AIELKMPALSPTMEEGTLTR) residues, respectively. The N-terminal amino acid sequences of the E2p and E3 components of the Z. mobilis PDH complex were identical to the amino acid sequences deduced from the nucleotide sequences of the pdhB and lpd genes. This result confirmed that the cloned genes encode the functionally active E2p and E3 components of the PDH complex.

TABLE 3.

Purification of the PDH complex of Z. mobilis

Step	Total activity (U)	Total protein (mg)	PDH complex sp acta (U/mg)	Purification (fold)
Cell extract	155	15,500	0.01	1
Protamine sulfate	150	15,030	0.01	1
6% PEG	220	7,350	0.03	3
10% PEG, 0.75 mM MgCl2	159	1,980	0.08	8
Anion-exchange column	62	60	1.04	104
Gel filtration	13.9	1.1	12.6	1,260

^aPDH complex activity was measured in cell extract and after the first two purification steps by the reduction of ferricyanide at 430 nm, because direct measurement of NADH formation by the PDH complex at 340 nm was obscured due to the reaction of pyruvate decarboxylase, which caused a rapid decrease in pyruvate concentration in the test mixture and a concomitant decrease in extinction (the ɛ₃₄₀ of pyruvate is 0.023 mM⁻¹ cm⁻¹). The enzyme activities at further purification steps were determined by the formation of NADH. 

FIG. 4.

SDS-PAGE analysis of the purified PDH complex of Z. mobilis. Purified PDH complex was subjected to electrophoresis on an SDS–10% gel and visualized with Coomassie blue R250. PDHC, purified PDH complex (6 μg); SDS-7B, prestained SDS-7B marker (Sigma). Sizes are indicated in kilodaltons.

Isolation and sequence analyses of the pdhAα and pdhAβ genes.

For pdhAα and pdhAβ gene probe constructions, two degenerate oligonucleotide mixtures (Table 2), corresponding to the known E1α and E1β amino-terminal peptide sequences, were used as primers in a PCR to provide nondegenerate probes for the detection of pdhAα and pdhAβ. The template for the amplification was Z. mobilis total DNA. The expected PCR products of 38 bp (E1α) and 53 bp (E1β) were cloned into the SmaI site of pUC18 and transformed to E. coli. The nucleotide sequences obtained from plasmid DNA of several recombinant clones matched in both cases the amino acid sequence obtained from protein sequencing. Corresponding oligonucleotides were DIG labeled for hybridization experiments. After Southern hybridization with Z. mobilis chromosomal DNA, treated with various restriction enzymes, size-selected plasmid libraries of EcoRI, EcoRV, and HindIII fragments were constructed in pBluescript SK or pUC18 and screened with the E1α and E1β gene probes. Three hybrid plasmids, plasmid pSKU41, carrying a 3.3-kb EcoRI fragment, plasmid pSKU13, carrying a 5-kb EcoRV fragment, and pSKU80, which harbored a 15-kb HindIII fragment, were isolated from positive clones (Fig. 1A).

Nucleotide sequence analysis of 3 kb of the EcoRV fragment revealed two ORFs with high similarity to the α and β subunits of heterotetrameric E1 components of PDH complexes (Fig. 1A). The first ORF (1,065 bp), named pdhAα, encoded a polypeptide of 354 amino acids, corresponding to a protein of 38.6 kDa, with highest sequence identity to the E1α subunit of the PDH complexes of humans (47% identity [30]) and S. cerevisiae (46.8% identity [4]). The predicted amino acid sequence contained the thiamine pyrophosphate binding site (Fig. 2A) (26) involved in binding the metal ion and the diphosphate group (35, 42). The second ORF (1,389 bp) started 2 bp downstream from the pdhAα gene. This ORF, named pdhAβ, encoded a polypeptide of 462 amino acids, corresponding to a protein of 49.8 kDa. The main part of the predicted polypeptide (amino acids 110 to 462) was closely related to the E1β subunits of PDH complexes of Arabidopsis thaliana (58% identity [36]) and S. cerevisiae (56% identity [41]). The unusual extension at the N terminus (amino acids 1 to 80) of the E1β subunit of Z. mobilis was identified as a lipoyl domain connected by a linker segment. This lipoyl domain contained the conserved lysine residue as a potential lipoylation site. The lipoyl domain of E1β of the Z. mobilis PDH complex showed about 72% identical amino acid residues to the lipoyl domain of its E2p subunit (Fig. 5), including the sequence motif PALSPTM. To the best of our knowledge, this is the first report of a PDH (E1β subunit) with an amino-terminal lipoyl domain.

FIG. 5.

Amino acid alignment of the lipoyl domains of the E1β and E2p components of the PDH complex of Z. mobilis. (a) Amino-terminal sequence of the E1β component; (b) amino-terminal sequence of the E2p component. The PALSPTM motif conserved within lipoyl domains of eukaryotic E2p and protein X components and the highly conserved lipoylation site are boxed.

The gene encoding enolase (eno) of Z. mobilis was detected by partial sequencing of the 5-kb insert of pSKU13 500 bp upstream of pdhAα (Fig. 1A). The eno gene was previously described by Burnett et al. (13). The pdhAαβ gene locus was further characterized by a detailed restriction map of the 15-kb HindIII fragment of pSKU80. This HindIII fragment encompasses a genomic region expanding approximately 7 kb upstream and 5 kb downstream of the pdhAα and -β genes. To gather information about the minimal distance between the pdhAα-pdhAβ and pdhB-ORF2-lpd gene regions, cross-hybridization experiments were performed with the 15-kb HindIII fragment, carrying the pdhAα and pdhAβ genes, and the 6.3-kb EcoRI and 5-kb EcoRV fragments, carrying the pdhB and lpd genes. No cross-hybridization reaction could be analyzed with a pdhA- or lpd-specific probe or with a probe (0.8-kb EcoRI/XbaI fragment) corresponding to the upstream region of pdhB. From these results, we conclude that the minimal distance between the two pdh gene regions of Z. mobilis extends beyond 7 kb.

Electron microscopy of Z. mobilis dihydrolipoamide acetyltransferase.

The fact that the PDH complex of Z. mobilis consisted of the subunits E1α, E1β, E2, and E3 strongly suggested a 60-mer structural core of this complex with an icosahedral symmetry, as found in all PDH complexes with this subunit composition. In addition, the amino acid sequence of the Z. mobilis E2p component possesses characteristic sequences which are usually found in PDH complexes with icosahedral symmetry. To elucidate the quaternary structure of the PDH complex from Z. mobilis, electron microscopic studies on the structural core-forming E2p component were performed. The enzyme was purified from cell extracts of recombinant Z. mobilis strains carrying plasmid pUN552, which carried the pdhB and lpd genes, resulting in copurification of E2 and E3 components of the PDH complex. In this enzyme probe, the specific activity of E2 was 7.8 U/mg of protein and the specific activity of E3 was 80 U/mg of protein. Projected electron microscopic images of this specimen are shown in Fig. 6. Most of the E2-E3 subcomplexes appeared to be dissociated during specimen preparation. Individual images of the negatively stained particles were visible and showed symmetrical views of the inner core with the characteristic patterns of the five-, three-, and twofold symmetries of a pentagonal dodecahedron. The experimentally determined structures were compared with the three-dimensional views of a computer-generated model of a pentagonal dodecahedron. Arrows indicate views with good coincidence with the projections along the five-, three-, and twofold symmetry axes (Fig. 6). These electron microscopic images confirmed that the PDH complex of Z. mobilis is arranged with icosahedral symmetry.

FIG. 6.

(A) Electron micrographic images of the inner core component of the Z. mobilis PDH complex (field of negatively stained E2p complexes with phosphotungstate [pH 7.0]). Views of the fivefold (a), twofold (b), and threefold (c) symmetry axes are indicated with arrows. The scale bar denotes 100 nm. (B) Computer-generated structural models of the five-, two-, and threefold symmetry axes of a pentagonal dodecahedron.

Purification and characterization of Z. mobilis LPD.

The lipoamide dehydrogenase (LPD) of Z. mobilis was purified for biochemical characterization via a simple purification protocol using a C-terminal His-tagged protein. For this purpose, the LPD was expressed from plasmid pQE709 in E. coli M15[pREP4] cells. The recombinant tagged LPD protein was purified to homogeneity in a single step by affinity chromatography on Ni-NTA-agarose. The specific activity of purified LPD was determined to 260 to 280 U/mg of protein, suggesting that the His tag had no effect on enzyme properties. The fluorescence spectrum of LPD taken from 300 to 600 nm shows a maximum at 457 nm, which is a typical feature of flavin-containing lipoamide dehydrogenases (71). With respect to the proposed, exclusively anabolic role of the PDH complex in Z. mobilis, we tested if the LPD used NAD, NADP, or both coenzymes as electron acceptors. The reduction of either NAD or NADP by purified LPD was monitored at 340 nm in the standard assay. The LPD enzyme transferred electrons from dihydrolipoamide selectively only to the coenzyme NAD. The K_m of LPD for NAD was 135 (±10) μM. A similar value (140 μM) was reported for the E. coli LPD enzyme (7). Thus, the anaerobic bacterium Z. mobilis obviously does not need an LPD enzyme with a higher affinity for NAD compared to an organism which uses the PDH complex mainly in aerobic metabolism.

DISCUSSION

This report shows that the PDH complex of the gram-negative bacterium Z. mobilis possesses a subunit composition different from that found in other gram-negative bacteria. The E1 component of the Z. mobilis PDH complex is composed of the heteromeric subunits E1α (38 kDa) and E1β (56 kDa), thus resembling the heterotetrameric E1α₂β₂ components of the PDH complexes of gram-positive organisms and eukaryotes. Electron microscopic analysis of the core complex of the Z. mobilis PDH revealed a pentagonal dodecahedron-like structure, in good agreement with the finding of Reed and Hackert that PDH complexes consisting of a four subunits (E1α, E1β, E2, and E3) aggregated with icosahedral symmetry (54). In these cases, the E2 core consisted of 60 molecules, forming a pentagonal dodecahedron, with around 30 E1 tetramers and 6 E3 dimers bound noncovalently to the edges and faces of the E2 core (54). Interestingly, a phylogenetic tree of E2 components reflected a structural relationship and protein sequence similarity of the PDH complex of Z. mobilis with the corresponding complexes of eukaryotic species. Z. mobilis E2p is most related to the E2p components of yeasts (S. cerevisiae and Neurospora crassa) among eukaryotes and less related to other gram-negative bacteria.

Two components of the PDH complex of Z. mobilis were found to have an N-terminal lipoyl domain. Amino-terminal lipoyl domains are a common feature of E2 components of 2-oxo acid dehydrogenase complexes, harboring a covalently bound lipoyl cofactor which functions as an intermediate carrier to couple the activities of the separate multienzyme components. A striking difference of E2 components of PDH complexes is the number of lipoyl domains per E2 chain, which varies from one to three, depending on the species. The dihydrolipoamide acetyltransferase of Z. mobilis contains a single amino-terminal lipoyl domain, as is the case for many other organisms, whereas the PDH contains a lipoyl domain at the N terminus of the E1β subunit, a unique feature among PDHs studied so far. However, lipoyl domains have not been found only as parts of E2 components. The protein X components of eukaryotic PDH complexes possess an N-terminal lipoyl domain (3). In addition, lipoyl domains were found to be connected to the N termini of E3 components of the PDH complexes from A. eutrophus (27), N. meningitidis (1, 17) and Mycoplasma capricolum (73) and the E3 component of the acetoin dehydrogenase enzyme system from Clostridium magnum (34). It was shown that the lipoyl domain of protein X can function in the overall complex reaction (51). In contrast to this, the role of lipoyl domains as part of E3 components is not yet known, but participation in the overall reaction was suggested (73). Multiple lipoyl domains in different complex components may provide extra lipoyl cofactors that could participate in catalysis and therefore improve specific activity. In contrast to this, the function of multiple lipoyl domains in a single E2 chain, studied for the E. coli PDH, is probably to extend the reach of the outermost lipoyl cofactor and improve the conformational mobility in order to facilitate substrate transfer between the active sites (22).

As described in this study, we have cloned the PDH complex encoding genes (pdhAα, pdhAβ, pdhB, and lpd) of Z. mobilis. The organization of these genes is atypical in that in the chromosome of Z. mobilis, pdhAα, pdhAβ, and pdhB, encoding the substrate-specific E1p and E2p components, are not clustered as are all other prokaryotic pdh genes. In Z. mobilis, the pdhA genes were located approximately 500 bp downstream of the eno gene. The pdhB gene was not located adjacent to the pdhA genes. In contrast to this, the pdhB gene was identified upstream of the lpd gene, and a 450-bp ORF of unknown function was found between the two genes. Hybridization experiments and sequence analysis revealed a minimal distance between the pdhAα-pdhAβ and the pdhB-ORF2-lpd gene loci of 7 kb. Since no genetic map of the Z. mobilis chromosome is available, we could not determine the relative localization of the pdh gene loci. Although the pdh genes are separated in the Z. mobilis chromosome, they encode the physiologically relevant enzyme components of the active PDH complex. However, the unusual organization of the pdh genes in Z. mobilis raises questions as to how transcription is controlled. The pdhAα and pdhAβ genes were separated by only 2 bp, suggesting that they are transcribed in a single operon. The long region between the eno and the pdhAα genes and the existence of a strong eno promoter (in contrast to the low expression of pdh genes) and a terminator-like structure downstream of eno (13) suggested a transcription start site upstream of pdhAα. No indication exists for the mode of transcription of the pdhB-ORF2-lpd cluster. However, since the stop codon for pdhB and the start codon for ORF2 overlapped by 4 bp, it is suggested that these genes are transcribed together.