The lipoamide dehydrogenase gene (lpdA) encoding the E3 subunits of both the pyruvate dehydrogenase and 2-oxoglutarate dehydrogenase complexes of Escherichia coli, is expressed from the upstream pdh and internal lpd promoters of the pdh operon (pdhR-aceEF-lpdA). Under aerobic conditions, the specific components of the 2-oxoglutarate dehydrogenase complex encoded by the sucAB genes in the sdhCDAB-sucABCD operon are expressed from the sdh promoter. The provision of lipoamide dehydrogenase subunits for assembly into the 2-oxoglutarate dehydrogenase complex could thus be controlled by co-regulation of the lpd promoter with the sdh promoter. Here, the transcription start point of the lpd promoter was defined by primer extension analysis, and an ArcA binding site, TGTTAACAAT, overlapping the lpd promoter and matching the consensus at 8 out of 10 positions, was identified by in vitro footprint analysis. PdhR was not bound to the lpd promoter nor was ArcA bound specifically to the pdh promoter. These results support the view that co-regulation of the lpd and sdh promoters is mediated primarily by ArcA.
The pyruvate dehydrogenase (PDH) complex of Escherichia coli contains multiple copies of three subunits: pyruvate dehydrogenase (E1p); lipoate acetyltransferase (E2p); and lipoamide dehydrogenase (E3). The corresponding genes are located at 2.7 min in the pdh operon (pdhR-aceE-aceF-lpdA; Fig. 1) which contains two major promoters: Ppdh, generating a 7.4-kb pdhR-lpdA readthrough transcript; and Plpd, which generates an independent 1.7 kb lpdA transcript [1, 2]. The pdh promoter is negatively autoregulated by PdhR, which binds to a specific operator sequence in the absence of pyruvate . The genes encoding the specific dehydrogenase (E1o, sucA) and succinyltransferase (E2o, sucB) subunits of the analogous 2-oxoglutarate dehydrogenase (ODH) complex are located at 16.3 min in the sdh-suc operon (sdhCDAB-sucABCD) which also encodes succinate dehydrogenase (SDH; sdhCDAB) and succinyl-CoA synthetase (sucCD). These genes are expressed primarily from the sdh promoter, but there is an internal suc promoter which seems to be responsible for the anaerobic and stationary phase expression of the sucABCD genes [3, 4]. The single lpdA gene is responsible for providing E3 subunits for assembly into both the PDH and ODH complexes. These complexes are differentially regulated and the mechanism which co-ordinates the synthesis of the E3 subunits with the individual requirements of each complex emerged from studying transcript synthesis under different growth conditions . Thus it appeared that for synthesis of the PDH complex, the lpdA gene is expressed together with the aceEF genes from Ppdh as the distal gene of the pdh operon, whereas for synthesis of the ODH complex it is expressed from the independent lpd promoter which in turn is co-regulated with the sucA and sucB genes, now known to be expressed from Psdh[3, 4].
Enzymological studies with an arcA mutant originally showed that the PDH and ODH complexes (and SDH) are anaerobically repressed by ArcA . In the case of the PDH complex, this could mean that either one or both of the two relevant promoters (Ppdh and Plpd) might be regulated by ArcA. Subsequent studies with lacZ fusions indicated that Ppdh is unaffected by arcA mutation whereas Plpd activity is anaerobically derepressed by arcA mutation . The latter promoter (Plpd) has also been shown to be activated by Fis and repressed by glucose via an ill-defined CRP-independent mechanism . In common with Plpd, the sdh promoter has been shown to be anaerobically repressed by ArcA, aerobically activated by Fis, and subject to CRP-independent repression by glucose , although there are also reports that Psdh is repressed by Fis  and subject to CRP-dependent glucose repression . Thus, in the absence of glucose, ArcA and Fis are the most plausible candidates for mediating the co-regulation of lipoamide dehydrogenase and 2-oxoglutarate dehydrogenase synthesis via the respective lpd and sdh promoters. This has been investigated here by mapping the start point of lpd transcription, and by in vitro studies on the binding of ArcA, Fis and PdhR to the lpd promoter, and of ArcA to the pdh promoter.
2Materials and methods
Bacterial strains and plasmids
E. coli DH5α was the routine transformation host and W3110 (prototroph) was used as the RNA source. Promoter-containing DNA was obtained from two plasmids, pGS698 for Ppdh and pGS704 for Plpd(Fig. 1). Two plasmids, pGS1146 and pGS1147, were constructed by sub-cloning the 0.55-kb RsaI fragment containing the lpd promoter from pGS704 into the SmaI site of pUC119 and defining the orientation of the insert by locating the asymmetric BglI site (Fig. 1).
Microbiological methods, DNA manipulation, protein purification and DNase I footprinting
Strains were cultured aerobically in L-broth with ampicillin (100 μg ml−1) where required, and DNA was isolated and manipulated by standard methods . His6-ArcA protein was purified and phosphorylated as described previously  and Fis protein was prepared from an amplified source, RJ2465 . The PdhR protein was isolated from BL21DE3(pGS680) by a modification of the published method . Cultures in L-broth were induced with IPTG (0.1 mg ml−1) for 90 min after growth to OD600=0.6. PdhR was first enriched from cell-free extracts by heparin-agarose chromatography, applying the samples in buffer A (20 mM Tris-HCl, 1 mM EDTA, 1 mM MgCl2, 1 mM sodium azide, 1 mM DTT, 0.1 mM PMSF, 10% glycerol, final pH 7.3) and eluting with an ammonium sulfate gradient (0–1 M in buffer A). The ammonium sulfate concentration of fractions containing PdhR were increased from 0.45 to 2.0 M for Phenyl Superose reverse-phase chromatography with an ammonium sulfate gradient (2-0 M in buffer A). The PdhR protein was eluted at 1.2 M and diluted for use without removing the salt.
For DNase I footprinting , 2-pmol samples of DNA (end-labelled with appropriate [α-32P]dNTP) were incubated for 10 min at 30°C with 0.5–4.0 μM phosphorylated ArcA, 4 μl of 5× bandshift buffer (0.5 M Tris-Cl pH 7.4, 0.5 M KCl, 50 mM MgCl2, 50% glycerol, 10 mM DTT; ), 10 μg BSA and 0.2 μg sheared calf thymus DNA (20 μl final volume), prior to adding MgCl2 (8 mM, final concentration) and digesting with 3 μl DNase I (10 U/ml; Boehringer Mannheim) for 1 min at 30°C. After ethanol precipitation, with 20 μg glycogen as carrier, samples were washed with 70% ethanol, dried under vacuum and fractionated in a 5% acrylamide 7 M urea sequencing gel, and then analysed by autoradiography.
RNA extraction and primer extension analysis
The hot acid phenol procedure  was used to extract RNA from exponential phase cultures (OD600=0.6) that had been rapidly cooled to 4°C in liquid N2. The method used for primer extension analysis  was modified to allow continuous incorporation of [α-32P]dCTP . Samples of total RNA (100 μg) from cultures of W3110 harvested at OD600=0.5–0.6, were used with 10 pmol primer and fractionated after processing by electrophoresis in 6% acrylamide/7 M urea gels alongside sequence ladders derived from the corresponding DNA and primer. The oligonucleotide primers for analysing lpd transcripts were: S560 (5892–5868) and S561 (6017–5994); co-ordinates based on GenBank V01498.
Results and discussion
Identification of the start points of lpd mRNA by primer extension analysis
Primer extension analysis with RNA from mid-exponential phase cultures of E. coli strain W3110 was used to obtain a precise identification of the start point of lpdA transcription (Fig. 2a). Transcription was shown to start primarily at the first of two adjacent G nucleotides (co-ordinates 5808 and 5809; GenBank V01498) just 7 bp downstream from the putative −10 motif of the lpd promoter (Fig. 2 ai; Fig. 3). The same start point was identified with a second oligonucleotide that primes from a different site (not shown). This start point is more optimally placed than the T and G nucleotides located 3 and 5 bp further downstream, that were tentatively assigned in previous studies using a less precise S1 mapping procedure . Four larger extension products starting at four consecutive nucleotides, TCTC (co-ordinates 5718–5721; GenBank V01498), were also detected in the same primer extension reactions (Fig. 2 aii). They correspond to mRNA molecules that start 8–11 bp downstream of a potential stem-loop, located immediately distal to the aceF coding region, and continue through the lpd promoter and the lpdA coding region to the lpd terminator (Fig. 2 aii; Fig. 3). These transcripts probably arise by endonucleolytic cleavage of the major pdhR-lpdA readthrough transcript (Fig. 1). Evidence for such processing in the aceF-lpdA intergenic region was observed previously  when mRNA molecules terminating at three sites lying within or close to the stem-loop (marked by filled circles in Fig. 3) were detected by S1 mapping. At the time, these mRNA molecules were thought to be formed by transcription termination immediately downstream of the aceF gene. However, no major transcripts terminating in the aceF-lpdA intergenic region can be detected by Northern blotting . The possibility that the longer lpdA transcripts are generated by a second lpd promoter, located within the sequence encoding the stem-loop, seems unlikely because this sequence lacks typical promoter elements. Nevertheless, the existence of such a promoter has not been ruled out.
Studies on the binding of ArcA, Fis and PdhR at the lpd promoter
DNase I footprinting with phosphorylated ArcA was used with two end-labelled 573 bp EcoRI-BamHI fragments from pGS1146 and pGS1147 (Fig. 1) to detect potential ArcA binding sites associated with the lpd promoter (Fig. 2b). A single site of protection was detected with both strands (Fig. 2b), the protection being typically stronger with the coding strand (Kd=2.0 μM) than the non-coding strand (Kd=2.7 μM). The footprints of the coding strand identify a hyper-sensitive site just upstream of the ArcA binding site (Fig. 2b), suggesting that the conformation of the DNA is perturbed when the regulator is bound. The protected region overlaps the −10 and −35 hexamers of Plpd, consistent with ArcA being an anaerobic repressor of lpdA gene expression . Furthermore, the protected region contains a sequence that matches the ArcA binding-site consensus, WGTTAATTAW , at 8 out of 10 positions (Fig. 3). The presence of a single ArcA site at the lpd promoter contrasts markedly with the presence of four ArcA binding sites associated with the sdh promoter . It is also interesting that the ArcA-site in the lpd promoter is located in an 18-bp palindromic sequence, AAAATTGTTAACAATTTT (Fig. 3), which is unique in the E. coli genome. There are eight closely related sequences, each having three mismatches, and two such sequences located upstream of the dnaK and yaaI genes exhibit hyphenated dyad symmetry of the type: AAAATTGnnnnCAATTTT.
Purified Fis protein (≥1.5 nM) retarded the same 573 bp lpd promoter fragment in gel-retardation assays. However, no protection could be detected by DNase I footprint analysis using up to 30 nM Fis (data not shown). This suggests that the observed retardation is probably due to non-specific Fis binding and it was concluded that the effects of Fis on lpdA expression may be indirect.
Evidence that the pyruvate-sensing repressor (PdhR) might regulate Plpd came from the 40% increase in β-galactosidase activity observed when an lpdA-lacZ fusion lacking the proximal segment of the pdh operon, is expressed in a pdhR deletion strain . However, purified samples of PdhR protein, which were shown to bind specifically to Ppdh DNA in gel retardation and footprinting studies, had no effect on the lpd promoter fragment in either gel retardation assays or footprint analysis under comparable conditions (data not shown). This is consistent with the absence of a potential PdhR binding site in the lpd promoter region. These observations suggest that the increase in independent lpdA-lacZ expression in the PdhR-deficient strain is an indirect effect and it was therefore concluded that PdhR controls the synthesis of the PDH complex solely by regulating Ppdh activity.
Attempts to demonstrate ArcA-binding at the PdhR promoter
Studies with pdhR-lacZ fusions indicated that Ppdh is not regulated by ArcA  even though a 1.7-fold anaerobic derepression of pyruvate dehydrogenase synthesis was observed in an arcA mutant . The 541 bp BamHI–HindIII fragment of pGS680 containing the pdh promoter (Fig. 1) was accordingly labelled with [α-32P]dGTP and used in gel retardation assays and footprint analysis. Phosphorylated ArcA retarded the Ppdh fragment, but only in direct proportion to the protein concentration, indicating that the binding is non-specific. There are three sequences within ±50 bp of the pdh transcription start point that match the ArcA-site consensus at 7 or 8 out of 10 positions. However, footprinting studies showed that the pdh promoter region is not protected by phosphorylated ArcA (data not shown). Therefore, based on this in vitro evidence, it was concluded that Ppdh is not regulated by ArcA. Presumably the increase in PDH activity observed in the arcA mutant is a secondary consequence of the arcA mutation.
Previous studies with an lpdA-lacZ fusion lacking the upstream genes have shown that lpd gene expression responds mainly to ArcA-mediated anaerobic repression, Fis-mediated activation under aerobic conditions, and CRP-independent repression by glucose . In the present in vitro studies, no specific binding site was detected for Fis, suggesting that its effect is indirect. However, a single binding site for phosphorylated ArcA, overlapping the −35 and −10 hexamers, was identified in the lpd promoter. This strongly suggests that in response to anaerobiosis, the formation and subsequent binding of phosphorylated ArcA simply blocks RNA polymerase binding and hence represses independent transcription of the lpdA gene. It was thus concluded that the co-expression of lipoamide dehydrogenase and 2-oxoglutarate dehydrogenase involves co-regulation of the lpd and sdh promoters, mediated primarily by ArcA and an ill-defined CRP-independent mechanism of glucose repression. PdhR, the pyruvate-sensitive repressor that crucially controls the synthesis of the PDH complex via Ppdh, also failed to bind to Plpd, indicating that although PdhR regulates the synthesis of lipoamide dehydrogenase via Ppdh, it does not regulate Plpd. The in vitro studies also confirmed that Ppdh activity is not regulated by ArcA.
We thank Dr. Ed Lin for helpful advice. The work was supported by a project grant from The Wellcome Trust (J.R.G.), a research fellowship (B-PD 11474-301) from the Swedish National Science Research Council (D.G.), and a BBSRC Advanced Fellowship (J.G.).