The hypA-E operon is involved in the maturation of all three NiFe hydrogenases in Escherichia coli. Two hyp promoters have been described; a σ54-dependent promoter upstream of hypA, and a σ70-dependent promoter (PhypA) within the hypA coding region. Here it is shown that the oxygen-responsive transcription factor FNR regulates PhypA under anaerobic conditions only. PhypA does not possess a canonical FNR recognition sequence, but two FNR half-sites are present. Studies using PhypA::lacZ fusions carrying lesions in one or both FNR half-sites indicated that although some residual anaerobic activity was retained by the promoter containing only the downstream FNR half-site, both half-sites are required for maximal PhypA activity in vivo. In vitro gel retardation analysis suggested that the primary interaction occurs at the downstream FNR half-site. Possible explanations for these observations and the implications for other FNR-regulated promoters are discussed.
Escherichia coli possesses three distinct NiFe hydrogenases (designated hydrogenase 1, 2 and 3, respectively)  and it has been suggested that a fourth hydrogenase (Hyf) may be present [2,3]. Hydrogenase 2 has a respiratory role by conserving the energy released from hydrogen oxidation . Hydrogenase 3 is part of the formate hydrogenlyase complex and functions as a gas-evolving hydrogenase in formate oxidation coupled to H+ reduction during fermentative growth . The role of hydrogenase 1 is less well established, although it clearly has a role in hydrogen oxidation. It has been suggested that hydrogenase 1, and perhaps hydrogenase 2, might be involved in recycling the hydrogen generated by the formate hydrogenlyase complex .
The five-gene hyp operon (hypA-E), which runs divergently from the formate hydrogenlyase complex genes (hyc operon), is involved in the maturation of all three NiFe hydrogenases [6,7]. Aspects of the transcriptional regulation of the hypA-E operon have been studied previously and two promoters have been identified . The first promoter is σ54-dependent, and is located upstream of the first gene (hypA) of the operon (Fig. 1). The second promoter is located within the hypA coding region (Fig. 1). Consequently a functional HypA protein is not formed when transcription initiates at the second promoter, but all the genes required for hydrogenase maturation (hypB-E) are transcribed (Fig. 1). Based on primer extension analysis both promoters had similar activities under anaerobic conditions, but under aerobic conditions the second internal hypA promoter was most active . It has been suggested that the internal promoter requires the anaerobic transcription factor FNR for activity under both aerobic and anaerobic conditions . The aerobic requirement for FNR is surprising because the iron–sulfur-containing holo-FNR protein activates the expression of anaerobic genes in response to oxygen-starvation, but does not bind at these target promoters when oxygen is available (reviewed in [8–11]). Nevertheless, it was suggested that the role of the FNR-dependent hyp promoter is to control the expression of the hydrogenase maturation genes (hypB-E) under respiratory conditions when the σ54-dependent promoter is silent .
The aim of the work described here was to further characterise the role of FNR in regulating hyp expression under aerobic and anaerobic conditions. It is shown that FNR acts an anaerobic activator of hyp expression. This FNR-mediated activation of hyp expression is unusual because there is no good match to a canonical FNR binding site in the hyp promoter region. However, it is shown that two FNR half-site motifs located upstream of the σ70-dependent promoter are necessary for FNR-mediated hypB-E expression in vivo.
2Materials and methods
2.1Bacterial strains, plasmids and oligonucleotides
The bacterial strains, plasmids and oligonucleotides used in this study are listed in Table 1.
Table 1. Bacterial strains, plasmids and oligonucleotide primers
hypA upstream site mutagenesis forward AACGGGCACTGGAATTTATCGAACAGCAGG
hypA upstream site mutagenesis reverse CCTGCTGTTCGATAAATTCCAGTGCCCGTT
hypA downstream site mutagenesis forward TGCATTTTGTTTTTATCTGGTTTGCCGCGG
hypA downstream site mutagenesis reverse CCGCGGCAAACCAGATAAAAACAAAAGGCA
2.2In vivo transcription assays
Cultures for in vivo transcription assays were grown in L-broth supplemented with 0.4% (w/v) glucose and appropriate antibiotics: ampicillin (200 μg ml−1); tetracycline (35 μg ml−1); chloramphenicol (25 μg ml−1) . Aerobic cultures (5 ml) were inoculated from an L-agar plate with appropriate antibiotics and shaken (250 rpm) in 250-ml conical flasks for 16 h at 37°C. Anaerobic cultures (10 ml) were inoculated from an L-agar plate with appropriate antibiotics and incubated in sealed 10-ml bottles for 16 h at 37°C, without shaking. In vivo transcription from the various hypA::lacZ fusions (Table 1) was estimated by measuring β-galactosidase specific activities according to Miller .
Plasmids were constructed and isolated using standard methodologies  and their authenticity was confirmed by DNA sequencing. Plasmid copy number was estimated by the method of Taylor and Brose . Amplification of the hyp operon promoter and site-directed mutagenesis were achieved by standard polymerase chain reaction (PCR) techniques using Expand™ high-fidelity PCR system (Roche). For gel retardation assays, DNA fragments (∼0.4 μg) obtained by restriction digestions of plasmid DNA were radiolabelled by end-filling using the Klenow fragment of E. coli DNA polymerase I (5 U) at 30°C for 30 min. Unincorporated nucleotides were removed using a QIAquick PCR purification kit and the pure radiolabelled DNA was stored at −20°C.
2.4Gel retardation assays
The FNR* protein was isolated as previously described . A typical gel shift reaction contained radiolabelled DNA (30 ng), FNR (0–3 μM), Tris–HCl, pH 8.0 (20 mM), glycerol (2.5% v/v), KCl (25 mM), EDTA (1 mM) and H2O to a final volume of 10 μl. Reactions were pre-incubated at 20°C for 10 min before the addition of loading solution (glycerol, 10% v/v; bromophenol blue, 0.03% w/v). Samples were applied to Tris–glycine buffered 5% (w/v) polyacrylamide gels. Following electrophoresis the gel was dried and exposed to Fugi RX X-ray film and for 16 h at −70°C. For competition experiments a 15-fold molar excess of the indicated cold competitor DNA compared to the labelled target DNA was added to the incubations.
The starting point for this work was the report that the σ70-dependent hyp operon promoter (PhypA) requires the anaerobic transcription factor FNR for expression under aerobic and anaerobic conditions . To investigate the action of FNR at PhypA a DNA fragment that included the hypA gene and extended 100 bp upstream of the start codon was amplified by PCR and cloned into pRW50 (Table 1) to create a low copy number plasmid (pGS1483) carrying an hypA::lacZ fusion for in vivo transcription studies (Fig. 1). The E. coli strain JRG1728 (Δlac, Δfnr) containing pGS1483 was then transformed with either pGS196, an fnr expression plasmid; pGS198, an equivalent plasmid encoding an FNR protein that is unable to sense and respond to oxygen-starvation, because it lacks three of the four cysteine residues that ligate the [4Fe-4S] cluster; or the vector pBR322t (Table 1). Initial experiments with JRG1728 (pGS1483, pGS196) showed that hypA expression was similar in exponential and stationary phase cultures, and that the growth conditions used did not affect the copy numbers of the plasmids used (not shown). Thus, in subsequent experiments cultures were grown for 16 h at 37°C. β-Galactosidase produced from the hypA::lacZ fusion indicated both FNR and Δ29FNR had no effect on expression under aerobic conditions (Fig. 2A). Under anaerobic conditions FNR activated hyp transcription by approximately seven-fold, and Δ29FNR had no effect (Fig. 2A). Thus, it appears that FNR acts only as an anaerobic activator of hyp expression. These results contrast with previous reports  that suggested that the activity of the internal FNR-dependent promoter was increased during aerobic growth.
Gel retardation assays were used to investigate whether FNR acts directly at the hyp promoter region. The same DNA fragment used to construct the hypA::lacZ fusion was end-labelled for use in gel retardation assays. Advantage was taken of the FNR* protein (FNR-D154A) which retains some activity in the presence of oxygen and has been widely used to simplify in vitro analyses of FNR:DNA interactions . The autoradiographs obtained indicated that FNR* binds at the hyp promoter region (Fig. 2B). Thus, the effects of FNR on hyp transcription are likely to be direct and not mediated through another FNR-regulated transcription factor.
To test the possibility that the previously reported aerobic effects of FNR were being masked by transcription from the σ54-dependent promoter, the hyp fragment was dissected into two subfragments. One region contained the upstream σ54−12 and −24 promoter elements and the other contained the internal FNR-dependent promoter (Fig. 1). These fragments were cloned into pRW50 and used in in vivo transcriptional assays, as before. Transcription from the σ54-dependent promoter was unaffected by the presence or absence of FNR (Fig. 3B). Even though the promoter activities under aerobic conditions were lower compared to the longer promoter used initially, once again, there was no indication of aerobic activation in the presence of FNR. However, surprisingly the FNR-dependent anaerobic activation (seven-fold) observed with the complete hyp promoter region was much reduced (1.8-fold) for the hypA::lacZ fusion containing only the σ70 promoter (Fig. 3A).
Inspection of the hyp promoter region failed to reveal any sequences that closely matched an FNR binding site (TTGATNNNNATCAA). However, two half-sites were identified (TTGAT) centred at −154 and −48 relative to the start site of the σ70-dependent promoter (Fig. 1). It was noted that in dissecting the hypA promoter these FNR half-sites had been separated and the pattern of hypA::lacZ expression described above suggested that the anaerobic activation of the internal hypA promoter might involve interaction of FNR with both half-site motifs. Therefore, each half-site within the full-length PhypA was inactivated by site-directed mutagenesis (Fig. 1). Measurement of β-galactosidase activities of cultures grown under aerobic and anaerobic conditions indicated that mutagenesis of the downstream (−48) half-site (PhypA+−), or both (−48 and −154) FNR half-sites (PhypA−?) abolished FNR-dependent anaerobic activation (Fig. 4). The lesion in the upstream (−154) half-site (PhypA−+) reduced FNR-dependent anaerobic induction from seven-fold, when both FNR half-sites were present, to two-fold (Figs. 2 and 4). Thus, the site-directed mutagenesis experiments supported the conclusions drawn from the promoter deletion experiments described above. Moreover, they suggest that FNR-dependent activation of hyp expression is completely dependent on the downstream half-site, and that both FNR half-sites are required for maximum anaerobic induction.
The mutagenesis data described above suggested that the two FNR half-sites acted syngergistically to promote hyp expression. Therefore, the altered promoter fragments used to construct the lacZ fusions (above) were used in gel retardation assays to determine whether binding to either FNR half-site was dependent on the presence of the other. Once again to simplify these experiments advantage was taken of the FNR* protein (FNR-D154A). As expected FNR* bound at the unaltered hypA promoter (PhypA) fragment (Fig. 5). Retarded complexes were also observed with the promoter fragments carrying only one intact FNR half-site (PhypA−+ and PhypA+−). However, when the promoter fragment that had both FNR half-sites mutated (PhypA−?) was used the assays contained free DNA and a smeared low-mobility complex, suggesting that, as expected, FNR* binding is impaired (Fig. 5, compare lanes 2 and 5). This was confirmed by testing the abilities of cold unaltered and mutant hypA promoter fragments to compete against labelled unaltered PhypA for FNR*. In these experiments it was found that addition of excess cold DNA carrying a lesion in the upstream FNR half-site (PhypA−+) competed as well as the unaltered promoter (PhypA) for FNR* (Fig. 5, lanes 6 and 7). However, DNA fragments carrying lesions in the downstream FNR half-site (PhypA+−) or in both half-sites (PhypA−?) were much less effective competitors (Fig. 5, lanes 8 and 9). Thus, it would appear that FNR binds preferentially to the downstream half-site. These observations suggest that the TTGAT sequences are the sites of FNR action and that FNR binding at the downstream half-site (Fig. 1) is the primary interaction.
In contrast to a previous report, it is shown here that in vivo transcription from the internal hypA promoter is FNR-dependent under anaerobic, and not aerobic, conditions. The reasons for this discrepancy are not obvious but may reflect the different techniques (primer extension compared to the promoter fusions used here) that were used to study hyp expression. It is possible that the aerobic cultures used for primer extension became sufficiently anaerobic during processing to allow some transcription from the internal hypA promoter. Alternatively, the native hypB-E transcript might be more stable under aerobic compared to anaerobic conditions. Further investigation will be needed to determine if this is the case.
In E. coli, FNR-regulated promoters contain DNA sequences related to the consensus FNR box, which consists of an imperfect palindrome (TTGATNNNNATCAA) . The internal hypA promoter (PhypA) is unusual in that it does not contain such an inverted repeat but rather has two FNR half-sites (TTGAT) separated by 102 bp (Fig. 1). It is shown that both half-sites are necessary for maximal FNR-dependent transcription from PhypA.
The simplest explanation for these observations, based on our knowledge of FNR:DNA interactions, is that maximal transcription from PhypA requires the binding of two FNR dimers, with one dimer bound at each FNR half-site. Such an arrangement could allow both dimers to make simultaneous contact with RNA polymerase to synergistically activate transcription as has been reported for an ansB promoter variant engineered to have tandem FNR sites . However, at other promoters, tandem FNR dimers have been shown to repress gene expression [18,21–23]. Moreover, the separation between the two FNR half-sites is greater than that at which the related transcription activator CRP has been shown to act synergistically when bound in tandem [24,25]. Alternatively, occupation of the upstream half-site by FNR may configure the promoter such that the downstream FNR dimer can interact productively with RNA polymerase. This would be consistent with the preferred binding of FNR* to the downstream half-site. Thus, the hyp promoter region might resemble the nirB promoter where binding of NarL/NarP reconfigures an inhibitory upstream complex to allow a conventionally located FNR dimer to activate transcription [26,27]. Further detailed in vitro analysis of hyp transcription will be needed to determine which, if any, of these possibilities is correct.
The internal hypA promoter is unusual in being dependent upon FNR recognising FNR half-sites, rather than a full FNR box, for anaerobic activity. Binding at an FNR half-site is not without precedent and has been shown to occur at two FNR-repressed E. coli promoters; hemA and at the upstream FNR site of the ndh promoter . However, the hypA promoter is the first example of an FNR-activated promoter that requires tandem FNR half-sites. Moreover, it is probably not the only member of the FNR-regulon to adopt this strategy. The E. coli caiF gene encodes a transcription activator of carnitine metabolism that has a similar promoter organisation to the hyp operon. It has an upstream σ54-dependent promoter and a downstream σ70-dependent promoter. Moreover, transcription from the σ70-dependent promoter is activated by FNR under anaerobic conditions . Inspection of the caiF promoter region reveals that, like PhypA, there are no good matches to the FNR consensus but there are two TTGAT half-sites centred at −61 and −143, between the σ54- and σ70-dependent transcript starts. Thus, the caiF promoter is architecturally similar to hyp and it is likely that the mechanism of anaerobic regulation based on recognition of tandem FNR half-sites has been adopted by the caiF promoter. It is suggested that tandem FNR half-sites may be a feature of other genes that are active under anaerobic conditions and use both σ54- and σ70-dependent promoters for expression.
We thank Dr Colin Scott for useful discussions and the Biotechnology and Biological Sciences Research Council (UK) for financial support.