RNA was isolated from cultures of Escherichia coli strain MG1655 and derivatives defective in fnr, narXL, or narXL with narP, during aerobic growth, or anaerobic growth in the presence or absence of nitrate or nitrite, in non-repressing media in which both strain MG1655 and an fnr deletion mutant grew at similar rates. Glycerol was used as the non-repressing carbon source and both trimethylamine-N-oxide and fumarate were added as terminal electron acceptors. Microarray data supplemented with bioinformatic data revealed that the FNR (fumarate and nitrate reductase regulator) regulon includes at least 104, and possibly as many as 115, operons, 68 of which are activated and 36 are repressed during anaerobic growth. A total of 51 operons were directly or indirectly activated by NarL in response to nitrate; a further 41 operons were repressed. Four subgroups of genes implicated in management of reactive nitrogen compounds, NO and products of NO metabolism, were identified; they included proteins of previously unknown function. Global repression by the nitrate- and nitrite-responsive two-component system, NarQ-NarP, was shown for the first time. In contrast with the frdABCD, aspA and ansB operons that are repressed only by NarL, the dcuB-fumB operon was among 37 operons that are repressed by NarP.
- fumarate and nitrate reductase regulon (FNR regulon)
- NarXL regulon
- regulation of anaerobic metabolism
The availability of whole genome sequences and the development of microarrays that include coding sequences for every gene in an organism have revolutionized our approach to studies of bacterial gene regulation. Studies of the transcriptional control of a single promoter have largely been replaced by attempts to answer global questions such as the number of genes regulated by a specific transcription factor, or the changes in gene expression that result from a particular stress or alteration in substrate availability. Early microarray studies were compromised by technical problems that have largely been overcome, but the end-products of many of the most recent studies are still simply lists of genes that are up- or down-regulated in response to a mutation, environmental stress, or change in nutrient availability. Why has so much effort resulted in so few seminal hypotheses or insights into microbial physiology?
In enteric bacteria, nitrate reduction to ammonia is essentially an anaerobic process. Adaptation to anaerobic growth is regulated by two global transcription factors, FNR (fumarate and nitrate reductase regulator) and ArcA (anaerobic respiratory control). FNR activates transcription of many genes in response to anaerobic conditions. However, FNR can also function as a repressor. In contrast, as oxygen becomes depleted and the ubiquinone pool becomes more reduced, the kinase activity of ArcB is activated, ArcA is phosphorylated, and many genes required for aerobic metabolism are repressed. These are not two independent networks, however, because FNR activates transcription of ArcA, albeit only weakly [1,2].
In two recent studies, microarray data have been analysed to reveal the extent of changes in the transcriptome as Escherichia coli K-12 adapts from aerobic to anaerobic growth. Salmon et al.  compared RNA isolated from cultures of strain MC4100 that had been grown aerobically or anaerobically, and from an fnr mutant that had been grown anaerobically. Kang et al.  grew both strain MG1655, for which the complete genome sequence is available, and an isogenic fnr mutant aerobically and anaerobically in a minimal salts medium and compared their transcriptome data with those from the previous study. In both groups of experiments, glucose was used as the carbon source for growth, and cluster analysis was used to analyse the data. In the former study, expression levels of 1445 genes changed in response to the availability of oxygen with 95% confidence of avoiding false positive results. The corresponding figure in the study of Kang et al.  was 962, but only 334 genes were common to both datasets, and of those, 123 appeared to be regulated in opposite directions. Thus 211 genes showed similar responses, only 10% of the 2073 genes for which changes were observed. Changes in expression levels of operons known from a range of genetic, biochemical and in vitro data to be dependent on FNR activation were not detected in these previous studies, and the nirB promoter in which there is a consensus binding site for FNR activation even appeared to be repressed by FNR .
Re-assessment of the FNR regulon of E. coli
Transcription of many operons regulated by FNR are co-dependent on other transcription factors [5,6] that respond to changes in carbon source, growth rate or nutrient availability. Furthermore, an fnr mutant is unable to grow anaerobically in simple media containing nitrate and only a non-fermentable carbon source such as glycerol, yet some FNR-dependent promoters are repressed in the presence of glucose. Two aims of the present study were to determine the number of operons significantly regulated by FNR or the nitrate-responsive two-component regulatory system, NarX-NarL. To minimize artifacts due to differences in growth rate between the fnr mutant and the parental strain resulting from inability of the mutant to utilize some terminal electron acceptors, bacteria were grown anaerobically in a minimal salts medium supplemented with glycerol, fumarate and TMAO (trimethylamine N-oxide). Where indicated, 20 mM potassium nitrate or 2.5 mM sodium nitrite was also added. Oligonucleotide microarrays were designed to avoid cross-hybridization between closely related genes; RNA was extracted from bacteria grown under a wide range of conditions and harvested at a defined stage of exponential growth. A common pool of reference RNA from multiple cultures of the parental strain, MG1655, that had been grown anaerobically in the absence of nitrate or nitrite was used on every microarray so that data could be compared between experiments. Each of the above procedures was validated in control experiments. As FNR-repressed genes will be expressed at low levels during anaerobic growth, some of the experiments were repeated with a pool of RNA from aerobically grown bacteria as the reference sample. This provided a check for false negative data resulting from rejection of microarray signals too weak to be considered reliable. In the first set of experiments, RNA was isolated from strain MG1655 and an fnr mutant that had been grown aerobically, or anaerobically in the presence or absence of nitrate or nitrite. Fumarate and TMAO were present in all cultures. The results from three biologically independent experiments for each of the seven types of culture were compared with the reference pool of RNA from the parental strain that had been grown in the absence of nitrate or nitrite. Only greater than 2-fold differences in microarray signals were considered to be biologically significant and results were adjusted to provide for a probability of a false positive identification of less than 5%.
A comprehensive literature search revealed that there is previous independent biochemical or genetic evidence that 29 operons are activated and 14 operons are repressed by FNR. In some of these cases, however, apparent regulation was based on promoter fusion data and might therefore be indirect, for example, due to an effect of FNR on arcA transcription. The current microarray data confirmed 32 of these 43 assignments, but alone did not confirm FNR activation of five operons (adhE, glpTQ, cydDC, hlyE and arcA) or FNR repression of six operons (hemA, narXL, tpx, yeiL, norVW and ubiCA). A total of 44 operons not previously known to be included in the FNR regulon were activated by FNR and a further 28 operons appeared to be repressed. For each of these operons, a potential FNR-binding site with at least an 8 out of 10 bp match to the consensus sequence could be identified. The first conclusion from this study is therefore that the FNR regulon includes at least 104, and possibly as many as 115, operons.
Only five of the previously identified FNR-activated operons, hcp-hcr, narK, narGHJI, fdnGHJI and nirBDC, were strongly induced by nitrate: this induction was FNR-dependent. Although, as expected, the fdnG and narG promoters were more strongly activated by nitrate than by nitrite , the hcp-hcr, narK and nirBDC operons were also strongly induced in the presence of nitrite. Most surprising was the fact that only one newly identified FNR-activated operon was significantly induced (and that only 2.8-fold) by nitrate. Nine of the newly identified FNR-dependent operons were repressed by nitrate; they included ynfEFGHI and ydhYVWXU encoding putative oxidoreductases (the former is an S-oxide reductase), as well as the fes and cadC promoters.
Effect of narXL and narXLP deletions on the transcriptome during anaerobic growth in the presence of nitrate
The NarX-NarL two-component regulatory system detects and regulates gene expression in response to the availability of high concentrations of nitrate [8,9]. The narX and narL genes were deleted from strain MG1655, and RNA was isolated from the mutant during anaerobic growth in the presence of nitrate, conditions that are optimal for regulation by phosphorylated NarL. A total of 51 operons were activated by NarL in response to nitrate; a further 41 operons were repressed. In E. coli, an alternative two-component regulatory system, NarQ-NarP, activates gene expression in response to nitrate and is especially sensitive to low concentrations of nitrate . The NarP regulon overlaps the NarL regulon because both NarL and NarP can bind to a NarP site, which is a 7-2-7 inverted repeat sequence where the seven conserved bases form the NarL heptamer . However, NarL can also bind to other sites not organized as a 7-2-7 inverted repeat, and multiple NarL heptamers occur in the regulatory regions of many nitrate-responsive genes. Consequently, some promoters, for example one of the two napF promoters regulating expression of the periplasmic nitrate reductase operon , and the nrfA promoter, are activated by NarP in response to nitrite or low concentrations of nitrate, but repressed by NarL when nitrate is abundant.
The narP gene was deleted from the narXL deletion strain, and RNA was isolated after anaerobic growth in the presence of nitrate. The microarray data revealed 14 operons that are activated by NarP in response to the availability of nitrate. As there was no previous convincing evidence that NarP can repress transcription of nitrate-responsive operons, it was both interesting and surprising that 37 operons were repressed by nitrate even in a narXL deletion strain relative to the ΔnarXL ΔnarP mutant, showing that these operons are repressed by NarP. Another interesting aspect of these results was that genes required for anaerobic dicarboxylate metabolism fell into two subgroups, both of which are repressed during growth in the presence of excess nitrate. Although the frdABCD, aspA and ansB operons were repressed only by NarL, the dcuB-fumB operon was strongly repressed by both NarL and NarP (Figures 1A and 1B). Similar differences were observed in the regulation by nitrate of genes for hydrogenase synthesis: the hyaABCDEF and hyc-ABCDEFGH operons were repressed by both NarL and NarP; the hyb operon was repressed only by NarL (Figures 1C and 1D).
Possible roles of genes of unknown function in detoxification of reactive nitrogen species
Multiple defence mechanisms protect bacteria against reactive nitrogen compounds such as nitric oxide (NO) and peroxynitrite, ensuring that the concentration of NO remains very low. NO activates NorR, the transcription factor that regulates the NO reductase genes, norVW [12⇓⇓–15]. The periplasmic nitrite reductase, NrfAB, is a highly effective NO reductase that reduces NO to ammonia . In aerated cultures of E. coli, HMP (flavohaemoglobin) oxidizes NO to nitrate . During anaerobic growth, transcription of the hmp gene is repressed by FNR, but severe exposure to NO stress results in inactivation of FNR and hence de-repression of hmp expression . HMP can then reductively protect E. coli against NO [14,18]. This multiplicity of defence mechanisms, coupled with the chemical instability of NO and its derivatives, makes it difficult to assign on the basis of mutant phenotypes protective functions to other proteins that might protect E. coli against reactive nitrogen compounds. One of the most fascinating aspects of the microarray data is that some proteins of unknown function were regulated co-ordinately with those already known to provide protection against reactive nitrogen species (Table 1). The YtfE protein was recently implicated in the NO stress response : in parallel with hmpA, ytfE transcription is strongly induced by nitrate, repressed by FNR and optimally expressed in an fnr mutant during growth in the presence of nitrite (Table 1). The similar regulation of the ygbA promoter suggests that this gene of unknown function might also encode a defence mechanism against reactive nitrogen compounds (results not shown). Expression of two other operons encoding proteins of unknown function, YeaR-YoaG and YibIH, is induced strongly by nitrite only in the absence of FNR (Table 1). Are these proteins also part of a defence mechanism that, like HMP, would be most important physiologically under extreme reactive nitrogen stress, when NO is sufficiently abundant to inactivate FNR?
Conversely, transcription of both the nap and nrf operons is totally dependent on FNR, repressed by nitrate but induced by nitrite, effectively providing protection against NO generated in the anaerobic gastrointestinal tract where nitrate is a scarce but valuable electron acceptor. On the assumption that similar correlations indicate linked functions, expression of hcp is co-ordinately regulated with nirB, suggesting that the enigmatic prismane (or hybrid cluster) protein encoded by hcp and its putative reductase encoded by hcr might also be involved in reactive nitrogen management (Table 1). Could it be another NO or even a peroxynitrite reductase? The microarray data lead us to propose that four subgroups of genes enable E. coli to survive a wide variety of conditions under which reactive nitrogen stress would otherwise be detrimental to growth, or even lethal.
A further striking result is that the regulation of six of the operons implicated in the management of reactive nitrogen species, hcp-hcr, ygbA, ygiS, yeaR-yoaG, yibHI and ytfE, is only partially dependent on NarL or NarP: nitrate induction from these promoters was detected even in the narXL narP deletion strain.
In summary, we have exploited at least five critical factors that limited previous attempts to define the FNR regulon of E. coli. They include (i) use of a highly specific oligonucleotide-based array that discriminates between homologous pairs of genes (typical examples are narG-narZ, narK-narU and frdA-sucA); (ii) avoidance of growing bacteria under conditions in which the target promoters are repressed; (iii) the use of a much wider range of growth conditions to increase the chance that regulatory mechanisms will be revealed, especially at promoters that are co-dependent on multiple transcription factors; (iv) ensuring that the growth medium minimizes artifacts due to differences in growth rates of cultures to be compared; (v) augmenting microarray data with bioinformatic analysis to assign genes to operons and operons into regulons. Even with these precautions, it is clear that microarray data alone are unable to define complete regulons, not only because it is unlikely that all relevant growth conditions will have been tested, but also because many changes in transcript levels will be due to secondary effects, as seen in the current experiments with changes due to ArcA repression being relaxed by an fnr mutation. There is little merit in attempting to use promoter fusions to confirm microarray data, because such studies provide an even less direct assessment of transcript levels than the microarrays they are designed to challenge. Similarly, quantitative real-time PCR does not provide independent complementary evidence for microarray data, but simply confirms that RNA has been isolated competently for the microarray studies and can be assayed by an independent method to give congruent data. The technical performance of microarray analysis is now rarely a significant problem. The ultimate test of whether a transcription factor regulates specific promoters must be to provide physical confirmation that the protein binds to the promoter of interest. To this end, we are currently testing selected predictions from the present study by chromatin immunoprecipitation. Preliminary results have confirmed the assignments so far tested.
This research was supported by funding from the Biotechnology and Biological Sciences Research Council grants 6/EGA16107 and P20180.
The 11th Nitrogen Cycle Meeting 2005: Independent Meeting held at Estación Experimental del Zaidín, Granada, Spain, 15–17 September 2005. Organized and Edited by E.J. Bedmar (Granada, Spain), M.J. Delgado (Granada, Spain) and C. Moreno-Vivián (Córdoba, Spain).
Abbreviations: ArcA, anaerobic respiratory control; FNR, fumarate and nitrate reductase regulator; HMP, flavohaemoglobin; TMAO, trimethylamine N-oxide
- © 2006 The Biochemical Society