Logo of narLink to Publisher's site
Nucleic Acids Res. Jan 2007; 35(1): 269–278.
Published online Dec 12, 2006. doi:  10.1093/nar/gkl1023
PMCID: PMC1802558

Transcription factor distribution in Escherichia coli: studies with FNR protein

Abstract

Using chromatin immunoprecipitation (ChIP) and high-density microarrays, we have measured the distribution of the global transcription regulator protein, FNR, across the entire Escherichia coli chromosome in exponentially growing cells. Sixty-three binding targets, each located at the 5′ end of a gene, were identified. Some targets are adjacent to poorly transcribed genes where FNR has little impact on transcription. In stationary phase, the distribution of FNR was largely unchanged. Control experiments showed that, like FNR, the distribution of the nucleoid-associated protein, IHF, is little altered when cells enter stationary phase, whilst RNA polymerase undergoes a complete redistribution.

INTRODUCTION

An army of more than 250 transcription factors controls gene expression in Escherichia coli. Some of these factors are operon-specific while others, known as global regulators, coordinate the expression of scores of promoters in response to specific environmental cues [reviewed in (13)]. The advent of whole-genome DNA sequencing, and associated advances in DNA microarray technology, has enabled investigation of the battery of genes regulated by each of these global factors. The E.coli FNR protein (regulator of fumarate and nitrate reduction) is the global transcription factor that manages the distribution of RNA polymerase in response to oxygen starvation. FNR senses oxygen via an N-terminal iron–sulfur cluster. Hence, in anaerobic conditions, FNR is able to bind to specific DNA targets at promoters and modulate transcription. In aerobic conditions, FNR is converted to a form, unable to bind these targets [reviewed in (4,5)]. Bioinformatic analysis has been used to search the E.coli genome for DNA sequences that resemble known FNR binding sites (6,7) and DNA microarrays have been used to study differences in the transcriptome that arise when the fnr gene is deleted from the genome (810). These studies illustrate the complexity of the FNR regulon and predict that, while FNR directly regulates ~100 transcription units, it indirectly affects up to 1000 genes. In this study, we used chromatin immunoprecipitation (ChIP), in conjunction with high-density microarrays (ChIP-chip), to measure the binding of FNR across the E.coli chromosome directly, and in vivo, for the first time. This allowed us to identify 63 DNA targets for FNR, some of which are adjacent to poorly expressed genes where FNR has minor regulatory effects.

In the second part of the study, we studied the distribution of FNR as growing E.coli cells enter stationary phase, and found that it is largely unchanged. Recall that transcription patterns change dramatically when cells cease to grow (11,12) but little is known about the distribution and binding of transcription factors in stationary phase cells. In control experiments, we showed that the binding pattern of IHF, a nucleoid-associated protein, is also unchanged, whereas the distribution of RNA polymerase is radically altered.

MATERIALS AND METHODS

Bacterial strains, plasmids and growth conditions

Bacterial strains used in this work are described in Table 1, together with the oligos used to generate different promoter fragments. For ChIP-chip experiments with FNR, MG1655 and JCB1011 cells were grown anaerobically in Luria–Bertani (LB) medium supplemented with 0.4% glucose. Supplementary Figure 1A shows growth of MG1655 and JCB1011 under these conditions, and the time points at which cells were harvested for ChIP-chip experiments. For ChIP-chip experiments with IHF and RNA polymerase, MG1655 cells were grown aerobically in M9 minimal medium to stationary phase (Supplementary Figure 1B). To compare the activity of different promoters:lacZ fusions in the presence and absence of FNR, we used E.coli JCB387 and the fnr derivative JRG1728.

Table 1
Bacterial strains, plasmids and oligonucleotides

ChIP

Bacterial cells were treated with formaldehyde, harvested, lysed and their nucleoprotein was extracted as described by Grainger et al. (13). Immunoprecipitation was then performed using mouse monoclonal antibodies against the 3× FLAG epitope (Sigma) or the β-subunit of RNA polymerase (Neoclone, Madison, USA) or rabbit polyclonal antibodies against IHF (donated by Steve Goodman). Note that the antibody against IHF is polyclonal and recognizes different IHF heterodimers. Immunoprecipitated DNA samples and total cell nucleoprotein samples were purified and labelled with Cy5 or Cy3 respectively, without amplification, as described by Grainger et al. (13).

Microarray analysis of immunoprecipitated DNA

Microarrays (Oxford Gene Technology) were designed and produced specifically to analyse DNA obtained from ChIP experiments with E.coli MG1655 and its derivatives (13). Labelled DNA obtained from immunoprecipitations was hybridized to the microarray as described previously (13). Arrays were then scanned and probes with low Cy5 and Cy3 values and isolated probes with a high Cy5/Cy3 intensity ratio were removed. Data shown are the average of two independent experiments and are presented as Supplementary Tables 1–3. Replicate datasets had a correlation co-efficient between 0.6 and 0.8.

ChIP-chip data analysis

The average Cy5/Cy3 intensity ratio calculated for each microarray spot was plotted against the corresponding position on the E.coli MG1655 chromosome, creating a profile of FNR binding (see Supplementary Table 1). We then searched the profile for ‘peaks’, formed by two or more consecutive probes, with a Cy5/Cy3 ratio clearly distinguishable from the background signal. A cut-off, corresponding to the lowest Cy5/Cy3 ratio observed for a peak at a known FNR target was set, and all probes that had an intensity ratio greater than this value were selected as FNR targets. When several adjacent probes (i.e. probes forming one peak) passed the cut-off, the target position was defined as the centre of the probe with the highest Cy5/Cy3 ratio.

To identify overlapping peaks for ChIP-chip datasets obtained using nucleoprotein from cells growing in different conditions, we aligned the averaged Cy5/Cy3 signals obtained for each condition and applied an equivalent cut-off to both datasets. We then counted the number of probes that passed the cut-off for both datasets (a leeway of one probe in either direction was allowed). To generate the data in Figure 5C, we selected the top 50, 100, 200, 400, 800, 1600, 3200 and 6400 probes from each dataset and determined the distribution of these probes between coding and non-coding DNA.

Construction and assay of promoter::lacZ fusions

Intergenic regions containing putative FNR targets were amplified by PCR from MG1655 genomic DNA using primers listed in Table 1. Fragments were digested with EcoRI and HindIII and cloned into pRW50, a low-copy-number lac expression vector, to generate promoter::lacZ fusions. β-galactosidase levels in JCB387 and JRG178 cells carrying these recombinants were measured by the Miller method (14). Activities shown are the average of three independent experiments, and error bars show one SD on either side of the mean. Cells were grown anaerobically in LB medium supplemented with 0.4% glucose. Assays were performed in triplicate.

Electrophoretic mobility shift assays (EMSA)

EMSA were carried out as detailed by Browning et al. (15). Purified promoter fragments were end-labelled with [γ-32P]ATP and ~0.5 ng of each fragment was incubated with varying amounts of purified FNR D154A, which allows binding in the presence of oxygen. The reaction buffer contained l0 mM potassium phosphate (pH 7.5), 100 mM potassium glutamate, 1 mM EDTA, 50 mM DTT, 5% glycerol and 25 mg ml−1 herring sperm DNA. The final reaction volume was 10 μl. After incubation at 37°C for 20 min, samples were run in 0.25× TBE on a 6% polyacrylamide gel (12 V cm−1) containing 2% glycerol and analysed using a Bio-Rad Molecular Imager FX and Quantity One software.

RESULTS

Isolation of DNA fragments associated with FNR in mid-log phase E.coli

Our aim was to use ChIP to measure the distribution of FNR across the chromosome of growing E.coli cells. To do this, we exploited strain JCB1011 whose fnr gene had been previously modified to encode FNR with a C-terminal 3× FLAG tag (10). Supplementary Figure 3A shows a western blot of total protein from strain JCB1011 and its parent, MG1655, probed with anti-FLAG or anti-FNR antibodies. The results show that intracellular levels of wild-type FNR and the FNR-3× FLAG fusion protein are similar and that the anti-FLAG antibody does not cross-react with other proteins. To check that the activity of FNR was unaffected by the 3× FLAG tag, we compared expression of five FNR-dependent promoters in JCB1011 and MG1655 (Supplementary Figure 3B) and anaerobic growth of the two strains (Supplementary Figure 1A). The results of these tests argue that the function of FNR is unaffected by the tag. Thus, JCB1011 and MG1655 cells were grown anaerobically in LB glucose medium to an OD650 of ~0.4, cultures were treated with formaldehyde, and cellular DNA was extracted and sonicated, yielding DNA fragments of ~500-1000 bp. After immunoprecipitation with anti-FLAG antibodies, DNA fragments from JCB1011 or control MG1655 cells were purified, labelled with Cy5 and Cy3, respectively, mixed and hybridized to the microarray. After washing and scanning, the Cy5/Cy3 signal intensity ratio was calculated for each probe. In parallel, the experiment was repeated using aerobically grown cells. Complete datasets are shown in Supplementary Table 1. Figure 1A gives an overview of the profile for FNR binding and some examples are shown in Figure 1B. Peaks for FNR binding are discrete and easily distinguishable from the background signal.

Figure 1
Distribution of FNR binding across the E.coli chromosome. (A) The figure shows an overview of results from ChIP-chip experiments that measure the profile of FNR binding across the E.coli chromosome during exponential growth in anaerobic conditions. Binding ...

Identification and sequence analysis of FNR targets

To determine the location of peaks for FNR binding in an unbiased manner, a Cy5/Cy3 cut-off was applied to the ‘anaerobic’ dataset. A total of 204 probes passed this cut-off, corresponding to 63 separate peak locations, all of which were in non-coding DNA or close to the 5′ end of a gene (Table 2). Of the 63 peaks identified, 20 correspond to locations listed as FNR targets in the current version of the Ecocyc database [www.ecocyc.org, (16)] and 9 of the rem-aining 43 peaks locate to FNR targets predicted by Constantinidou et al. (10). To identify FNR binding sites at the 43 loci not currently listed by Ecocyc, we selected and then combined 500 bp DNA sequences corresponding to the centre of each peak. We then used AlignACE to search for sequence motifs present in these DNA sequences. A sequence logo representing the motif that we identified is shown in Figure 1C. The motif clearly matches the known FNR consensus binding motif of TTGAT(n)4ATCAA. Using this approach, we were able to identify FNR binding sites at 33 of the 43 putative FNR targets identified by our ChIP-chip analysis. The base sequences and locations of the binding sites that we identified are shown in Table 2. Recall that the activity of FNR is triggered by lack of oxygen (4,5). The control dataset from cells grown aerobically showed that binding at the different targets was suppressed (e.g. see Figures 1B and and2A2A).

Table 2
FNR targets identified by ChIP-chip analysis
Figure 2
Association of FNR with a novel group of DNA targets. (A) The in vivo DNA binding profile of FNR across intergenic regions upstream of the hsdR and mrr (i), dbpA (ii), yccF and helD (iii) and nohA (iv) genes. Binding is illustrated by Cy5/Cy3 signals ...

Association of FNR with some previously uncharacterized DNA targets

Five of the targets for FNR binding are adjacent to genes encoding proteins involved in the manipulation of RNA or DNA and, in each case, a likely FNR binding site was identified (see Table 2 section E; Figure 2A). DNA fragments covering each of these targets were amplified and end-radiolabelled, or cloned into the lac expression vector pRW50 to create promoter::lacZ fusions. To detect the binding of FNR to the predicted target promoters in vitro, the radiolabelled DNA fragments were used in EMSA assays with purified FNR protein (Figure 2B). In all cases, addition of purified FNR retarded the migration of the purified DNA fragments. To investigate the effect of FNR on transcription from each target promoter, each of the pRW50 encoded promoter::lacZ fusions was transformed into E.coli strain JCB387 or the fnr derivative JRG1728. Expression of lacZ in each strain was measured. Our data (Figure 2C) show that deletion of fnr had marginal affects on transcription and that the activity of each cloned promoter was low in our conditions.

Distribution of FNR in stationary phase

The analysis of FNR binding was repeated using stationary phase cultures of JCB1011 and MG1655. The dataset for the FNR experiment is presented in Supplementary Table 1, alongside the data from growing cells, and an overview is shown in Figure 3A. The results show that the profile of FNR binding in stationary phase is similar to the profile in growing cells (e.g. are shown in Figure 3B).

Figure 3
Comparison of chromosome-wide FNR distribution in growing and stationary phase E.coli. (A) The figure shows an overview of results from ChIP-chip experiments that measure the profile of FNR binding across the E.coli chromosome during rapid growth (green) ...

As controls for this experiment, similar analyses were performed with the nucleoid-associated protein, IHF and with RNA polymerase, which is known to be redistributed in stationary phase E.coli. Note that previously we had used ChIP-chip to study IHF and RNA polymerase in exponentially growing cells (17). The datasets for IHF and RNA polymerase binding are shown in Supplementary Tables 2 and 3, respectively, alongside data generated using growing E.coli cells (17). The results show that the profile of IHF binding in stationary phase is similar to the profile in growing cells while the profile of RNA polymerase is radically altered.

Similarities between the datasets for FNR, IHF and RNA polymerase binding, in growing and stationary phase cells, were quantified by calculating correlation coefficients (Figure 4A) and by comparing the position of probes passing the cut-off for each dataset (Figure 4B). For FNR and IHF, the correlation between ChIP-chip datasets profiling binding in growing and stationary phase cells is high [Figure 4A (i) and (ii)]. Much less correlation is observed when ChIP-chip datasets for RNA polymerase binding in stationary phase and mid-log phase cells are compared [Figure 4A (iii)]. Consistent with this, for FNR and IHF, many of the same probes pass the cut-off for both datasets. In contrast, few probes pass the cut-off for both the mid-log and stationary phase RNA polymerase experiments (Figure 4B).

Figure 4
Comparison of chromosome-wide FNR, IHF and RNA polymerase distribution in growing and stationary phase E.coli. (A) Correlation between ChIP-chip datasets profiling the chromosome-wide distribution of FNR (i) IHF (ii) and RNA polymerase (iii) in growing ...

DISCUSSION

In this work, we have applied ChIP-chip technology to produce the first chromosome-wide direct analysis of DNA binding in vivo by the global E.coli transcription regulator, FNR. The advantage of this approach to studying the FNR regulon is that it avoids complications due to genes that are indirectly controlled by FNR or genes that are regulated by multiple transcription factors. Moreover, FNR binding at sites adjacent to poorly transcribed genes, or genes where FNR has little impact on transcription, can be detected and the effects of environmental conditions can be studied. We identified 63 locations at which FNR binds to the E.coli chromosome, including a group of five targets adjacent to genes encoding proteins that manipulate DNA and RNA. None of these five targets were identified as FNR regulated by previous transcriptome analyses (8,9,10), none are listed as FNR targets by the Ecocyc database (16), and, to our knowledge, FNR-dependent regulation of such proteins has not been documented. At 10 of the 63 targets of FNR binding, we were unable to identify a match to the canonical FNR binding sequence. These may be locations at which FNR binds cooperatively with another factor. We note that the profile of FNR binding presented here consists of discrete peaks (Figure 1A). In contrast, the published binding profile for the related transcription activator, CRP, is far more complex, due to the existence of ~10 000 low affinity binding sites for CRP scattered throughout the genome (6,13). Consistent with this, Robison et al. (6) predicted only ~500 low affinity targets for FNR in the E.coli chromosome. Although our experiments identified 63 discrete targets for FNR binding, these targets include only 20 out of the 65 validated targets listed in the Ecocyc database (16). This highlights an important limitation of the ChIP-chip methodology. Its inability to detect all FNR-DNA interactions is likely to be due to inefficient crosslinking at some locations, or epitope masking (18).

Although major changes in gene expression and nucleoid structure occur when E.coli cells enter stationary phase (11,12,19), it is not known if changes in the distribution of global DNA binding proteins occur. Our observation that the DNA binding profile for FNR is largely unaltered in stationary phase cells shows that the compaction of the stationary phase chromosome does not occlude FNR binding sites. This surprising result prompted us to examine the nucleoid-associated protein IHF and we came to the same conclusion; the distribution of the protein is similar in growing and stationary phase cells. In sharp contrast, as expected, the binding profile of RNA polymerase was completely altered. In growing cells, most RNA polymerase is associated with ~90 transcription units encoding factors required for protein synthesis, motility and ATP production (13). In stationary phase, this RNA polymerase is liberated and is distributed more equitably between the different genes (Figure 5A). Interestingly, for most transcription units, RNA polymerase binding is skewed toward the 5′ end (e.g. are shown in Figure 5B) and thus, in stationary phase, the proportion of RNA polymerase bound to non-coding parts of the genome is increased (Figure 5C). Consistent with this, Lee and Gralla (20) showed that some σS-dependent promoters have the ability to trap RNA polymerase. We note that, in their study of RNA polymerase distribution in stationary phase S.cerevisiae, Radonjic and colleagues identified a similar phenomenon (21). This suggests that trapping of RNA polymerase at promoters may be an evolutionarily conserved mechanism to regulate transcription in response to growth rate.

Figure 5
Increased association of RNA polymerase with non-coding DNA during stationary phase. (A) The figure shows an overview of results from ChIP-chip experiments that measure the profile of RNA polymerase binding across the E.coli chromosome during stationary ...

SUPPLEMENTARY DATA

Supplementary Data are available at NAR online.

[Supplementary Data]

Acknowledgments

The authors thank Steve Goodman and Jeff Green for providing antisera and Jeff Cole, Bob Landick, David Lee, Tim Overton, Bernhard Palsson and Joseph Wade for helpful discussions. This work was supported by a Wellcome Trust programme grant, a BBSRC Japan Partnering Award, a grant-in-aid of scientific research from the Japanese Ministry of Education, Culture, Sports, Science and Technology, and the Daiwa Anglo-Japanese Foundation. Funding to pay the Open Access publication charges for this article was provided by our Wellcome Trust programme grant.

Conflict of interest statement. None declared.

REFERENCES

1. Browning D.F., Busby S.J. The regulation of bacterial transcription initiation. Nature Rev. Microbiol. 2004;2:57–65. [PubMed]
2. Martinez-Antonio A., Collado-Vides J. Identifying global regulators in transcriptional regulatory networks in bacteria. Curr. Opin. Microbiol. 2003;6:482–489. [PubMed]
3. Babu M.M., Teichmann S.A. Evolution of transcription factors and the gene regulatory network in Escherichia coli. Nucleic Acids Res. 2003;31:1234–1244. [PMC free article] [PubMed]
4. Browning D., Lee D., Green J., Busby S. Secrets of bacterial transcription initiation taught by the Escherichia coli FNR protein. In: Hodgson D., Thomas C., editors. Signals, Switches, Regulons and Cascades: Control of Bacterial Gene Expression. Cambridge University Press, Cambridge, UK; 2002. pp. 127–142.
5. Green J., Paget M. Bacterial redox sensors. Nature Rev. Microbiol. 2004;2:954–966. [PubMed]
6. Robison K., McGuire A.M., Church G.M. A comprehensive library of DNA-binding site matrices for 55 proteins applied to the complete Escherichia coli K-12 genome. J. Mol. Biol. 1998;284:241–54. [PubMed]
7. Tan K., Moreno-Hagelsieb G., Collado-Vides J., Stormo G. A comparative genomics approach to prediction of new members of regulons. Genome Res. 2000;11:566–584. [PMC free article] [PubMed]
8. Salmon K., Hung S.P., Mekjian K., Baldi P., Hatfield G.W., Gunsalus R.P. Global gene expression profiling in Escherichia coli K12. The effects of oxygen availability and FNR. J. Biol. Chem. 2003;278:29837–29855. [PubMed]
9. Kang Y., Weber K.D., Qiu Y., Kiley P.J., Blattner F.R. Genome-wide expression analysis indicates that FNR of Escherichia coli K-12 regulates a large number of genes of unknown function. J. Bacteriol. 2005;187:1135–1160. [PMC free article] [PubMed]
10. Constantinidou C., Hobman J.L., Griffiths L., Patel M.D., Penn C.W., Cole J.A., Overton T.W. A reassessment of the FNR regulon and transcriptomic analysis of the effects of nitrate, nitrite, NarXL and NarQP as Escherichia coli K12 adapts from aerobic to anaerobic growth. J. Biol. Chem. 2006;281:4802–4815. [PubMed]
11. Selinger D.W., Cheung K.J., Mei R., Johansson E.M., Richmond C.S., Blattner F.R., Lockhart D.J., Church G.M. RNA expression analysis using a 30 bp resolution Escherichia coli genome array. Nat. Biotechnol. 2000;18:1262–1268. [PubMed]
12. Weber H., Polen T., Heuveling J., Wendisch V.F., Hengge R. Genome-wide analysis of the general stress response network in Escherichia coli: sigmaS-dependent genes, promoters, and sigma factor selectivity. J. Bacteriol. 2005;187:1591–1603. [PMC free article] [PubMed]
13. Grainger D.C., Hurd D., Harrison M., Holdstock J., Busby S.J. Studies of the distribution of Escherichia coli cAMP-receptor protein and RNA polymerase along the E. coli chromosome. Proc. Natl Acad. Sci.USA. 2005;102:17693–17698. [PMC free article] [PubMed]
14. Miller J. Experiments in Molecular Genetics. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1972.
15. Browning D.F., Grainger D.C., Beatty C.M., Wolfe A.J., Cole J.A., Busby S.J. Integration of three signals at the Escherichia coli nrf promoter: a role for Fis protein in catabolite repression. Mol. Microbiol. 2005;57:496–510. [PubMed]
16. Keseler I.M., Collado-Vides J., Gama-Castro S., Ingraham J., Paley S., Paulsen I.T., Peralta-Gil M., Karp P.D. EcoCyc: a comprehensive database resource for Escherichia coli. Nucleic Acids Res. 2005;33:334–337. [PMC free article] [PubMed]
17. Grainger D.C., Hurd D., Goldberg M., Busby S.J. Association of nucleoid proteins with coding and non-coding segments of the Escherichia coli genome. Nucleic Acids Res. 2006;34:4642–4652. [PMC free article] [PubMed]
18. Buck M.J., Lieb J.D. ChIP-chip: considerations for the design, analysis and application of genome-wide chromatin immunoprecipitation experiments. Genomics. 2004;83:349–360. [PubMed]
19. Kim J., Yoshimura S.H., Hizume K., Ohniwa R.L., Ishihama A., Takeyasu K. Fundamental structural units of the Escherichia coli nucleoid revealed by atomic force microscopy. Nucleic Acids Res. 2004;32:1982–1992. [PMC free article] [PubMed]
20. Lee S.J., Gralla J.D. Osmo-regulation of bacterial transcription via poised RNA polymerase. Mol. Cell. 2004;14:153–162. [PubMed]
21. Radonjic M., Andrau J.C., Lijnzaad P., Kemmeren P., Kockelkorn T.T., van Leenen D., van Berkum N.L., Holstege F.C. Genome-wide analyses reveal RNA polymerase II located upstream of genes poised for rapid response upon S. cerevisiae stationary phase exit. Mol. Cell. 2005;18:171–183. [PubMed]
22. Blattner F.R., Plunkett G., III, Bloch C.A., Perna N.T., Burland V., Riley M., Collado-Vides J., Glasner J.D., Rode C.K., Mayhew G.F., et al. The complete genome sequence of Escherichia coli K-12. Science. 1997;277:1453–1474. [PubMed]
23. Page L., Griffiths L., Cole J.A. Different physiological roles for two independent pathways for nitrite reduction to ammonia by enteric bacteria. Arch. Microbiol. 1990;154:349–354. [PubMed]
24. Spiro S., Guest J.R. Inactivation of the FNR protein of Escherichia coli by targeted mutagenesis in the N-terminal region. Mol. Microbiol. 1988;2:701–707. [PubMed]
25. Lodge J., Fear J., Busby S., Gunasekaran P., Kamini N.R. Broad host range plasmids carrying the Escherichia coli lactose and galactose operons. FEMS Microbiol. Lett. 1992;74:271–276. [PubMed]

Articles from Nucleic Acids Research are provided here courtesy of Oxford University Press
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...