Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Dps is a nucleoid-associated protein that plays a major role in condensation of the Escherichia coli chromosome in stationary phase. Here we show that two other nucleoid-associated proteins, Fis and H-NS, can bind at the dps gene promoter and downregulate its activity. Both Fis and H-NS selectively repress the dps promoter, preventing transcription initiation by RNA polymerase containing σ70, the housekeeping σ factor, but not by RNA polymerase containing σ38, the stationary-phase σ factor. Fis represses by trapping RNA polymerase containing σ70 at the promoter. In contrast, H-NS functions by displacing RNA polymerase containing σ70, but not RNA polymerase containing σ38. Dps levels are known to be very low in exponentially growing cells and rise sharply as cells enter stationary phase. Conversely, Fis levels are high in growing cells but fall to nearly zero in stationary-phase cells. Our data suggest a simple model to explain how the Dps-dependent super-compaction of the folded chromosome is triggered as cell growth ceases.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In exponentially growing cultures of E. coli, Dps levels are near zero, but they can rise to ∼175 000 molecules per cell in stationary phase (Talukder et al., 1999). Transcription of dps is driven from a single promoter that can be recognized by RNA polymerase containing either σ70, the major housekeeping σ factor (Eσ70), or σ38, the stationary-phase factor (Eσ38). Although OxyR can activate transcription by Eσ70, and IHF can enhance transcription by Eσ38 (Altuvia et al., 1994; Jeong et al., 2006), the molecular basis of how dps expression is regulated by growth phase is not known.
In a recent study, we exploited chromatin immunoprecipitation to monitor the distribution of different nucleoid-associated proteins across the chromosome of exponentially growing E. coli K-12 (Grainger et al., 2006), and we found that both Fis and H-NS bind to targets near the dps promoter. As it is known that Fis and H-NS can act as transcription factors at some promoters (reviewed by McLeod and Johnson, 2001; Dorman and Deighan, 2003), and also that their level varies with growth phase (Talukder et al., 1999), we have made an in-depth study of the action of Fis and H-NS at the dps promoter. We were especially intrigued by the possible role of Fis, which is abundant during early exponential phase (> 20 000 copies per cell) but is undetectable in starved cells (Talukder et al., 1999). Fis binds as a dimer to degenerate 15-base-pair target sites, most of which are located in non-coding promoter DNA (Robison et al., 1998; Grainger et al., 2006). Many promoters regulated by Fis are also regulated by other nucleoid-associated proteins and significant overlap between the groups of genes regulated by Fis and H-NS has been identified (discussed in Dorman, 2007). H-NS is most abundant in rapidly growing cells but is also found at significant levels in stationary phase (Talukder et al., 1999).
In this article, we report that Fis and H-NS bind to adjacent sites within the core dps promoter and selectively shut down transcription by Eσ70 but not by Eσ38. While H-NS represses transcription via a simple blocking mechanism, Fis forms an unproductive closed complex with Eσ70. The Fis-Eσ70–dps promoter complex can block transcription by Eσ38. Thus, Fis and Eσ70 act as co-repressors of dps transcription.
In vivo deletion analysis of the dps promoter
We first sought to identify a minimal dps promoter that was fully active and subject to growth rate regulation. Thus, we cloned DNA fragments of different length, carrying the dps promoter, into the low-copy-number lac expression plasmid, pRW50, resulting in dps promoter::lacZ fusions (Fig. 1A). Recombinant plasmids were transformed into the E. coli host strain JCB387 and expression from the dps promoter constructs was assayed in mid-log phase and overnight cultures by measuring β-galactosidase activities. Figure 1A shows data with plasmids carrying the full-length dps400 fragment and with the shorter dps300, dps200 and dps100 fragments. The truncated dps300 and dps200 promoter fragments were less active than the dps400 promoter but the dps100 promoter fragment, which covers the core promoter elements, was fully active, suggesting that upstream DNA sequences are not required for maximal levels of transcription. In further assays, we compared growth-phase regulation of the full-length (dps400) and core (dps100) promoter. Data in Fig. 1B show that there is little difference in their regulation, suggesting that determinants for growth rate regulation reside in the core dps promoter. Consistent with this, Lomovskaya et al. (1994) have also reported that the core dps promoter is active and responsive to environmental stimuli such as osmotic shock.
Effect of RpoS, Fis and H-NS on in vivo triggering of the dps promoter
As we had previously detected binding of both Fis and H-NS at the dps promoter (Grainger et al., 2006), and because these proteins have been implicated in the growth-phase regulation of transcription at other targets (Typas et al., 2007a) we measured growth phase-dependent triggering of the dps100 promoter in cells lacking Fis, H-NS or, as a control, RpoS (Fig. 2). As expected, in cells lacking RpoS, transcription from the dps100 promoter was reduced and promoter activity rose slowly as cells grew (Fig. 2A). In cells lacking Fis, we observed an increase in levels of β-galactosidase activity and found that triggering the dps100 promoter occurred more rapidly and at a lower OD650 (Fig. 2B). Similarly, when cells were grown at 25°C, deletion of H-NS resulted in an increase in promoter activity and rapid triggering of dps transcription at a lower OD650 (Fig. 2C, i). No H-NS effect was observed when cells were grown at 37°C (Fig. 2C, ii) consistent with previous observations that H-NS activity is regulated by temperature (Ceschini et al., 2000; Ono et al., 2005). Thus, Fis and H-NS appear to contribute as much to growth-phase regulation as the stationary-phase RNA polymerase.
Binding of Fis and RNA polymerase to the dps promoter
The 15-base-pair consensus DNA target for Fis binding is highly degenerate and principally defined by a G at position 1 and a C at position 15 (Hengen et al., 1997), with an A at position 5 and a T at position 11 also contributing to Fis binding (Shao et al., 2008). Scrutiny of the dps100 promoter sequence identified two putative Fis sites, FisI and FisII, located at positions −33 to −19 and −17 to −3 respectively (Fig. 3A). Figure 3B shows a DNase I footprint experiment to investigate the binding of purified Fis protein to dps promoter DNA fragments. Fis binding results in changes in DNA cleavage by DNase I in the predicted sites for Fis. In particular, binding of Fis rendered base pairs at positions −33 and −9 hyper-sensitive to DNase I attack. To confirm Fis binding to the FisI and FisII sites, we conducted electrophoretic mobility shift assays (EMSA). Figure 3C (part i) shows the results of an EMSA experiment where the dps100 fragment was incubated with increasing concentrations of Fis. Two distinct Fis–DNA complexes can be observed, presumably corresponding to Fis binding to one target or to two targets simultaneously. We next examined Fis binding to the dps100–19G or dps100–17A promoter fragments that carry single-base substitutions at position 15 of FisI or position 1 of FisII respectively. EMSA experiments with each fragment show a single Fis–DNA complex (Fig. 3C, ii and iii). Hence the 19G and 17A mutations reduce Fis binding at FisI and FisII respectively. The dps100–19G fragment had a lower affinity for Fis than the dps100–17A fragment showing that FisI is the primary binding site for Fis. Binding of Fis to a dps100 derivative carrying both the 17A and 19G substitutions was almost completely abolished (Fig. 3C, iv). Finally, we compared the affinity of Fis for the dps100 DNA fragment and a DNA fragment derived from the E. coli nrfA promoter, which contains two previously characterized binding sites for Fis (Browning et al., 2005). The data show that, although dps100 has a lower affinity for Fis, similar concentrations of Fis are required to retard the migration of the dps100 and nrfA DNA fragments and that the affinity of the dps100 FisI site falls between that of the two nrfA Fis binding sites (Fig. S2).
As the DNA targets for Fis at the dps promoter overlap the −10 and −35 elements, we reasoned that Fis might repress transcription by blocking the binding of RNA polymerase. To test this, we used EMSA to measure the binding of RNA polymerase to naked dps promoter DNA and to dps promoter DNA saturated with Fis. As the dps promoter can be recognized by both Eσ70 and Eσ38, we tested both forms of RNA polymerase. Our experiment, illustrated in Fig. 4, showed that both Eσ70 and Eσ38 bind to the dps promoter (lanes 3 and 5), resulting in several complexes with much lower mobility than the complexes formed with Fis (lane 2). Surprisingly, when RNA polymerase was added to DNA fragments saturated with Fis, RNA polymerase binding was not inhibited (lanes 4 and 6) and we observed distinct low mobility protein–DNA complexes, which must correspond to ternary Fis–RNA polymerase–DNA complexes.
Regulation of dps promoter activity by Fis in vitro
To investigate the effects of Fis on transcription from the dps promoter in vitro, the dps100 fragment was cloned upstream of the factor-independent λoop transcription terminator in plasmid pSR. Hence, transcription from the dps100 promoter produces a 130-base transcript that can be quantified after electrophoresis. Figure 5 shows the results of an in vitro transcription assay, using pSR plasmid carrying the dps100 fragment as a template, with different combinations of Eσ70, Eσ38 and Fis. Bands corresponding to two transcripts are visible in each lane. The larger 130-base transcript initiates from the dps100 promoter while the smaller 108-base RNAI transcript from the plasmid replication origin acts as an internal control. Lanes 1 and 2 show transcripts produced by Eσ38 and Eσ70 respectively. While both forms of RNA polymerase produce the dps100 transcript, Eσ38 is more proficient. As expected, together, Eσ38 and Eσ70 produced an additive amount of transcript (lane 3). We then repeated the experiment in the presence of Fis (lanes 4–6). The results show that, while Fis only has a small negative effect on dps100 transcription by Eσ38 (compare lanes 1 and 4), it reduced transcription by Eσ70 by 10-fold (compare lanes 2 and 5). Surprisingly, in the presence of both Eσ70 and Eσ38, dps100 transcription was greatly reduced by Fis (compare lanes 3 and 6). Thus, although Fis alone cannot repress transcription by Eσ38, a combination of Fis and Eσ70 blocks transcription by the stationary-phase RNA polymerase.
Taken together, our results suggested that Fis represses transcription from the dps promoter by binding in conjunction with RNA polymerase and inhibiting transcription at a step prior to promoter escape. Interestingly, Eσ70 appears to be particularly sensitive to Fis and can act as a ‘co-repressor’. To understand the role of Fis further, we conducted KMnO4 footprinting experiments to monitor open complex formation by RNA polymerase. These experiments, illustrated in Fig. 6, show that some opening of the dps promoter at position −12 occurs in the absence of RNA polymerase (lane 1). More extensive opening of the promoter is seen in the presence of RNA polymerase and we found that Eσ38 (lane 2) was more proficient at producing the open complex than Eσ70 (lane 3). Titration of Fis into reactions containing either Eσ38 (lanes 4–7) or Eσ70 (lanes 8–11) resulted in a decrease in open complex formation, suggesting that Fis functions by inhibiting promoter opening. Eσ70 is more sensitive to Fis than Eσ38 (Fig. 6B) and we suggest that this is because Eσ70 is less able to form an open complex at the dps promoter. We note that lower concentrations of Fis were required in KMnO4 footprinting, compared with in vitro transcription, assays. This may be because of differences in the nature of the DNA template (i.e. linear as opposed to supercoiled plasmid DNA).
Fis binding to the dps promoter from related bacteria
It is unlikely that Fis can occupy the FisII site in the ternary Fis–RNA polymerase–dps promoter complex because FisII completely overlaps the −10 hexamer (Fig. 3A). To investigate this further, we aligned the dps promoter sequence of E. coli K-12 with that from some related bacteria (Fig. 7A). Most of the conserved DNA sequences aligned with core promoter elements and conserved sequences were also found in the two binding sites for Fis. For FisI, the conserved base pairs are located at the extremities of the binding site. In contrast, for FisII, the conserved sequences are located in the centre of the site (also corresponding to the −10 hexamer). Recall that DNA-binding specificity of Fis is largely dictated by the first and final positions in the target sequence (Hengen et al., 1997), we therefore speculate that FisI is better conserved than FisII.
The Salmonella typhimurium LT2 dps promoter has a T at position −3 (position 15 of FisII) and is thus predicted to have non-functional FisII site (Fig. 7A). To test this, a DNA fragment equivalent to the E. coli K-12 dps100 promoter fragment was generated, end-labelled, and Fis binding was measured by an EMSA. Results, illustrated in Fig. 7B, confirm that dps100 fragment from S. typhimurium contains only one high affinity Fis binding site. We then cloned the S. typhimurium dps100 fragment into pSR plasmid and used our in vitro transcription assay to measure the effects of Fis on promoter activity, as above. Results illustrated in Fig. 7C show that the S. typhimurium dps promoter exhibits a similar response to Fis as the E. coli K-12 promoter, suggesting that (for S. typhimurium at least) the FisII site is dispensable.
In an additional experiment we assayed the E. coli dps100–19G promoter in our in vitro transcription assay and found that, in addition to disrupting Fis binding, the −19G mutation altered the basic properties of the promoter. For this reason, we have not characterized this promoter variant further. The data (Fig. S3) show that Eσ70 had reduced activity at the −19G promoter and was able to repress transcription by Eσ38 in the absence of Fis.
Binding of H-NS to the dps promoter and regulation by H-NS in vitro
DNase I footprinting was used to investigate the binding of purified H-NS to the E. coli K-12 dps promoter (Fig. 8A). At low H-NS concentrations, protection extends from positions −13 to +12 (compare lanes 1 and 2). At higher H-NS concentrations (lanes 3–7), the region of protection extends further upstream to position −17 and several sites of hyper-sensitivity to DNase I attack become visible within the H-NS footprint. The protected region contains a 10-base-pair DNA element, centred at position +2.5, with a 7/10 match to the recently identified consensus target for H-NS binding (Bouffartigues et al., 2007; Lang et al., 2007), with the three non-matching base pairs corresponding to the second preference at that position (Fig. 8B).
To investigate the effects of H-NS on transcription from the dps promoter in vitro, the dps400 fragment was cloned upstream of the factor-independent λoop transcription terminator in plasmid pSR. Figure 9 shows the results of in vitro transcription assays with different combinations of Eσ70, Eσ38 and H-NS.
First, we measured the effect of H-NS on transcription driven by Eσ70 at 22°C and 37°C (Fig. 9A). We found that H-NS repressed transcription by Eσ70 at 22°C (lanes 1–4) but not 37°C (lanes 5–8), consistent with our in vivo observations. Recall that Fis and Eσ70 were able to co-repress transcription by Eσ38. Thus, we next measured the effect of H-NS on transcription by Eσ38 at 22°C, with and without Eσ70 (Fig. 9B). H-NS had no effect on transcription by Eσ38 (lanes 3–5) and a combination of H-NS and Eσ70 has little effect on Eσ38-dependent transcription (lanes 9–11). Similar results were obtained with the shorter dps100 promoter fragment (data not shown).
As H-NS and Eσ70 could not act as co-repressors, we reasoned that, in contrast to Fis, H-NS was likely to repress transcription by Eσ70 via a simple blocking mechanism. To test this, we used an EMSA to investigate the binding of H-NS, Eσ70 and Eσ38 to the dps promoter (Fig. 10). The experiment shows that, when the dps promoter is saturated with H-NS, binding of Eσ70 is greatly reduced (compare lanes 3 and 4). Conversely, we found that H-NS only had a small negative effect on the binding of Eσ38 and that a rearrangement of the complexes formed by Eσ38 occurred when H-NS was present (compare lanes 5 and 6).
Transcription from E. coli dps promoter is growth rate regulated and dependent on the stationary-phase RNA polymerase, Eσ38 (Altuvia et al., 1994; Fig. 2). However, as for many Eσ38-dependent promoters, it is not clear how promoter selectivity is achieved; the dps promoter has a good match to the extended −10 element recognized by Eσ70 and, in vitro, Eσ70 and Eσ38 have similar affinities for the dps promoter (Fig. S4). In our previous study of chromosome-wide DNA binding by nucleoid proteins we observed clear in vivo binding signals for both Fis and H-NS at the dps promoter (Grainger et al., 2006). As Fis and H-NS had previously been shown to play important roles at σ38-dependent promoters, we conducted a detailed study of Fis and H-NS action at the dps promoter.
Taken together, our results show that σ factor selectivity at the dps promoter is defined by both its DNA sequence and the local nucleoprotein environment. Thus, although Eσ70 is able to bind to the promoter it is deficient in its ability to form an open complex and initiate transcription. Both Fis and H-NS further reduce transcription initiation by RNA polymerase containing σ70 while RNA polymerase containing the stationary-phase σ38 factor is hardly affected. Interestingly, Fis and H-NS have both previously been shown to selectively regulate transcription by Eσ70 and Eσ38, although by mechanisms that differ to those presented here (reviewed in Typas et al., 2007a). For example, at the proP2 promoter, Fis selectively activates transcription by Eσ38 but not Eσ70, due to the unusual organization of the core promoter elements for RNA polymerase (Typas et al., 2007b). A complex mechanism of repression by H-NS has been identified at the hdeAB promoter, where H-NS stabilizes a repression loop that represses transcription by Eσ70 but not Eσ38 (Shin et al., 2005). Here, at the dps promoter, Fis binds to a site (FisI) in the spacer region between the −10 and −35 hexamers to trap Eσ70 at the promoter, while H-NS prevents promoter binding by Eσ70 (see Fig. 11). Thus Fis, but not H-NS, can act together with Eσ70 as a co-repressor, and this can block transcription by Eσ38. Consistent with our model, Oshima et al. (2006) previously found that RNA polymerase was bound at the dps promoter, but not coding region, in growing E. coli cells.
Our model predicts that H-NS mediated repression of the dps promoter can be overcome by Eσ38 and that repression by Fis is reduced as cell density increases and Fis levels decrease. However, Fis can be found in cells during oxidative stress and cells that have been grown to stationary phase in specific conditions (Cróinín and Dorman, 2007). In such situations it is important that transcription from the dps promoter is not blocked by Fis. Thus, we suggest that the FisI binding site has a relatively low affinity for its cognate factor to ensure that it is only occupied during rapid growth, when Fis levels are at their peak. In conclusion, expression of Dps protein is the major factor responsible for super-compaction of the nucleoid and its expression is induced as E. coli cells enter stationary phase and Eσ38 levels increase. As Fis and H-NS are abundant during exponential growth, our results have important implications for the regulation of growth-phase transitions in nucleoid structure and highlight the importance of transcriptional cross-talk between nucleoid proteins.
Strains, plasmids and oligonucleotides
Bacterial strains and plasmids are listed in Table S1. Standard techniques for recombinant DNA manipulations were used throughout. Table S2 lists primers used to amplify sections of the dps promoter in such a way that they are flanked by EcoRI and HindIII restriction sites. After digestion, fragments carrying dps promoter derivatives were cloned into pSR, sequenced and then subcloned into pRW50. By convention, promoter sequences are numbered with respect to the transcription start point (+1) and with upstream and downstream locations denoted by ‘−’ and ‘+’ prefixes respectively. The full sequences of the cloned promoter fragments are shown in Fig. 1.
Construction of MG1655Δhns
A 1.18 kb PCR product containing hns and upstream sequence was generated using primers hns7 and hnsR, and digested with SphI and BamHI. The hns promoter and 5′ end of the structural gene were removed by digesting the PCR product with FspI, which cut twice, yielding three DNA fragments. The SphI–FspI and FspI–BamH fragments were cloned into the sacB-based suicide vector pCVD442 to form pMDG6 and transformed into SM10λpir (Donnenberg and Kaper, 1991). pMDG6 was introduced into MG1655 rpsL by conjugation (Donnenberg and Kaper, 1991). Merodiploids resulting from the conjugation were selected by incubation on LB supplemented with 100 μg ml−1 streptomycin and 100 μg ml−1 ampicillin. Resolution of the merodiploids was selected by incubating overnight in LB medium followed by growth at 30°C on LB medium containing 5% (w/v) sucrose. Sucrose-insensitive colonies were tested for deletion of the hns promoter by PCR using the hns7 and hnsR primers. Successful construction and integrity of the mutant was confirmed by comparative genomic hybridization (data not shown).
DNA fragments containing the dps promoter were cloned into pRW50 to generate dps::lacZ fusions. β-Galactosidase levels in cells carrying these recombinants were measured by the Miller (1972) method. Activities are shown in Miller units and are the average of three or more independent experiments with a standard deviation of < 10%. Cells were grown aerobically in LB media and assays were performed in either JCB387 or the derivative JCB3871Δfis.
Core E. coli RNA polymerase was purchased from Epicenter (Madison). Preparations of σ70 and σ38 were made by overexpression of the cloned rpoD and rpoS genes in BL21 DE3 cells (Tang et al., 1995). Inclusion bodies were then solublized in 6 M Guanidine HCl, before being dialysed into buffer containing 20 mM Tris, 100 mM NaCl and 10% Glycerol. Proteins were bound to a DEAE anion exchange column (Amersham) and eluted with a linear gradient to 1 M NaCl. RNA polymerase holoenzyme was reconstituted by incubating core RNA polymerase with equimolar amounts of σ70 and σ38 at room temperature for 20 min. When a combination of Eσ70 and Eσ38 was required, both holoenzymes were first reconstituted separately and then mixed. Note that neither σ factor was in excess in these preparations, thus avoiding problems of σ factor competition. Fis protein was purified according to the method of Pan et al. (1996) and H-NS, purified according to the method of Smyth et al. (2000), was donated by John Ladbury (UCL).
DNase I and KMnO4 footprinting
Purified AatII–HindIII DNA fragments were derived from maxi-preparations (using a Qiagen maxiprep kit) of plasmid pSR carrying either the dps200 or dps400. Fragments were labelled at the HindIII end using [γ-32P]-ATP and polynucleotide kinase. Footprints were performed at 37°C for Fis and 22°C for H-NS, as in our previous work (Savery et al., 1996). DNA fragments were used at a final concentration of 10–40 nM in buffer containing 20 mM Tris pH 7, 10 mM MgCl2, 100 μM EDTA and 120 mM KCl. DNase I footprints also contained 12.5 μg ml−1 Herring sperm DNA as a non-specific competitor. Note that protein stocks of Fis and H-NS were diluted in reaction buffer, containing competitor DNA, prior to addition to the reaction. Footprints were analysed on a 6% DNA sequencing gel (molecular dynamics). The results of all footprints and EMSA experiments were visualized by exposing the dried gel against a Fuji phosphor screen and analysed using a phosphorimager and Quantity One software. The KMnO4 footprint shown in Fig. 6 was quantified by measuring the intensity of bands observed upon the addition of RNA polymerase (between promoter positions −4 and −10), using the control lane 1 as a background signal. Similar results were obtained in a repeat experiment.
Electrophoretic mobility shift assays
DNA fragments for EMSA experiments were generated by PCR amplification using the appropriate DNA primers and pSR derivatives as a template. PCR products were purified, cut with HindIII and end-labelled using [γ-32P]-ATP and polynucleotide kinase. DNA fragments were then incubation with purified proteins, at 37°C for Fis and 22°C for H-NS, in buffer containing 20 mM Tris pH 7, 10 mM MgCl2, 100 μM EDTA, 120 mM KCl and 12.5 μg ml−1 Herring sperm DNA. Note that protein stocks of Fis and H-NS were diluted in reaction buffer, containing competitor DNA, prior to addition to the reaction. Reactions were loaded under tension onto a 5% polyacrylamide gel, run in 0.5× TBE at 160 V for 2–4 h and analysed as described above.
In vitro transcription assays
The in vitro transcription experiments were performed as described previously (Savery et al., 1998) using the system of Kolb et al. (1995). A Qiagen maxiprep kit was used to purify supercoiled pSR plasmid carrying the dps100 or dps400 dp promoter inserts. This template (16 μg ml−1) was pre-incubated with purified Fis (at 37°C) or H-NS (at 22°C unless stated otherwise) in buffer containing 20 mM Tris pH 7.9, 5 mM MgCl2, 500 μM DTT, 50 mM KCl, 100 μg ml−1 BSA, 200 μM ATP, 200 μM GTP, 200 μM CTP, 10 μM UTP with 5 μCi [α-32P]-UTP. The reaction was started by adding purified E. coli Eσ70 and/or Eσ38. Labelled RNA products were analysed on a denaturing polyacrylamide gel. Transcripts were quantified using Quantity One software.
We thank John Ladbury for purified H-NS, Joseph Wade, Lars Westblade and Nigel Savery for helpful discussions, and the Wellcome Trust for funding this work with a programme grant awarded to S.J.W.B.