Differential regulation of Escherichia coli topoisomerase I by Fis

Authors


*E-mail shoshy@cc.huji.ac.il; Tel. (+972) 2675 7212; Fax (+972) 2675 7308.

Summary

We previously reported that the P1 promoter of topA encoding topoisomerase I of Escherichia coli is activated in response to oxidative stress, in a Fis-dependent manner. Here we show that Fis regulation of topA varies with the intracellular concentrations of Fis. Thus, when Fis levels are low, hydrogen peroxide treatment results in topA activation, whereas at high Fis levels hydrogen peroxide treatment renders topA P1 inactive. In vivo DMS footprinting indicates that only at low Fis levels, when exposed to the stress, the region of the topA promoter changes and P1 becomes active. Potassium permanganate experiments indicate that low levels of Fis activate P1 transcription by facilitating the formation of open complexes, while high levels of this protein shut off the promoter. DNase I footprinting show that Fis binds the promoter region of topA at eight sites with different affinities. One low affinity site overlaps the −10, −35 hexamers of RNA polymerase. We propose that in response to oxidative stress, when present at low levels, Fis binds the promoter region of topA at its high affinity sites, thereby facilitating the recruitment of RNA polymerase to P1, while at high levels, Fis occupies the low affinity sites as well, and thus prevents the binding of RNA polymerase. Our results indicate that the oxidative stress response varies in response to changes in growth phase and nutritional environment.

Introduction

Changes in the nutritional environment require a major reconfiguration of cellular responses, which in Escherichia coli are governed to a large extent by Fis. Fis (factor for inversion stimulation) is a small homodimeric protein involved in a number of processes that require the assembly of complex nucleoprotein structures containing bent DNA (reviewed in Finkel and Johnson, 1992). Initially, Fis was identified by its ability to stimulate Hin- Gin- and Cin-mediated DNA inversion, and Lambda recombination (Johnson et al., 1986; Koch and Kahmann, 1986; Haffter and Bickle, 1987; Thompson et al., 1987). Subsequent studies showed that in addition to mediating specialized site-specific DNA recombination events, Fis acts in a wide range of cellular processes including initiation of DNA replication after temperature upshift (Gille et al., 1991; Filutowicz et al., 1992), cold shock response (Brandi et al., 1999), catabolism of sugars and nucleic acids (Gonzalez-Gil et al., 1996; Browning et al., 2004; 2005), DNA supercoiling (Schneider et al., 1997), oxidative stress (Weinstein-Fischer et al., 2000) and the translation apparatus regulating the expression of components such as rRNA, tRNA and M1 RNA (rnpB), the catalytic subunit of E. coli RNase P (Ross et al., 1990; Kim et al., 2004). Fis also appears to affect gene expression by stabilizing local DNA architecture at promoters (Schneider et al., 2001). The number of Fis binding sites at promoters regulated by Fis varies from one site at leuV and nrf (Ross et al., 1999; Browning et al., 2005), three to five sites at rrn P1 (in Schneider et al., 2003), four at rnpB (Kim et al., 2004) and at least seven sites at the promoter of H-NS (Falconi et al., 1996). In many cases, additional proteins act together with Fis to control the expression of these genes and both repression and activation were detected (reviewed in Finkel and Johnson, 1992).

In accordance with its multiple role, the expression of fis itself is governed by a variety of control mechanisms. As a result, Fis protein levels vary dramatically during cell growth ranging from the most abundant transcriptional regulator in early exponential cells, to almost undetectable levels by late logarithmic growth phase (Ball et al., 1992; Nilsson et al., 1992; Ninnemann et al., 1992; Azam and Ishihama, 1999). The expression pattern of Fis is primarily attributed to transcription regulation and not to changes in mRNA turnover rates. Fis transcription is subject to a plethora of control mechanisms including, autoregulation, stringent control, the availability of its initiating NTP and the CRP transcription factor (Ball et al., 1992; Ninnemann et al., 1992; Nasser et al., 2001; Walker and Osuna, 2002; Mallik et al., 2004; Walker et al., 2004). In addition, fis transcription is positively regulated by the DNA-bending protein, IHF, and is enhanced by increases in negative supercoiling (Pratt et al., 1997; Schneider et al., 2000). A recent study has demonstrated that fis is also regulated at the level of translation by BipA, a new member of the elongation factor GTPase super-family, whose expression is growth rate dependent (Owens et al., 2004).

Previously, we showed that the response of E. coli to hydrogen peroxide involves transcription activation of the topA P1 promoter, leading to a transient decrease in the negative supercoiling of the DNA. We also showed that Fis, a nucleoid-associated protein, mediates the peroxide-dependent activation of topA P1 and that a fis deficient mutant strain is extremely sensitive to hydrogen peroxide (Weinstein-Fischer et al., 2000). Here, we extend our characterization of the control of topA expression by Fis. We show that the hydrogen peroxide-dependent regulation of topA by Fis varies with the intracellular concentrations of Fis. Thus, low levels of Fis induce the transcription of topoisomerase I, whereas high levels of Fis result in its repression.

Results

topA P1 activation by hydrogen peroxide correlates with low levels of Fis

We previously reported that the P1 promoter of topA is induced by hydrogen peroxide in a Fis-dependent manner. The intracellular concentration of Fis in cells grown in rich media varies dramatically during the logarithmic growth phase (Ball et al., 1992; Nilsson et al., 1992; Ninnemann et al., 1992; Azam and Ishihama, 1999; Owens et al., 2004). Fis protein expression sharply increases upon subculturing of stationary phase cells (upshift), and then decreases steeply with growth (Ball et al., 1992; Nilsson et al., 1992; Ninnemann et al., 1992; Azam and Ishihama, 1999; Owens et al., 2004). These shifts in the protein level correspond to the shifts in fis mRNA levels (Ball et al., 1992). We examined the shifts in Fis protein levels in cells diluted in rich media and found that in accordance with the previous reports, the expression of Fis increases upon subculturing (OD600 of 0.05–0.08), and then decreases with growth (OD600 of ≥ 0.2) (Fig. 1A). Thus, to learn about the relationships between Fis levels and topA induction by hydrogen peroxide, we examined topA transcription in cells exposed to hydrogen peroxide upon subculturing, when Fis protein levels are high, and during late logarithmic growth phase, at low levels of Fis. Interestingly, we found, that while fis mRNA and protein levels were high upon subculturing (early exponential; EE) and low during late exponential (LE) phase (Fig. 1B and C, lanes 1 and 3), the pattern of topA P1 expression was the opposite. No expression of the topA P1 promoter could be detected when the cells were exposed to hydrogen peroxide right after dilution, when fis levels were high (Fig. 1B, lane 2). topA P1 activation by hydrogen peroxide was detected only later on, when fis mRNA and protein levels declined (Fig. 1B, lane 4) indicating that the peroxide-dependent activation of topA correlated with decreased levels of Fis. In addition, we found that the peroxide treatment of these cells resulted in a further decrease in fis expression at both time points (Fig. 1B and C).

Figure 1.

topA activation and Fis levels.
A. Fis levels shift during exponential growth. Immunoblot analysis. MC4100 overnight cultures diluted 1/100 in fresh LB medium were grown to early exponential (OD600 of 0.05–0.08; EE) and late exponential (OD600 of 0.2–0.6; LE) phase. Samples corresponding to equal amounts of cells were then analysed by SDS-PAGE and probed with α-Fis antibody as described in Experimental procedures.
B and C. topA P1 activation correlates with low fis mRNA (B) and protein (C) levels. MC4100 overnight cultures were diluted and grown to early exponential (EE) and late exponential (LE) phase as described in Experimental procedures and then exposed to H2O2 (1 mM). Samples were taken from the cultures prior to (–) and at 10 min of treatment (+).
B. Primer extension reactions were carried out using topA (401) and fis (650) specific primers.
C. Immunoblot analysis using α-Fis antibody.
D. Effects of Fis overexpression on topA P1 activation. MC4100 overnight cultures carrying the plasmids pPtac lacI (lanes 1–4) or pPtac-fis lacI (lanes 5–8) were diluted and grown to OD600 of 0.2. Total RNA was extracted from an untreated culture (lane 1) and cultures exposed to IPTG at OD600 of 0.1 and grown to OD600 of 0.2 (1 mM; lane 2) or to H2O2 (1 mM for 10 min at OD600 of 0.2; lanes 3, 5, 7) as well as from cultures exposed to IPTG prior to the addition of H2O2 (lanes 4, 6, 8). The analysis of two colonies of MC4100/pPtac-fis lacI is shown (lanes 4, 5 and 6, 7). Primer extension reactions were carried out using a topA (401) specific primer.

To examine whether high levels of Fis were responsible for the lack of topA P1 activation upon exposure to hydrogen peroxide, we cloned the fis gene downstream of the tac promoter, under the control of the LacI repressor and repeated the experiments in cells carrying an inducible fis-expressing plasmid. Cultures carrying plasmid Ptac-fis-lacI or the empty vector Ptac-lacI were treated with IPTG and exposed to hydrogen peroxide during late exponential phase when the chromosomally encoded fis gene is expressed at very low levels. RNA samples were extracted from treated and untreated cells and subjected to primer extension. Figure 1D (lanes 3 and 4) shows the activation of topA P1 by hydrogen peroxide in cells carrying the control plasmid Ptac-lacI. In contrast, no transcription of topA was detected in cells carrying the plasmid Ptac-fis-lacI, even in the absence of IPTG, suggesting that the unregulated basal level of fis produced from the multi-copy number plasmid suffices to inhibit this induction (Fig. 1D lanes 5–8). These results show that the Fis-dependent induction of topA P1, in response to oxidative stress, correlates with a decrease in Fis levels, and that high expression of Fis during late log phase when endogenous Fis levels are low prevents this induction.

Fis binds to the promoter region of topA at eight sites

Previously, we demonstrated that Fis binds three sites at the promoter region of topA (I, II and III) centred at approximately −62, −93 and −129 upstream of the transcription start site of P1 (Weinstein-Fischer et al., 2000). Our finding that topA activation by hydrogen peroxide correlates with low levels of Fis and that high Fis levels inhibit this induction, prompted us to extend our characterization of Fis binding to the promoter region of topA. DNase I footprint analysis using purified Fis, demonstrated that Fis binds to five additional sites; two (B and A) downstream and three (IV, V and VI) upstream of the previously identified sites (Fig. 2). Site B is centred at +20 downstream of the transcription start site, while site A overlaps the RNA polymerase −10, −35 hexamers of P1 carrying two possible Fis-binding sequences centred at −10 and −30 (Fig. 3A and B). The other three sites (IV, V and VI) were mapped further upstream. Sites IV and VI each carry two Fis consensus sequences, centred at −157/−178 and −230/−256, respectively, while site V carries one Fis binding sequence centred at −199 (Fig. 3B). The binding of Fis to the promoter also induced enhanced DNase I cleavage, possibly reflecting Fis-induced DNA distortion as has been noted for other Fis targets (Finkel and Johnson, 1992).

Figure 2.

DNase I footprint of Fis bound to wild-type topA promoter region. The fragment of the full-length topA promoter region was amplified using primers 686 and 687. The brackets indicate the sites (B to VI) protected on the top and bottom strands. Footprint reactions were carried out at 25°C as described in Experimental procedures. Fis protein was present at the concentrations indicated. Stars indicate enhanced DNase I cleavage within the Fis sites.

Figure 3.

A. Schematic representation of the promoter region of topA. The sites B to VI are indicated on the full-length fragment. Solid lines represent DNA fragments used in the footprint and the gel mobility assays. The position of primers used to amplify the DNA fragments is shown. The numbering indicates the position of the far-left and the far-right 5′ ends of the primers, upstream or downstream, of the transcription start site of P1.
B. Sequence of topA promoter region. The brackets indicate the protected sites B to VI on both strands. Asterisks indicate enhanced DNase I cleavage within the Fis sites. The similarities of the Fis sites to the consensus derived by Finkel and Johnson (Finkel and Johnson, 1992) [Gnn(c/t)(A/g)(a/t)(a/t)(T/A)(t/a)(t/a)(T/c)(g/a)nnC] are indicated by lines between the top and bottom strands. Dots indicate poorly conserved positions in Fis sites. Two possible Fis consensus sites are shown in the protected regions IV and VI. The bases matching the −35 and −10 hexamers of the σ70 consensus (TTGACA and TATAAT, respectively) are underlined within boxes. Also indicated are the transcription start sites that correspond the promoters P2, P4 and Px1 (see text).

The use of increasing concentrations of Fis enabled us to conclude that Fis binds these sites with different affinities; sites III and VI were bound by Fis with the greatest affinity, whereas the rest of the sites showed decreased binding affinity. Interestingly, site A that overlaps the −10, −35 hexamers appeared to be at the end of the scale with the lowest binding affinity.

Fis binds the topA promoter region in a cooperative manner

To analyse the pattern of Fis binding and the relative importance of the different regions, we examined its binding to the topA promoter region by electrophoretic mobility shift assay (EMSA). A fragment carrying the eight characterized binding sites (Fig. 3A; primers GO8-401) was amplified and labelled by polymerase chain reaction (PCR) and the DNA was incubated with purified Fis. The native gel shows that incubation of the full-length fragment with 1 μM of Fis resulted in the formation of eight high molecular weight complexes, a result that is in accordance with our mapping of Fis binding at the topA promoter region (Fig. 4, top left panel). Incubation of this fragment with lower amounts of Fis (0.1–0.5 μM) caused the appearance of one strong, and two weak bands with decreased mobility that represent binding to high and low affinity sites respectively. The pattern of topA binding obtained with increasing amounts of Fis further supports the observation of the high and low affinity sites at this promoter region and indicates that Fis progressively associates in a cooperative manner with multiple binding sites present in each DNA fragment. The calculated Hill coefficient of Fis binding to this fragment is 2.07 ± 0.52. Hill coefficient values represent degrees of cooperative binding; values of 1 represent little or no cooperative binding, whereas values greater than 1 represent positive cooperative binding (Hill, 1910). The Hill coefficient of 2 supports the observation that Fis binding to the promoter region of topA is of a cooperative nature.

Figure 4.

Fis binds topA promoter region in a cooperative manner (EMSA). End labelled DNA fragments of the topA promoter region were incubated with increasing concentrations of Fis as described in Experimental procedures. The fragments carrying sites B-VI, I-VI, B-III and I-III were PCR-amplified using primers GO8-401, GO8-402, 418-401, 418-402 respectively. The full-length fragment, B-VI, carries eight Fis binding sites. The fragment I-VI lacks the downstream sites B and A. The fragment B-III has been deleted for the upstream sites IV, V and VI. The fragment I-III carries the middle three sites and lacks both the upstream IV, V and VI and the downstream sites B and A.

To examine the relative importance of the different regions on the pattern of Fis binding, we deleted the downstream sites A and B (Fig. 3A, primers 686-402 and GO8-402) leaving sites I to VI intact. We found both by EMSA (Fig. 4 top right panel) and by DNase I footprinting (Fig. 5A) that the deletion of these sites had a relatively small effect on the binding of Fis to the remaining sites. The calculated Hill coefficient of Fis binding to this fragment is 1.75 ± 0.32. In contrast, as demonstrated by the mobility shift assays, the deletion of the upstream sites IV, V and VI (Fig. 4, bottom left panel and Fig. 3A, primers 418-401) had a significant effect on the cooperative pattern of Fis binding (Hill coefficient value of ≈ 1). Although the deleted topA fragment included five Fis binding sites (B, A, I, II, III), only one strong and one weak band with decreased mobility could be observed when this DNA fragment was incubated with high Fis levels. DNase I footprinting experiment carried out using the same DNA fragment demonstrated that the deletion of these upstream sites (IV, V and VI) (Fig. 3A, primers 418-687) decreased the binding affinity of Fis to the remaining sites, except for site III that bound Fis with a similar high affinity (Fig. 5B). Based on the footprint data and the data obtained from the mobility assay of this fragment, it can be concluded that the one complex obtained in the mobility assay results from the binding of Fis to site III. A similar pattern of binding was obtained when this fragment was deleted further for the B, A region (Fig. 3A, primers 418-402 and Fig. 4, bottom right panel) further indicating that the B, A region has no effect on the pattern of Fis binding. Together the data show that Fis binds the topA promoter region at multiple sites, in a cooperative manner. The presence of the upstream sites (IV to VI) grants the cooperative nature of this binding, while the downstream sites (sites A, B) have a marginal effect on the binding pattern of Fis to this region.

Figure 5.

DNase I footprint of Fis bound to mutated topA promoter.
A. Fis binding to topA promoter region carrying sites I to VI. The deletion of the downstream sites B and A was obtained by using the primers 686-402 (see also Fig. 4A).
B. Fis binding to topA promoter region carrying sites B to III. Deletion of the upstream sites IV, V and VI was obtained using primers 418-687. The brackets indicate the sites protected on the top and bottom strands. Footprint reactions were carried out at 25°C as described in Experimental procedures. Fis protein was present at the concentrations indicated. Stars indicate enhanced DNase I cleavage within the Fis sites.

The binding of Fis affects the activity of a promoter located upstream of P1

Previous transcription studies to map the 5′ untranslated region of topA have identified, in addition to the P1 promoter, three additional transcription start sites denoted P2, P4 and Px1 (Tse-Dinh and Beran, 1988; Lesley et al., 1990; Qi et al., 1997). Px1 was found to be induced in stationary phase, while P2 and P4 were characterized as logarithmic growth phase promoters (Qi et al., 1997). We observed that the weak Fis affinity site IV, overlaps the sequence of P4, while binding site III overlaps the −35 hexamer of promoter P2 (Fig. 3B). These observations prompted us to examine the activity of these promoters under conditions of low and high Fis levels. RNA analysis carried out with samples taken from cultures upon subculturing and during late exponential phase showed that P4 was active upon subculturing when Fis levels were high but more so during late logarithmic phase, under conditions of low Fis (Fig. 6). In fis-deficient cells, P4 transcription level did not change with cell growth, while overexpression of Fis from a multi-copy number plasmid, rendered P4 completely inactive (Fig. 6 and not shown). Presumably, P4 that overlaps the low affinity site IV is sequestered under conditions of high Fis. Unlike P1, exposure to hydrogen peroxide led to a decrease in the activity of P4 in wild type as well as in fis-deficient cells, indicating that independent of Fis, P4 becomes inactive under oxidative stress conditions (Fig. 6). In our hands, the P2 promoter remained inactive under the conditions examined as well as in fis-deficient cells (not shown).

Figure 6.

topA P1, P4 and fis RNA levels. RNA samples were extracted from wild type and fis-deficient cultures at early (EE) and late exponential (LE) phase, before and after exposure to H2O2 (1 mM for 10 min). Primer extension reactions were carried out using topA (401) and fis (650) specific primers.

In vivo detection of P1 promoter activity by DMS footprinting

To closely examine the changes occurring at the promoter region of topA, in cells exposed to hydrogen peroxide during the logarithmic growth phase, we conducted in vivo dimethylsulphate (DMS) protection assays. DMS modifies mainly guanines and to a lesser extent, adenines in the major groove of the DNA, and provides a sensitive method for probing protein footprints in vivo (Sasse-Dwigh and Gralla, 1991). Wild type and fis-deficient cells that harbour a plasmid carrying the entire promoter region of topA (pPtopA B-VI) were diluted and grown to early and late exponential phase and then the cultures were exposed to hydrogen peroxide for 10 min prior to the addition of DMS. Plasmid DNA was extracted and treated with piperidine to cleave unprotected methylated bases and the products were analysed by primer extension. The gel shows that exposure of wild-type cells to hydrogen peroxide at the early stages of upshift does not change the primary methylation protection pattern that is observed in the untreated cultures (Fig. 7 compare lanes 2 and 3). In contrast, exposure of these cells at a later stage of the exponential growth resulted in a significant change in the protection pattern (Fig. 7 compare lanes 4 and 5). The changes were detected at the region of site A that overlaps the −10 and −35 hexamers of P1, site I, site II, as well as downstream of P1 transcription start site (site B). These results indicate that the region of the topA promoter changes and P1 becomes active when exposed to oxidative stress during late exponential phase but remains inactive when exposed to hydrogen peroxide at the early stages of upshift. Hydrogen peroxide exposure of this promoter in late logarithmic fis-deficient cells, had no effect on the pattern of methylation, because no changes could be detected (Fig. 7 lanes 6 and 7), suggesting that in the absence of Fis, the promoter remains inactive. In addition, the P1 promoter in the fis-deficient cells exhibits excessive methylation, indicating that in the absence of Fis-binding, this region is mostly exposed and unprotected from DMS. Interestingly, in the absence of hydrogen peroxide, the methylation pattern of the core promoter region (site A) in subcultured cells was similar to the pattern observed in late log cells (Fig. 7 compare lanes 2 and 4). Possibly, the differential binding of Fis to this region cannot be detected by this assay or, under normal growth, Fis-binding is similar in early and late exponential phase cells, and it changes in late log cells, under conditions of oxidative stress.

Figure 7.

In vivo detection of P1 promoter activity by DMS footprinting. Bacterial cultures (wild type and fis-deficient) carrying plasmid pPtopA B-VI were diluted and grown in fresh LB to early (EE) and late exponential (LE) phase. Thereafter, the cultures were exposed to hydrogen peroxide prior to the addition of DMS and the plasmid DNA was extracted and treated as described in Experimental procedures. Asterisks and underlines denote unprotected methylated bases and protected non-modified residues in late log wild-type cells exposed to hydrogen peroxide, compared with non-treated cells respectively (lane 5). The left and right brackets indicate the position of the Fis binding sites and of the hexamers respectively. The sequences of −10 and −35 hexamers are boxed. The numbering indicates the position upstream (–) or downstream (+) of the transcription start site of P1.

Sites B to VI are necessary and sufficient for the hydrogen peroxide dependent control of P1

To examine the importance of an intact Fis binding region for the peroxide-dependent topA P1 induction, we cloned the promoter region of topA including sites B to VI, on a multi-copy number plasmid (pPtopA B-VI) and analysed topA mRNA levels upon subculturing and during late exponential phase, prior to and after exposure to hydrogen peroxide. Primer extension assays carried out using a primer specific to the plasmid-encoded topA promoter region, demonstrated that Fis-binding sites B to VI were sufficient to enable induction of topA transcription when treated during the late exponential phase (Fig. 8). Similar to the chromosomal-encoded topA gene, the cloned P1 promoter remained inactive when exposed to the stress right after subculturing. Mutating the P1 promoter region by deleting sites IV to VI (pPtopA B-III) to change the cooperative pattern of Fis binding, abolished topA P1 induction (Fig. 8). Taken together these data indicate that the eight characterized Fis binding sites B to VI are sufficient to carry out the control of topA P1 induction in response to hydrogen peroxide exposure. Furthermore, sites IV to VI are necessary for both, co-operative binding and hydrogen peroxide-dependent P1 activation.

Figure 8.

Sites B-VI are sufficient and necessary for P1 activation. The plasmid pPtopA B-VI carries the full-length region of the topA promoter including sites B to VI, while pPtopA B-III carries a truncated promoter lacking the upstream sites IV to VI. topA mRNA levels were examined in cells upon subculturing (EE) and during late exponential phase (LE) prior to and after exposure to hydrogen peroxide (1 mM for 5 or 10 min). Primer extension assays were carried out using a primer specific to the plasmid-encoded topA promoter region.

Fis facilitates initiation complex formation at topA P1

To learn about the mechanism of topA P1 activation by Fis, we measured the extent of promoter opening in the presence of Fis using potassium permanganate (KMnO4) that preferentially targets pyrimidine residues in the untwisted region of DNA. KMnO4 footprinting was carried out using a plasmid that harbours the entire topA promoter region (pPtopA B-VI), in the presence of RNA polymerase, low concentrations of nucleoside triphosphates and different Fis concentrations. We found that addition of RNA polymerase alone had no effect on the reactivity of KMnO4, indicating that under the conditions examined, RNA polymerase could not bind the promoter region of topA (Fig. 9). Upon addition of a low concentration of Fis, an enhanced reactivity of KMnO4 within the regions of −10 and −35, at positions −10, −11, −12 and −31, −32 and −34 was observed. Also, bases +4, +1, −1, −2, −4 and −20 were reactive, suggesting that DNA unwinding had extended to the start site thus, forming a transcription bubble. Increasing the concentration of Fis abolished this enhancing effect. The results of the KMnO4 indicate that Fis facilitates initiation complex formation at the P1 promoter of topA.

Figure 9.

KMnO4 footprinting. Plasmid pPtopA B-VI (0.5 μg) extracted from cultures exposed to H2O2 at OD600 of 0.2 (1 mM for 10 min) was incubated (37°C for 30 min) with 0.5 mM NTP, purified Fis (10–50 nM) and 4 units of E. coli RNA polymerase (RNP). Thereafter, the samples were treated with 2 μl of 10 mM KMnO4 for 30 s as described in Experimental procedures. Primer extension to probe the modified bases in the template strand was carried out using 32P-end labelled primer 401A. The modifications were monitored on 8% sequencing gels. The modified residues are in bold. The sequences of −10 and −35 hexamers are boxed. The right brackets indicate the position of the hexamers. The numbering indicates the position upstream (–) or downstream (+) of the transcription start site of P1.

Discussion

Exposure of E. coli cells to hydrogen peroxide was previously shown by us to lead to topA P1 transcription activation in a Fis-dependent manner (Weinstein-Fischer et al., 2000). In this study, we show that induction of the hydrogen peroxide-dependent topA P1 promoter correlates with low levels of Fis. An increase in Fis levels due to subculturing, or to Fis overexpression from a multi-copy number plasmid, prevents this activation. In vivo DMS footprinting indicates that only at low Fis levels, when exposed to the stress, the region of the topA promoter changes and P1 becomes active. No changes in the pattern of DMS-induced methylation can be detected when P1 is exposed to the stress at high Fis levels or in fis-deficient cells, indicating that under conditions of high Fis or in its absence, the promoter remains inactive. In vitro KMnO4 footprinting shows that low levels of Fis facilitate the recruitment of RNAP to the P1 promoter but increased levels of Fis, abolish this enhancing effect. Thus, the concentration of Fis modulates the response of the topA P1 promoter to hydrogen peroxide.

Given the differential regulation of topA by Fis, we extended our characterization of Fis binding to the topA promoter and found that in vitro Fis binds this region in a cooperative manner at eight sites centred at −230/−256, −199, −157/−178, −129, −93, –62, −10/−30 and +20 relative to the transcription start site of P1 (sites VI, V, IV, III, II, I, A and B respectively). In vivo activity assays show that these Fis binding sites (B to VI) are sufficient to carry out the control of topA P1 induction in response to hydrogen peroxide exposure. Deleting the far downstream sites (A and B) had a marginal effect on the binding pattern of Fis to this promoter. In contrast, the deletion of the far-upstream sites (VI, V and IV) had a significant effect on both, the cooperative pattern of Fis-binding, and the ability of P1 promoter to respond to hydrogen peroxide. A similar type of functional organization of Fis sites was described for rrnA P1 (Rochman et al., 2004). A strong Fis binding site located far upstream of a cluster of Fis sites was found to facilitate the recruitment of additional Fis molecules in the nucleoprotein complex, thereby allowing the assembly of the transcription initiation complex at rrnA P1. Here, we find that the far upstream sites are necessary for both, the recruitment of additional Fis molecules to the topA promoter region and its activation.

Fis binds its sites at the topA promoter region with different affinities; the upstream sites III and VI exhibit the highest binding affinity to Fis, while site A that overlaps the −10, −35 hexamers of RNAP shows the lowest binding affinity to Fis. The location of site A that overlaps the RNAP recognition sequences and the weak binding affinity of Fis to this site, suggests that when Fis is expressed at high levels, it succeeds in occupying the entire range of sites along the promoter region including site A, and thus, directly interferes with the binding of RNA polymerase. A similar type of negative regulation was documented for Fis activity in several other cases. Fis modulates DNA supercoiling by repressing the transcription of the two subunits of gyrase, gyrA and gyrB. At gyrA, Fis interferes with polymerase binding preventing open complex formation, whereas at gyrB promoter, it prevents the isomerization of open complexes driven by nucleoside triphosphates, probably making polymerase escape less feasible (Schneider et al., 1999). Another example is the promoter of the nrfA operon that is activated by FNR and repressed by Fis and IHF. Fis represses PnrfA transcription by binding to a site that overlaps the −10 hexamer of RNAP, whereas IHF binds to a site that overlaps the site of the activator FNR (Browning et al., 2000; 2002; 2005). Fis and IHF each act at this promoter to repress the FNR-dependent open complex formation. The expression of acs encoding acetyl-coenzyme A synthetase is controlled by the regulator proteins CRP, Fis and IHF (Browning et al., 2004). CRP activates acs transcription by binding tandem sites located upstream of the major promoter, acsP2, whereas, both IHF and Fis each act at different times during growth as anti-activators of CRP. Fis binds the promoter region of acs at multiple sites; three high affinity sites, I, II and III are located upstream of the promoter (centred at −264, −98 and −59, respectively), and two low affinity sites positioned adjacent to Fis II and within the RNAP binding region. Thus, it was proposed that during exponential growth, Fis represses acs expression indirectly by repressing crp transcription and directly by binding to sites that overlap the binding sites of CRP and the −10 hexamer of RNAP.

While negative regulation by Fis has been characterized extensively only a few examples of activation were reported. As a positive regulator, Fis activates transcription from the ribosomal RNA promoters by contacting the C-terminal domain of the α-subunit of RNAP (Reviewed in Hirvonen et al., 2001; Aiyar et al., 2002; Schneider et al., 2003). At the promoter of tyrT, the wrapping of the upstream DNA around RNA polymerase helps it to overcome the melting barrier imposed by the GC-rich discriminator sequence within the promoter interval region (Muskhelishvili et al., 1997; Pemberton et al., 2002). Fis rescues the transcription of tyrosine tRNA1 at non-optimal super-helical densities by enhancing this effect. This DNA microloop that is formed by the upstream region and is stabilized by Fis, promotes the formation of a productive initiation complex (Auner et al., 2003). In Shigella and enteroinvasive E. coli, the proteins Fis and HN-S affect the expression of virF (Falconi et al., 2001). Fis binds to four sites at the promoter of virF of which two overlap an HN-S specific site that is responsible for the thermo-regulation. At permissive temperature, Fis exercises direct positive transcription control to induce virF expression. In Salmonella typhimurium, Fis is involved in the control of SPI-1 genes acting as a positive regulator of hilA (Wilson et al., 2001). The only example of dual regulation by Fis reported so far relates to HN-S expression. Fis was shown to bind the hns promoter region at seven sites (from −282 to +25) of which one overlaps the −10 hexamer of RNAP and one maps downstream to the transcription start site, centred at +25 (Falconi et al., 1996). In vitro assays demonstrated that low concentrations of Fis stimulated transcription of hns, while high Fis levels resulted in a decrease in HN-S expression. When both proteins, Fis and HN-S were present increasing amounts of Fis alleviated the inhibitory effect of HN-S leading to transcription stimulation (Falconi et al., 1996).

Here we present evidence for dual regulation of topA P1 by Fis in response to oxidative stress. The mechanism of this dual regulation is intriguing and not yet clear. Based on our data we suggest that when present at high levels, Fis cooperatively binds the entire promoter region including site A, at all conditions, preventing the binding of RNAP, under both normal growth and stress conditions. However, when present at low levels, under normal growth, Fis is ineffectively bound to parts of the promoter region of topA, but under conditions of oxidative stress, the binding pattern of Fis changes, leading to the recruitment of RNAP to P1. The change in the pattern of Fis binding could be due to changes in DNA topology or, due to auxiliary proteins present in the peroxide-treated late exponential cells that affect the binding of Fis to this region. An indication that the topology of this region changes under oxidative stress conditions is provided by the observation that the upstream P4 promoter becomes inactive under oxidative conditions, independent of Fis. Furthermore, our in vivo DMS data show that exposure of late log cells to hydrogen peroxide results in a change in the methylation protection pattern at the region of P1.

In the absence of hydrogen peroxide, using in vivo DMS assays, we could not detect changes in the methylation pattern of the core promoter region between early and late log phase cells. Either, the differential binding of Fis to this region could not be detected by this assay, or it suggests that under normal growth, the binding pattern of Fis, in early and late log phase cells, is similar, independent of Fis levels. However, the data of the P4 promoter showing that under normal growth P4 transcription correlates with Fis levels, corroborates our model that the initial binding of the topA promoter region by Fis does change with Fis levels.

Nutritional shift-up induces an increase in DNA supercoiling that is followed by relaxation (Balke and Gralla, 1987; Schneider et al., 2000). The initial effect is due, at least in part, to an increase in the ATP/ADP ratio and DNA gyrase activation, which in turn lead to an increase in Fis levels (Drlica, 1992). Previously, we have shown that topA activation by Fis is an important step in the response of bacteria to oxidative stress. Therefore, the inability of topA to respond to the stress upon subculturing in the presence of high Fis levels is puzzling. It is possible that an increase in the DNA supercoiling enhances the fitness of bacteria and/or that the binding of high Fis directly protects DNA against damage. A long-term experimental evolution study demonstrated that an increase in the super-helical density of the DNA increases the fitness of bacteria suggesting that cells upon subculturing might be more resistant (Crozat et al., 2005). This study also showed that mutations leading to minor changes in the level and the activity of Fis and TopA, respectively, are beneficial to bacteria. Here, we show that shifts in both Fis and topA levels that lead to changes in DNA architecture improve the fitness of E. coli towards oxidative stress, indicating once again, that these regulatory proteins play key roles in the response of bacteria to changes in the environment and stress conditions.

Experimental procedures

Strains, plasmids and media

Strains were routinely grown at 37°C in Luria–Bertani (LB) medium. Ampicillin (amp, 50–100 μg ml−1), or kanamycin (kan, 25 μg ml−1) were added where appropriate.

Plasmid construction

The construct pSA41 (pPtopA B-VI), plasmid pSA15 was digested with EcoRI and the site was filled in with DNA polymerase I large (Klenow) fragment. The plasmid was then digested with XbaI and the topA promoter fragment carrying Fis binding sites B to VI (499 bp) was subcloned into unique XhoI (filled in) and XbaI sites of pZE12 (Lutz and Bujard, 1997). To construct pSA42 (pPtopA B-III), a truncated topA promoter fragment (358 bp) was amplified from pSA15 by PCR using primers EcoRI 418 (5′-CGA ATT CAG TGT GAC GCT TTC GTC) and GO9 (5′-GGC GGA TCC CGG ATG TGA CCG ACG CT) and subcloned into unique EcoRI and BamHI sites of pGEM-3 (Promega). The fragment was then retrieved with EcoRI and XbaI and cloned into the corresponding sites of pZE12. The construct pPtac-fis lacI (pSA43) the fis gene was amplified from MC4100 chromosomal DNA by PCR using primers 763 (5′-CGA ATT CAT GTT CGA ACA ACG CGT) and 764 (5′-GCG TCG ACC CAT GCC GAG TAG CGC C). The 358 bp fragment was subcloned into unique EcoRI and SalI sites of pKK177-3-lacI (pSA10; Schlosser-Silverman et al., 2000).

Immunoblots

Bacterial overnight cultures diluted 1/100 in fresh LB medium were grown at 37°C to early and late exponential phase (OD600 of 0.05–0.6). Where indicated, 1 mM H2O2 was added for 10 min. The cultures were centrifuged and equal absorbance units were suspended in 2 × SDS loading buffer (30 mM Tris-HCl pH 7.0, 6% glycerol, 1.2% sodium dodecyl sulphate, 0.1% Bromophenol blue, 0.45 M β-mercaptoethanol). The proteins were then separated on 15% SDS-PAGE and transferred to nitrocellulose blotting membranes (Sartorius) by electroblotting. The membranes were probed with α-Fis antibody and incubated with secondary anti-rabbit IgG antibody conjugated to horseradish peroxidase (Enco Scientific Services). Detection was carried out with SuperSignal substrate (Pierce).

RNA analysis

Bacterial overnight cultures diluted 1/100 in fresh LB medium, were grown to early and late exponential phase (OD600 of 0.04–0.07 and ≥ 0.2, respectively) and then treated as indicated in the legends to the figures. For total RNA extraction, the cultures were harvested and the cell pellets were resuspended in 50 μl of 10 mM Tris-HCl pH 7.5 containing 1 mM EDTA. Lysozyme was added to 0.9 mg ml−1, and samples were subjected to three freeze-thaw cycles. Total RNA was extracted using Ultraspec-RNA (BIOTECX-Laboratory) according to the manufacturer's instructions except that 1 ml of TriPure reagent was used for 4 × 109 cells. The RNA samples (30 μg) were subjected to primer extension (at 42°C for 45 min) using AMV-RT (Promega) and end-labelled primers for chromosomal topA 401 (5′-CAT ATT CAC CTT ACC), plasmid-encoded topA 846 (5′-CGT TTT ATT TGA TGC CTC) and fis 650 (5′-AGG TCT GTC TGT AAT GCC AG). The extension products, together with sequencing reactions primed with the same end-labelled primer, were separated on a 6% sequencing gel.

DNase I footprinting

The region of topA promoter (392 bp) was PCR-amplified using 32P end-labelled top strand primer 686 (5′-CAA GGA CAT TAG TCT AC) and unlabelled bottom strand primer 687 (5′-CTC AAC GAT GAC AAG AG) or 32P end-labelled bottom strand primer 687 and unlabelled 686. The region carrying the sites I to VI (274 bp) was PCR-amplified using 32P end-labelled top strand primer 686 and unlabelled bottom strand primer 402 (5′-GCG GAT ATC AAC CCC T). The region carrying the sites B to III (271 bp) was PCR-amplified using 32P end-labelled top strand primer 418 (5′-AGG TAC AGT GTG ACG CT) and unlabelled bottom strand primer 687 or 32P end-labelled bottom strand primer 687 and unlabelled 418. The fragments were separated on 4% polyacrylamide gels (16 cm × 16 cm) and purified by incubation overnight at 37°C in 500 μl of 0.5 M NH4OAc and 1 mM EDTA pH 8.0. The DNA was subjected to chloroform extraction followed by ethanol precipitation in the presence of 2 μl of Quick Precip (EDGE Biochemicals) and resuspended in 15 μl of H2O. Footprint reactions were carried out at 25°C for 30 min using purified Fis and 1 μl of DNA in 1 × DNase I buffer (Promega-RQI carrying 40 mM Tris-HCl pH 8.0, 10 mM MgSO4, 1 mM CaCl2) supplemented with 100 mM NaCl, 1 mM DTT and 100 μg ml−1 BSA (here denoted Fis-binding buffer). Fis (kindly provided by Georgi Muskhelishvili) was diluted in protein dilution buffer (10 mM Tris-HCl pH 7.5, 500 mM NaCl, 50% glycerol and 0.1% triton X-100). The fragments were partially digested by DNase I (Promega) (0.2 U for 3 min). The reactions were stopped using 10 μl of stop solution. The results obtained with top and bottom strands are shown.

Electrophoretic mobility shift assay

The fragments carrying sites B to VI, I to VI, B to III and I to III were PCR-amplified using primers GO8-401, GO8-402, 418-401, 418-402 respectively. The fragments were labelled and purified as described in the section of DNase I footprinting. The binding reactions (total of 10 μl) were carried out at 25°C for 25 min using purified Fis and 1 μl of DNA in 1 × Fis-binding buffer (see above). The binding reactions supplemented with 7.5 μl of loading buffer (50% glycerol and 0.025% Bromophenol blue) were analysed on 4% polyacrylamide gels in 0.5 × TBE buffer (0.7 V cm−2 for 1.5–3 h).

In vivo DMS

Bacterial cultures carrying plasmid pPtopA B-VI were grown in LB at 37°C, to early and late exponential phase (OD600 of 0.04–0.07 and ≥ 0.2, respectively) and then treated with 1 mM hydrogen peroxide (Sigma) for 10 min. Thereafter, the cells were exposed to 10 mM DMS (Fluka) for 5 min and the plasmids were isolated using Qiagen mini spin columns. Pellets were resuspended in 100 μl of 1 M Piperidine (ALDRICH) and incubated at 90°C for 30 min. The plasmids were purified with G50-80 columns (Amersham Pharmacia Biotech). The DNA 10 μl was used as a template for 40 cycles of linear PCR reaction (1.5′ 94°C, 5′ 46°C, 2′ 72°C, 3′ 72°C) in a final volume of 35 μl, with 401 A end-labelled primer. PCR products together with sequencing reactions primed with the same end-labelled primer were separated on 8% sequencing gel.

KMnO4 footprinting

Plasmid pPtopA B-VI (0.5 μg) purified (gravity column, Qiagen) from cells grown to OD600 of 0.2 and treated with hydrogen peroxide (1 mM for 10 min) was incubated (37°C for 30 min) with 0.5 mM NTP, purified Fis (10–50 nM) and 4 units of E. coli RNA polymerase (freshly diluted in 1 × Fis binding buffer) in 25 μl of 1 × Fis binding buffer (40 mM Tris-HCl pH 8.0, 10 mM MgSO4, 1 mM CaCl2, 100 mM NaCl, 1 mM DTT and 100 μg ml−1 BSA). Thereafter, the samples were treated with 2 μl of 10 mM KMnO4 (Sigma) for 30 s. The reactions were then quenched with 2 μl of βME 14 M followed by two rounds of phenol-chloroform extraction. The DNA was then precipitated in ethanol and resuspended in 15 μl of water. Primer extension to probe the modified bases in the template strand (7.5 μl) was carried out using 32P-end labelled primer 401 A (5′-CCC ATA TTC ACC TTA CC). The extension products (5′-at 95°C, 1′ at 95°C, 1′ at 43°C, 1′ at 72°C, 3′ at 72°C; 10 cycles; 25 μl reaction volume) were precipitated with ethanol in the presence of 0.3 M NaOAc and resuspended in loading dye (98% deionized formamide, 2% 0.5 M EDTA pH 8.0, 0.0025 g Bromophenol blue and 0.0025 g Xylene cyanol-FF). The modifications were monitored on 8% sequencing gels.

Acknowledgements

We thank Georgi Muskhelishvili for purified Fis and Fis antibodies. We gratefully acknowledge Maya Weiss for assisting with the Western blot and Hilla Giladi for editorial comments. Supported by The Israel Science Foundation founded by The Israel Academy of Sciences and Humanities, grant number 663/02, The United States-Israel Binational Science Foundation, grant number 2001032, The Israel Science Foundation-Bikura Program, grant number 1342/05, The Israeli Ministry of Science, grant number 3/2559 and by BACRNAs, a Specific Targeted Research Project supported by European Union's FP6 Life Science, Genomics and Biotechnology for Health, LSHM-CT-2005-018618 (S.A.).

Ancillary