Regulation of nrf operon expression in pathogenic enteric bacteria: sequence divergence reveals new regulatory complexity

Summary

The Escherichia coli K‐12 nrf operon encodes a periplasmic nitrite reductase, the expression of which is driven from a single promoter, pnrf. Expression from pnrf is activated by the FNR transcription factor in response to anaerobiosis and further increased in response to nitrite by the response regulator proteins, NarL and NarP. FNR‐dependent transcription is suppressed by the binding of two nucleoid associated proteins, IHF and Fis. As Fis levels increase in cells grown in rich medium, the positioning of its binding site, overlapping the promoter −10 element, ensures that pnrf is sharply repressed. Here, we investigate the expression of the nrf operon promoter from various pathogenic enteric bacteria. We show that pnrf from enterohaemorrhagic E. coli is more active than its K‐12 counterpart, exhibits substantial FNR‐independent activity and is insensitive to nutrient quality, due to an improved −10 element. We also demonstrate that the Salmonella enterica serovar Typhimurium core promoter is more active than previously thought, due to differences around the transcription start site, and that its expression is repressed by downstream sequences. We identify the CsrA RNA binding protein as being responsible for this, and show that CsrA differentially regulates the E. coli K‐12 and Salmonella nrf operons.

Introduction

The Escherichia coli K‐12 nrf operon encodes the NrfA periplasmic formate‐dependent nitrite reductase, which is responsible for reducing nitrite to ammonium ions to support bacterial growth under anaerobic conditions (Darwin et al., 1993). In addition to its role in anaerobic respiration, the NrfA nitrite reductase can also reduce the toxic molecule nitric oxide (NO) and contributes to the ability of E. coli and Salmonella enteric serovar Typhimurium to detoxify NO anaerobically (Poock et al., 2002; Gilberthorpe and Poole, 2008; Mills et al., 2008; van Wonderen et al., 2008). As enteric bacteria are exposed to both nitrite and NO, during their transition through the mammalian gastrointestinal track, this makes the NrfA nitrite reductase an important enzyme in the anaerobic environment of the gut.

Transcription of the E. coli K‐12 nrf operon is driven from a single promoter (pnrf) and expression is induced by the global transcription activator protein, FNR, in the absence of oxygen (Page et al., 1990; Tyson et al., 1994). FNR binding to a single DNA site, centred at position −41.5 (i.e., between positions −41 and −42 relative to the transcription start site, +1), is sufficient for maximal pnrf induction (Tyson et al., 1994; Browning et al., 2002). However, FNR‐dependent activation is suppressed by the binding of two nucleoid‐associated factors, IHF (integration host factor) and Fis (factor for inversion stimulation). The nrf promoter contains three DNA sites for IHF (IHF I to III) and three DNA sites for Fis (Fis I to III) (see Figs 1 and 2) (Browning et al., 2002; Browning et al., 2005; Browning et al., 2006). Binding of IHF to IHF I and Fis to Fis I both repress FNR‐dependent transcription, whilst the occupancy of IHF III has a stimulatory effect (Browning et al., 2002; Browning et al., 2005; Browning et al., 2006) (Fig. 1). The nrf promoter is also regulated in response to nitrite and nitrate ions by the two homologous response regulators, NarL and NarP (Tyson et al., 1994; Darwin et al., 1997; Wang and Gunsalus, 2000). Both NarL and NarP bind to the same site positioned at −74.5 and their association with pnrf displaces IHF from IHF I, resulting in nitrite‐dependent activation and maximal pnrf expression (Fig. 1) (Browning et al., 2002; Browning et al., 2005; Browning et al., 2006). In addition, expression from pnrf is repressed when cells are grown in rich medium (Page et al., 1990; Tyson et al., 1997). This repression is mediated by Fis binding to Fis I. As the cellular concentration of Fis surges under nutrient rich conditions, this results in greater occupancy of Fis I, shutting down pnrf expression irrespective of other environmental cues (Ball et al., 1992; Browning et al., 2005).

Figure 1.

Organization of the E. coli K‐12 pnrf53 promoter fragment.

The figure shows a schematic representation of the E. coli K‐12 pnrf53 promoter fragment and the important elements involved in its regulation. All numbering is in relation to the transcription start of pnrf (+1). The FNR and NarL/NarP binding sites are represented by inverted arrows, whilst the IHF and Fis binding sites are depicted by boxes. The central base pair of each DNA binding site is given, the transcription start site is indicated by an arrow and the location of the nrfA ATG initiation codon is shown. Expression from pnrf is completely dependent on FNR‐dependent activation, which is repressed (−ve) by IHF and Fis binding to IHF I and Fis I, respectively, and stimulated (+ve) by IHF binding to IHF III. NarL/NarP counteract the repressive effects of IHF, bound to IHF I, by displacing IHF from the promoter region. The location of the weak acsP1 promoter is indicated by an arrow (Browning et al., 2002, 2005).

Figure 2.

Alignment of the nrf promoter sequences from different enteric bacteria.

The figure shows the sequence of the E. coli K‐12 pnrf53 fragment from positions −209 to +131, aligned with the nrf promoter regions from EHEC, UPEC, EAEC, EPEC, S. flexneri (SFX) and S. enterica serovar Typhimurium (STM). The location of the transcription start site for pnrf53 is indicated by lower case text. The location of FNR and NarL/NarP binding sites are represented by inverted arrows, whilst IHF and Fis binding sites are depicted by boxes. The CsrA binding sequences and GGA motifs in the E. coli K‐12 and S. enterica serovar Typhimurium leader sequences are highlighted by grey boxes. The insertion of sequences within the upstream promoter region of S. enterica serovar Typhimurium promoter is indicated. Differences between the pnrf53 fragment and other promoters are highlighted in black. The extended −10 consensus sequence (TGnTATAAT) is aligned with the pnrf −10 promoter elements (Browning and Busby, 2016) and the location of the p + 102A and p + 104A substitutions, introduced into the pnrf53 and pnrf53 STM promoter fragments, respectively, is indicated by an arrow.

During their evolution, bacterial pathogens are exposed to the particular environmental conditions within their host organism. Over time, their genomes accumulate mutations, many of which will have no effect, whilst others can change protein function or gene regulation. To understand how this process has shaped the expression of the nrf operon, we have examined the nrf operon promoters from a number of different enteric pathogens (Fig. 2), particularly focusing on the enterohaemorrhagic E. coli (EHEC) and Salmonella enteric serovar Typhimurium promoters. Using this approach, we have uncovered differences in the regulatory strategies used at these promoters in these organisms and identify the global regulator, CsrA, as an additional regulator of this complex operon in enteric bacteria.

Results

Analysis of nrf operon promoters from pathogenic E. coli and Salmonella enterica strains

Previously, we generated the E. coli K‐12 pnrf53 promoter fragment, which carries the nrf promoter sequences from −209 upstream of the transcription start site (+1) to +131 downstream (Figs 1 and 2). This fragment contains all the necessary DNA sequence required for anaerobic and nitrite induction (Tyson et al., 1994). The alignment of pnrf53 DNA sequence with the corresponding nrf operon sequences from different enteric pathogens indicated that there are some base pair differences in the transcription factor binding sites, the core promoter regions and translational initiation signals at many of the promoters, with sequence differences being particularly extensive for the S. enterica serovar Typhimurium promoter (Fig. 2). As these differences could affect the expression of the nrf operon in these bacteria, we generated similar pnrf53 promoter fragments for EHEC, uropathogenic E. coli (UPEC), enteroaggregative E. coli (EAEC), enteropathogenic E. coli (EPEC), Shigella flexneri and S. enterica serovar Typhimurium (Fig. 2 and Supporting Information Table S1) (Browning et al., 2006). All fragments were cloned into the low copy number lac expression vector pRW50 (Lodge et al., 1992), to generate lacZ transcriptional fusions, and transformed into our wild‐type Δlac E. coli K‐12 strain, JCB387. The expression of β‐galactosidase in JCB387 cells, carrying each promoter, was then determined when cultures were grown in minimal medium aerobically, anaerobically and anaerobically in the presence of nitrite. Results in Fig. 3 show that expression from all pnrf derivatives was induced by anaerobiosis and further increased in the presence of nitrite. Most pnrf53 derivatives displayed similar expression levels to the E. coli K‐12 promoter, however, the EHEC promoter, pnrf53 EHEC, was more active anaerobically, whilst, the S. enterica serovar Typhimurium promoter (pnrf53 STM) was less active, as previously demonstrated (Browning et al., 2006; Browning et al., 2010). As the pnrf53 EHEC promoter possesses a single base pair difference in its −10 promoter element (TATACT) (Fig. 2), which improves its resemblance to the −10 consensus sequence (TATAAT) (Browning and Busby, 2016) and is a considerable distance from the FNR biding site, we examined whether expression was still completely dependent on FNR, by determining the promoter activity of constructs in the Δfnr strains, JRG1728 and JCB387 Δfnr. Results in Table 1 show that, whilst the majority of pnrf53 derivatives possessed little promoter activity in JRG1728, the pnrf53 EHEC promoter displayed considerable FNR‐independent activity in both JRG1728 and JCB387 Δfnr. In support of this Western blots, using anti‐NrfA antiserum and whole cell lysates from aerobically grown cells, detected considerably more NrfA protein in EHEC strain EDL933 than in the E. coli K‐12 strains RK4353 and MG1655 (Supporting Information Fig. S1). Thus, the improvement of the pnrf EHEC −10 element allows some NrfA expression to occur in the absence of FNR and in the presence of oxygen. Note that amino acid sequences of the EHEC FNR, Fis and various RNA polymerase subunits are either identical or extremely similar to those of E. coli K‐12 (Supporting Information Fig. S2).

Figure 3.

Expression of nrf promoters from different enteric bacteria in strain JBC387.

The figure shows the β‐galactosidase activities of wild‐type JCB387 cells carrying pRW50, containing pnrf53 promoter fragments from various enteric bacteria (see Fig. 2). Cells were grown aerobically and anaerobically in minimal salts medium and where indicated 2.5 mM sodium nitrite was added. β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass, each activity is the average of three independent determinations and standard deviations are shown.

Table 1.

Expression of nrf promoters from different enteric bacteria in JRG1728 and JCB387 Δfnr strains.

		β‐Galactosidase activity a
Promoter b	JCB387	JRG1728	JCB387 Δfnr
pnrf53	5119 ± 46	54 ± 2	36 ± 1
pnrf53 EHEC	8140 ± 177	412 ± 15	296 ± 13
pnrf53 UPEC	5200 ± 216	71 ± 3	nd c
pnrf53 EAEC	5568 ± 142	51 ± 2	nd
pnrf53 EPEC	5564 ± 72	55 ± 1	nd
pnrf53 SFX	5298 ± 112	65 ± 6	nd
pnrf53 STM	2435 ± 36	69 ± 1	55 ± 1

a. β‐galactosidase activities were measured in the JCB387 and two Δfnr stains, JRG1728 and JCB387 Δfnr, carrying pRW50 containing different pnrf53 fragments. Cells were grown anaerobically in minimal salts medium and β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass. Each activity is the average of three independent determinations and standard deviations are shown.

b. The first column lists the pnrf53 fragments used.

c. nd: not determined.

The EHEC pnrf promoter is insensitive to nutrient quality

The E. coli K‐12 nrf promoter is repressed in rich medium, being down‐regulated by Fis binding to Fis I (Page et al., 1990; Tyson et al., 1997; Browning et al., 2005). To determine if the pnrf EHEC promoter was similarly regulated, we examined its promoter activity in the narL narP strain JCB3884 in minimal and rich media. Note that strain JCB3884 was used in this experiment to remove any effects that NarL or NarP activation have on promoter activity. Results in Table 2 show that expression from the pnrf53 and pnrf53 STM promoter fragments was similarly repressed in rich medium growth conditions (5.8‐ and 6.7‐fold respectively). However, expression from the pnrf53 EHEC promoter fragment was relatively insensitive to medium composition, showing only a 2.2‐fold decrease. Thus, we conclude that improvement of the −10 element makes the EHEC promoter less sensitive to nutrient quality.

Table 2.

Repression of pnrf promoters from different enteric bacteria in rich medium.

	β‐Galactosidase activity a
Promoter b	Minimal medium	Rich medium	Ratio c
pnrf53	3254 ± 320	562 ± 19	5.8
pnrf53 STM	1121 ± 107	168 ± 5	6.7
pnrf53 EHEC	4766 ± 327	2146 ± 153	2.2

a. β‐galactosidase activities were measured in the narL narP strain JCB3884, carrying pRW50 containing different pnrf53 promoter fragments. Cells were grown anaerobically in either minimal salts medium or rich medium (Lennox broth plus 0.4% glucose). β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass, each activity is the average of three independent determinations and standard deviations are shown.

b. The first column lists the pnrf53 fragments used.

c. The ratio column indicates the fold repression for each promoter in rich medium.

Previously, we demonstrated that mutations that disrupt Fis binding to Fis I relieved repression under nutrient rich conditions (Browning et al., 2005). As this −10 element base change in the EHEC promoter lies just outside the Fis I binding site (Fig. 2), we used gel retardation assays to examine the binding of Fis to the E. coli K‐12 and EHEC pnrf97 promoter fragments, which both carry pnrf DNA sequences from −87 to +10 (Figs 4A and 5A). Results in Fig. 4A show that both pnrf97 fragments bound purified Fis similarly, with Fis I being occupied first and lower affinity Fis sites at higher concentrations, as previously observed in gel retardation and DNase I footprinting experiments (Browning et al., 2002, 2005). Thus, we conclude that differential Fis binding is not the reason why the EHEC nrf promoter can bypass repression under nutrient rich growth conditions.

Figure 4.

Gel retardation assays using pnrf97 promoter fragments. The figure shows gel retardation assays of pnrf97 fragments from E. coli K‐12 and EHEC, with purified Fis, FNR DA154 and RNA polymerase.

A. End‐labelled pnrf97 fragments were incubated with increasing concentrations of purified Fis protein: lanes 1–5, pnrf97 EcoRI‐HindIII fragment; lanes, 6–10, pnrf97 EHEC EcoRI‐HindIII fragment. The concentration of Fis protein in each reaction was: lanes 1 and 6, no protein; lanes 2 and 7, 50 nM; lanes 3 and 8, 100 nM; lanes 4 and 9, 150 nM; lanes 5 and 10, 200 nM.

B. End‐labelled pnrf97 fragments were incubated with purified FNR DA154 and increasing concentrations of purified RNA polymerase: lanes 1–8, pnrf97 EcoRI‐HindIII fragment; lanes 9–16, pnrf97 EHEC EcoRI‐HindIII fragment. The concentration of FNR protein in each reaction was: lanes 1, 6–8, 9 and 14–16, no protein; lanes 2–5 and 10–13, 2.7 µM. The concentration of RNA polymerase in each reaction was: lanes 1, 2, 9 and 10, no protein; lanes 3, 6, 11 and 14, 25 nM; lanes 4, 7, 12 and 15, 50 nM; lanes 5, 8, 13 and 16, 100 nM.

C. End‐labelled pnrf97 fragments were incubated with purified Fis, FNR DA154 and RNA polymerase: lanes 1–7, pnrf97 EcoRI‐HindIII fragment; lanes 8–14, pnrf97 EHEC EcoRI‐HindIII fragment. The concentration of Fis was: lanes 1, 3, 5, 7, 8, 10, 12 and 14, no protein; lanes 2, 4, 6, 9, 11 and 13, 200 nM. The concentration of RNA polymerase in each reaction was: lanes 1–4 and 8–11, no protein; lanes 5–7, 500 nM; lanes 12–14, 200 nM. The concentration of FNR protein in each reaction was: lanes 1, 2, 7–9 and 14, no protein; lanes 3–6 and 10–13, 2.7 µM.

Figure 5.

Expression of pnrf97 promoter fragments from different enteric bacteria.

A. The panel shows a schematic representation of the E. coli K‐12 pnrf53 and pnr97 promoter fragments. The location of FNR and NarL/NarP binding sites are represented by inverted arrows, whilst IHF and Fis binding sites are depicted by boxes.

B. The panel shows the β‐galactosidase activities of wild‐type JCB387 and JCB3884 (narL narP) cells carrying pRW224, containing pnrf97 promoter fragments (sequences −87 to +10) from E. coli K‐12, S. enteric serovar Typhimurium and EHEC (see Fig. 2). Cells were grown aerobically and anaerobically in minimal salts medium and where indicated 2.5 mM sodium nitrite was added.

C. The panel shows the β‐galactosidase activities of JCB3884 (narL narP) cells carrying pRW224, containing either the pnrf97 STM or the pnrf97 STM/p14C promoter fragments (sequences −87 to +10). The p14C mutation introduces a point mutation at position −14 in pnrf97 STM to disrupt the extended −10 promoter motif. Cells were grown aerobically and anaerobically in minimal salts medium. β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass, each activity is the average of three independent determinations and standard deviations are shown.

As it is likely that the improvement of the EHEC pnrf −10 element is responsible for this alteration in regulation, we examined the binding of purified FNR and RNA polymerase to each pnrf97 promoter fragment. Note that in this experiment FNR carries the DA154 substitution, which renders FNR active under aerobic conditions (Wing et al., 2000). As expected, FNR bound to each fragment, producing a single shifted species (Fig. 4B). The inclusion of 25 to 100 nM RNA polymerase resulted in super‐shifted RNA polymerase/FNR/DNA complexes for the pnrf97 EHEC fragment (Fig. 4B; lanes, 11 to 13). However, similar complexes were not detected at these concentrations of RNA polymerase for the E. coli K‐12 pnrf97 fragment (Fig. 4B) and were only observed at high concentrations of 500 nM (Fig. 4C; lane 5). Thus, as expected, RNA polymerase binds more strongly to the EHEC pnrf promoter in the presence of FNR. To investigate the effect of Fis on these complexes, we first pre‐incubated end‐labelled pnrf97 fragments with purified Fis and/or FNR before adding RNA polymerase. Note that for the E. coli K‐12 pnrf97 fragment RNA polymerase was used at a concentration of 500 nM, whilst for pnrf97 EHEC a concentration of 200 nM was used. Results detailed in Fig. 4C, confirm that Fis and FNR can bind simultaneously to each promoter fragment (lanes, 4 and 11) and that FNR facilitates the binding of RNA polymerase (lanes, 5 and 12). When the E. coli K‐12 pnrf97 promoter fragment was pre‐incubated with Fis and FNR and then challenged with RNA polymerase, we were able to detect super‐shifted species (lane, 6). However, the diffuse nature of these complexes suggests that they are unstable during electrophoresis and that Fis interferes with RNA polymerase binding. Conversely, for the pnrf97 EHEC fragment, stable RNA polymerase‐containing complexes were detected when both FNR and Fis were present in the reaction mix (lane, 13), indicating that Fis has less effect on the ability of RNA polymerase to bind to the EHEC promoter. Thus, we conclude that the EHEC pnrf promoter can bypass this repression as its improved −10 element allows RNA polymerase to out compete Fis when binding at the EHEC nrf promoter.

We note that for the E. coli K‐12 pnrf97 promoter fragment, RNA polymerase shifted species were observed in the absence of FNR (Fig. 4C: lane 7). The pnrf97 fragment contains the weak divergent acsP1 promoter (Fig. 1), which is totally repressed by FNR binding (Browning et al., 2002, 2005) and, therefore, the complexes observed are due to occupation of the acsP1promoter at the higher RNA polymerase concentrations used for this fragment. Note that in vitro transcription experiments confirmed that FNR supresses transcription from acsP1, whilst activating transcription from pnrf (Supporting Information Fig. S3), thus, corroborating our assignment of the RNA polymerase complexes observed in Fig. 4.

Sequences downstream of position +10 repress expression from the Salmonella nrf promoter

The S. enterica serovar Typhimurium nrf promoter is the least active of the promoters tested and previously we demonstrated that sequences upstream of position −87 do not influence expression as the Salmonella promoter lacks the stimulatory upstream IHF III site (Fig. 2) (Browning et al., 2006). During the course of this study, we cloned the smaller pnrf97 promoter fragments (−87 to +10) from E. coli K‐12 and EHEC into the low copy number lac expression vector pRW224, to generate lacZ transcriptional fusions (Fig. 5A). As a control we also generated the S. enterica serovar Typhimurium version of this construct, i.e., pnrf97 STM (Supporting Information Table S1). Constructs were transformed into strains JCB387 and JCB3884 (narL narP) and promoter activity was determined in cells grown in minimal medium. Surprisingly, results in Fig. 5B indicated that anaerobic expression from the pnrf97 STM transcriptional fusion was much higher than the E. coli K‐12 derivative, resembling that of pnrf97 EHEC. This suggests that the Salmonella promoter is stronger than previously thought and that expression from the longer pnrf53 STM fragment might be repressed by an additional factor, which binds downstream of position +10. Note that the introduction of the p14C mutation into the pnrf97 STM promoter fragment, which disrupts the extended −10 element of the promoter, completely abolished promoter activity, indicating that new promoter elements had not been generated during construction of this fragment (Fig. 5C).

Expression of the E. coli K‐12 and the Salmonella nrf operons are regulated by CsrA

CsrA is a sequence‐specific RNA binding protein, which directly represses the translation of many E. coli and Salmonella genes and can indirectly repress the transcription of others (Lawhon et al., 2003; Vakulskas et al., 2015). CsrA binds to the consensus sequence CAGGA(U/A/C)G within mRNAs, often found overlapping the ribosome binding sites of the genes it regulates (Liu et al., 1997; Vakulskas et al., 2015). Inspection of the S. enterica serovar Typhimurium nrfA sequence, around the ATG translation initiation codon, suggested that it might contain a CsrA binding sequence (Fig. 6A). To investigate this, point mutations, at positions +103 and +104, were introduced into the pnrf53 STM fragment to disrupt the important GGA motif of the CsrA binding site (Fig. 6A). DNA promoter fragments were cloned into pRW50 to generate lacZ transcriptional fusions and β‐galactosidase activity was then examined in JCB3884 (narL narP). Results in Fig. 6B show that disruption of the potential CsrA binding site elevated anaerobic expression ∼twofold, suggesting that CsrA might repress expression from the pnrf53 STM fragment. Note that amino acid sequence of the S. enterica serovar Typhimurium CsrA is identical to that of E. coli K‐12 (Supporting Information Fig. S2).

Figure 6.

Analysis of CsrA‐dependent regulation of nrf operon expression.

A. The panel shows the sequence of the S. enterica serovar Typhimurium pnrf53 STM promoter fragment from positions +84 to +133, aligned with the CsrA binding site consensus (Liu et al., 1997; Vakulskas et al., 2015). The location of the nrfA ATG initiation codon is underlined and sequence differences between pnrf53 STM and the E. coli K‐12 fragment are highlighted in black (see also Fig. 2).

B. The panel shows the β‐galactosidase activities of JCB3884 (narL narP) cells carrying pRW50, containing the pnrf53 STM promoter fragments harbouring substitutions at positions +103 and +104 (see panel A) cloned as lacZ transcriptional fusions.

C. The panel shows the β‐galactosidase activities of CF7789 and TRCF7789 (crsA) cells, carrying various pnrf53 and pnrf53 STM fragments cloned into pRW50 as lacZ transcriptional fusions. The p + 102A and p + 104A substitutions disrupt the potential CsrA binding sites in the pnrf53 and pnrf53 STM fragments respectively.

D. The panel shows the β‐galactosidase activities of CF7789 and TRCF7789 (crsA) cells, carrying pnrf53 and pnrf53 STM fragments cloned into pRW224 as lacZ translational fusions. In all experiments, cells were grown aerobically and anaerobically in minimal salts medium and β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass. Each activity is the average of three independent determinations and standard deviations are shown for all data points. In panels (C) and (D) the fold increase in β‐galactosidase activity, due to the p + 102A or p + 104A substitutions, is indicated in brackets.

To examine whether the pnrf53 STM fragment is regulated by CsrA, β‐galactosidase expression, from the pnrf53 STM and the pnrf53 STMp + 104A fragments (positions −246 to +133), cloned into pRW50, was examined in the Δlac strain CF7789 and in TRCF7789 in which csrA is disrupted (Romeo et al., 1993). As the sequence surrounding the E. coli K‐12 nrfA ribosome binding site is similar to that of Salmonella (Fig. 2) we also altered the GGA motif in the E. coli K‐12 promoter (i.e., the p + 102A substitution). Note that both the p + 104A and p + 102A substitutions maintain the arginine codon at this position in the nrfA mRNA when compared to the wild‐type sequence (i.e., AGG verses AGA). Results in Fig. 6C show that expression from pnrf53 STM p + 104 was elevated ∼twofold in CF7789, whilst no increase was observed in TRCF7789. This suggests that CsrA inhibits expression from the pnrf53 STM construct. Conversely, the expression from the E. coli pnrf53 p + 102A fragment was indistinguishable from that of the wild‐type pnrf53 fragment in the both strains, indicating that CsrA does not regulate expression from the E. coli K‐12 pnrf53 fragment. As CsrA predominantly influences translation (Vakulskas et al., 2015), we also cloned each pnrf53 derivative into pRW224 to generate translational fusions and β‐galactosidase expression was again determined in CF7789 and TRCF7789 (csrA). The disruption of the GGA motif in both the pnrf53 STM and the E. coli K‐12 pnrf53 fragments resulted in an increase of anaerobic expression in CF7789, which was absent in TRCF7789 (Fig. 6D). This indicates that CsrA represses expression from both the E. coli and Salmonella nrf constructs, when cloned as lacZ translational fusions, and suggests that CsrA regulates these two nrf operons differently.

To confirm that CsrA regulates both the E. coli K‐12 and Salmonella nrf operons we examined the effect that over‐expressing CsrA has on expression from the E. coli K‐12 pnrf53 and pnrf53 STM fragments. A C‐terminal 6His tagged version of csrA (csrA‐6his) (Dubey et al., 2005) was, therefore, cloned into the expression vector pQE60 NdeI to generate pQE60/csrA (Raghunathan et al., 2011). Induction analysis confirmed that CsrA‐6His could be detected in E. coli K‐12 JM109 (lacI^q Δlacz) cells carrying pQE60/csrA (Supporting Information Fig. S4). Therefore, JM109 cells, carrying either pQE60 NdeI or pQE60/csrA, were transformed with pRW244 carrying various pnrf53 and pnrf53 STM fragments, cloned as translational fusions. Cells were grown anaerobically in minimal salts medium, containing 0.4% glucose, and the effect of leaky uninduced CsrA‐6His expression was determined by measuring β‐galactosidase expression. Results in Table 3 show that CsrA caused a large decrease in expression from the pnrf53 and pnrf53 STM wild‐type fragments, when compared to cells carrying the empty pQE60 NdeI vector. However, for pnrf53 and pnrf53 STM fragments, carrying substitutions in the CsrA binding site (i.e., pnrf53 p + 102A and pnrf53 STM p + 104A respectively) the effect of CsrA expression was considerably less. This confirms that CsrA represses expression of both E. coli K‐12 and Salmonella nrfA using the CsrA binding sites identified by this study.

Table 3.

Repression of pnrf promoter derivatives by CsrA‐6His expression.

	β‐Galactosidase activity a
Promoter b	pQE60 NdeI	pQE60/csrA	Ratio c
pnrf53	516 ± 32	41 ± 3	12.6
pnrf53 p+102A	1314 ± 90	643 ± 84	2
pnrf53 STM	399 ± 14	17 ± 1	23.5
pnrf53 STM p+104A	2269 ± 43	338 ± 1	6.7

a. β‐galactosidase activities were measured in strain JM109 (lacI^q Δlac) carrying pRW224 containing different pnrf53 translational fusions and pQE60 derivatives. Cells were grown anaerobically in minimal salts medium supplemented with 0.4% glucose to control CsrA‐6His expression. β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass, each activity is the average of three independent determinations and standard deviations are shown.

b. The first column lists the pnrf53 fragments used.

c. The ratio column indicates the fold repression due to the leaky expression of CsrA‐6His.

Sequences surrounding the transcription start site of the Salmonella nrf promoter are responsible for its elevated promoter activity

Results in Fig. 5 indicate that the anaerobic expression from the Salmonella pnrf97 STM fragment is elevated in comparison to the E. coli K‐12 pnrf97 fragment. To identify which parts of the Salmonella promoter that were responsible for this, we generated chimeric pnrf97 fragments, in which the upstream and downstream sequences from pnrf97 STM were introduced into the E. coli K‐12 pnrf97 fragment (Fig. 7A). Fragments were cloned into pRW224 to generate lacZ transcriptional fusions, and β‐galactosidase activities were determined in JCB3884 (narL narP) cells, during aerobic and anaerobic growth in minimal medium. Results in Fig. 7B show that the downstream differences surrounding the transcription start site were predominantly responsible for the increased promoter activity. When individual differences were introduced into the E. coli K‐12 pnrf97 fragment (i.e., the p3A, p + 1A and p + 4T substitutions) only those at positions +1 and +4 led to small increases in expression, in comparison to pnrf97 (Fig. 7B). Thus, we conclude that the differences around the transcription start site are responsible for the higher promoter activity observed for the Salmonella pnrf97 fragment and that none of the differences alone are responsible for the elevation observed.

Figure 7.

Expression of chimeric pnrf97 promoter fragments.

A. The panel shows the sequence alignment of various wild‐type and chimeric pnrf97 fragments (positions −87 to +10). The position of the transcription start site for E. coli K‐12 pnrf97 is indicated by lower case text. The location of FNR and NarL/NarP binding sites are represented by inverted arrows, whilst IHF I and Fis I binding sites are depicted by boxes. Differences between the E. coli pnrf97 fragment and the pnrf97 STM derivatives are highlighted in black.

B. The panel shows the β‐galactosidase activities of JCB3884 (narL narP) cells carrying pRW224, containing various pnrf97 promoter fragments. Cells were grown aerobically and anaerobically in minimal salts medium. β‐galactosidase activities are expressed as nmol of ONPG hydrolysed min⁻¹ mg⁻¹ dry cell mass, each activity is the average of three independent determinations and standard deviations are indicated.

Discussion

Both the E. coli and Salmonella formate‐dependent Nrf nitrite reductases can support anaerobic growth on nitrite and detoxify NO, making them important for survival under the anoxic conditions experienced in the intestines of warm blooded animals (Pope and Cole, 1982; Poock et al., 2002; Lundberg et al., 2004; Gilberthorpe and Poole, 2008; Mills et al., 2008; van Wonderen et al., 2008). Here, we show that the regulation of the nrf promoters from various pathogenic bacteria is similar to that of E. coli K‐12, being induced by anaerobiosis and nitrite (Fig. 3). However, in spite of maintaining this general pattern of regulation, the core promoters of the EHEC and Salmonella promoters are more active than that of E. coli K‐12 (Fig. 5B). Our results also demonstrate that the EHEC nrf promoter displays considerable FNR‐independent activity (Table 1), and that the operon is expressed in EHEC in the presence of oxygen (Supporting Information Fig. S1). The EHEC promoter only differs from its E. coli K‐12 counterpart due to a single base pair improvement in the −10 element at position −7 (Fig. 2), which is an important base in the −10 hexamer (Shultzaberger et al., 2007; Browning and Busby, 2016). This difference increases the affinity of RNA polymerase for the EHEC promoter and enables RNA polymerase to out‐compete Fis when binding, decreasing the effect of this repression (Fig. 4: Table 2). Alignment of promoter sequences from 22 EHEC O157:H7 strains (Supporting Information Fig. S5A) indicated that this improvement was present in all strains examined, however, it was absent from the closely related E. coli O55:H7 and EHEC O157:H‐ strains (Supporting Information Fig. S5B) (Rump et al., 2011; Sadiq et al., 2014). This indicates that this alteration in pnrf appears to be specific to EHEC O157:H7. We suggest that, as EHEC O157:H7 is a commensal of cattle, and as cattle feeds often contain elevated nitrate levels, the altered expression patterns of nrf may facilitate growth in the presence of nitrite and NO, derived from metabolism in this niche (Lundberg et al., 2004; Callaway et al., 2009; Cockburn et al., 2013).

The core promoter of the Salmonella nrf promoter was also much stronger than that of E. coli K12, with anaerobic expression rivalling that of the EHEC promoter (Fig. 5B). Chimeric promoter constructs demonstrated that this was primarily due to sequence differences around +1, but that no specific difference was absolutely responsible (Fig. 7B). At bacteria promoters, the occurrence of an A as an initiating nucleotide is favoured over C and this difference could partly explain the strength of the Salmonella promoter (Jeong and Kang, 1994; Walker and Osuna, 2002; Vvedenskaya et al., 2015). Furthermore, during transcription initiation RNA polymerase also recognizes bases around the transcript start, i.e., the core recognition element, and, thus, differences in this region of the Salmonella promoter could influence recognition of this element and in turn affect promoter activity (Zhang et al., 2012; Bae et al., 2015; Vvedenskaya et al., 2016). Alignment of nrf promoter sequences from different S. enterica serovars (Supporting Information Fig. S6) indicated that many serovars possessed the same differences around +1 as S. enterica serovar Typhimurium, indicating that these improvements are largely conserved. However, two S. enterica serovars (Dublin and Newport) possess a G at position +4, which is the same as the E. coli K‐12 promoter at this position (Supporting Information Fig. S6). Thus, we would expect in these Salmonella serovars expression from pnrf to be lower.

Our data indicates that although the EHEC and S. enterica serovar Typhimurium core promoters have similar anaerobic expression levels (Fig. 5B) they achieve it using different mechanisms, highlighting the ability of bacteria to mix and match their promoter elements to achieve similar transcription outputs (Miroslavova and Busby, 2006; Hook‐Barnard and Hinton, 2007). The EHEC solution to increasing promoter strength, by altering the −10 element, decreases the dependency of the promoter on FNR and its sensitivity to repression during growth in rich media. For the Salmonella promoter, changing bases around the transcription start site increases promoter activity but maintains the overall regulation seen at the E. coli K‐12 promoter (Tables 1 and 2). Thus, increasing promoter strength to similar levels can result in very different patterns of regulation.

The RNA binding protein, CsrA, regulates the expression of many genes in E. coli K‐12 and S. enterica serovar Typhimurium, controlling diverse traits such as metabolism, biofilm formation, motility and virulence (Lawhon et al., 2003; Edwards et al., 2011; Vakulskas et al., 2015). Our results demonstrate that CsrA modulates anaerobic respiration in both E. coli and Salmonella by regulating expression of the formate‐dependent NrfA periplasmic nitrite reductase. Consistent with our results, pull down experiments with 6His tagged CsrA identified nrfA mRNA as a target for CsrA in E. coli (Edwards et al., 2011). The positioning of the CsrA binding site suggests that CsrA likely represses translation, perhaps by preventing ribosomal access to the nrfA ribosome binding site in both bacterial species (Figs 2 and 6D). However, it is clear from our work with transcriptional fusions that Salmonella nrfA expression is more complicated (Fig. 6C) and that CsrA may influence nrfA transcription elongation, perhaps by causing premature transcription termination, as has been observed for pgaA in E. coli, or alternatively it may affect transcript stability (Figueroa‐Bossi et al., 2014; Vakulskas et al., 2015). As CsrA exists as a homo‐dimer it is able to associate with two sites separated by up to 63 nucleotides (Mercante et al., 2009). It is of note that the nrfA untranslated leaders, in both bacteria, contain additional GGA motifs, which could serve as auxiliary low affinity CsrA targets (Fig. 2). This, coupled with any differences in mRNA secondary structure, could account for the apparent differential CsrA regulation observed between the E. coli and Salmonella constructs.

In E. coli, CsrA availability and activity is controlled by the small non‐coding RNAs (sRNA) CsrB and CsrC (CsrB/C), which both contain multiple CsrA binding motifs that sequester CsrA away from its target transcripts (Vakulskas et al., 2015). Production of CsrB/C sRNA is increased in response to the accumulation of metabolic carboxylic acids, e.g., formate and acetate (Suzuki et al., 2002; Vakulskas et al., 2015) and CsrB/C turnover is accelerated by the presence of glucose (Leng et al., 2016). Thus, it has been proposed that when a preferred carbon source is exhausted and metabolic carboxylic acids build up, CsrB/C levels increase and sequester CsrA to promote the switch from exponential to stationary phase (Leng et al., 2016). As CsrA regulates gene expression in response to nutrient quality, we propose that CsrA has been co‐opted at nrf to reinforce operon regulation in response to this signal. For example, during anaerobic growth in rich medium, nrf operon transcription is sharply inhibited by Fis (Browning et al., 2005). Under these conditions CsrB/C sRNA turnover should be increased and CsrA ‘freed’ to repress its targets, including nrfA. However, during growth in poor media nrf operon transcription in maximal as Fis levels are lower, particularly in stationary phase (Ball et al., 1992; Ali Azam et al., 1999). If this is coupled with the accumulation of carboxylic acids, we predict that increased CsrB/C sRNA production and decreased turnover, would sequester CsrA from the nrf transcript to allow its increased translation. As NrfA‐mediated nitrite reduction is also dependent on formate, it is possible that CsrA regulation is also a way of linking nrf operon expression to formate levels. Thus, we propose the use of CsrA and Fis together at nrf ensures that the translational and transcriptional regulation complement and reinforce one another.

It is clear that the regulation of the E. coli K‐12 nrf operon is complicated, with a total of six global regulators (i.e., FNR, NarL, NarP, IHF, Fis and now CsrA) coordinating expression in response to environmental and metabolic conditions (Browning et al., 2002, 2005, 2006). Much of this regulation is conserved between the different pathogenic bacteria studied here. However, it is clear that regulation of the EHEC and S. enterica serovar Typhimurium nrf operons have been fine‐tuned, presumably in response to the particular niches occupied by each species in the intestines of their host organisms. Although CsrA has been implicated in nrfA regulation, it does not account for all the repression observed for the Salmonella pnrf53 STM construct and at present the role of the long untranslated leader, which is conserved between species, is unclear. Thus, it is likely that additional regulatory mechanisms may operate to control expression of this complex and important operon in enteric bacteria.

Experimental procedures

Bacterial strains, growth conditions, plasmids and primers

The bacterial strains, plasmids and promoter fragments used in this work are listed in Supporting Information Table S1 and oligonucleotides are listed in Supporting Information Table S2. Standard methods for cloning and manipulating DNA fragments were used throughout (Sambrook and Russell, 2001). By convention, locations at the nrf promoter are labelled with the transcript start point designated as +1, and with upstream and downstream locations prefixed ‘−’ and ‘+’ respectively. Single base substitutions in pnrf are denoted pNX, where N is the position of the substitution relative to the transcript start and X is the substituted base in the non‐template strand of the promoter. For routine DNA manipulations and as a source of DNA fragments for gel retardation, fragments were cloned into plasmid pSR (Kolb et al., 1995) and for in vitro transcriptions, its derivative, pLSR was used (El‐Robh and Busby, 2002). To measure promoter activities, fragments were cloned into the lac expression vectors pRW50 and pRW224 (Lodge et al., 1992; Islam et al., 2011). Derivatives of pSR, pLSR and pQE60 were maintained in host cells using media supplemented with 100 μg ml⁻¹ ampicillin, whilst derivatives of pRW50 and pRW224 were maintained with 15 μg ml⁻¹ tetracycline. Cells were grown in either minimal medium (minimal salts with 0.4% glycerol, 10% Lennox broth, 40 mM fumarate) (Pope and Cole, 1982) or in Lennox broth (2% (w/v) peptone, (Merck), 1% (w/v) yeast extract (Fisher Scientific) and 170 mM NaCl) supplemented with 0.4% glucose (Squire et al., 2009). Where indicated, a final concentration of 2.5 mM sodium nitrite was added to cultures.

Promoter fragment and plasmid construction

The DNA sequences of the nrf promoters from pathogenic enteric bacteria were compiled from xBASE2 (http://xbase.warwick.ac.uk/) (Fig. 1) (Chaudhuri et al., 2008). The nrf promoter DNA from EHEC, UPEC, EAEC, EPEC, was amplified by PCR using the nrfA Up and nrfA Down primers (Supporting Information Table S2) with genomic DNA as template. The S. flexneri pnrf DNA was amplified using nrfSFX Up and nrfA Down primers. PCR products were restricted with EcoRI and BamHI and cloned into pRW50 to generate lacZ transcriptional fusions.

The wild‐type pnrf97 EHEC and pnrf97 STM fragments were generated by PCR, using the primer pairs nrfA E87/nrfO157 H10 and nrfSTM E87/nrfSTM H10, with the relevant genomic DNA as template. The chimeric pnrf97 STM up and pnrf STM down promoter fragments were also synthesized by PCR, using primer pairs nrfSTM E87/nrfA H10 and nrfA E87/nrfSTM H10 with pRW224/pnrf97 STM and pRW224/pnrf97, respectively, as template. The p3A, p + 1A and p + 4T substitutions were introduced into the E. coli K‐12 pnrf97 fragment by PCR using primers nrfSTM p3A, nrfSTM p + 1A and nrfSTM p + 4T with primer nrfA E87 and pRW224/pnrf97 as template. The extended −10 motif in the pnrf97 STM promoter fragment was disrupted by PCR amplifying the Salmonella promoter region using primers nrfSTM E87 and nrfSTM p14C with pRW224/pnrf97 STM as template. All pnrf97 promoter derivatives were restricted with EcoRI and HindIII and cloned into pRW224 to generate lacZ transcriptional fusions.

The pnrf53 and pnrf53 STM fragments, carrying substitutions in the CsrA binding site, were generated using megaprimer PCR (Sarkar and Sommer, 1990). In the first round of PCR primers nrfA p + 102A, nrfSTM p + 103, nrfSTM p + 104, were used in conjunction with primer D10527 and the relevant pRW50/pnrf53 construct. Each PCR product was then used with primer D10520 and the same template to generate the final PCR product, which was restricted with either EcoRI and HindIII for cloning into pRW50 to generate lacZ transcriptional fusions or EcoRI and BamHI for pRW224 to generate lacZ translational fusions.

To express CsrA in vivo, the DNA encoding a C‐terminal 6His tagged version of csrA (csrA‐6his) was excised from plasmid pCSB12 (Dubey et al., 2005) and cloned into vector pQE60 NdeI (Raghunathan et al., 2011), using NdeI and BamHI, to generate plasmid pQE60/csrA. Plasmid pQE60/csrA was maintained in E. coli K‐12 JM109 cells, grown in the presence of 0.4 to 1% glucose, to limit the leaky expression of CsrA.

Assays of nrf promoter activity

To assay the expression from pnrf derivatives cloned into the lac expression vectors pRW50 and pRW224, different host strains were transformed and β‐galactosidase activity was measured as described in our previous work (Jayaraman et al., 1987). Cells were grown in either minimal salts medium (Pope and Cole, 1982) or rich medium [Lennox broth supplemented with 0.4% glucose (Squire et al., 2009)]. Where indicated, a final concentration of 2.5 mM sodium nitrite was added to cultures. For aerobic growth, cells were shaken vigorously to an OD₆₅₀ of 0.2 to 0.3, whilst, for anaerobic growth, they were held static in growth tubes to an OD₆₅₀ of 0.4 to 0.6. β‐galactosidase activities are reported as nmol of ONPG hydrolysed in our assay conditions min⁻¹ mg⁻¹ dry cell mass and each activity is the average of three independent determinations.

Western blotting

To examine the expression of NrfA protein in E. coli K‐12 and EHEC strains, bacteria were either grown anaerobically and aerobically in minimal malts medium at 37°C. For anaerobic growth conditions, 50 ml of bacterial culture was grown without shaking to an OD₆₀₀ of 0.5 to 0.6, whilst for aerobic conditions 10 ml cultures were shaken vigorously to an OD₆₀₀ of 0.3 to 0.4. The preparation of normalized total cellular protein samples, their resolution by SDS‐PAGE gels and their Western blotting was carried out as detailed in our previous work (Browning et al., 2013). NrfA proteins were detected using anti‐NrfA antiserum raised in rabbit (kindly provided by Jeff Cole) and blots were developed using the ECL Western Blotting Detection System (GE Healthcare).

To examine the expression of CsrA‐6His protein, JM109 cells, carrying pQE60/csrA were grown aerobically in 10 ml of Lennox broth supplemented with glucose, where appropriate, until an OD₆₀₀ of ∼0.4. Protein expression was then induced by the addition of IPTG (Isopropyl β‐D‐1‐thiogalactopyranoside) to a final concentration of 1 mM for 3 hrs. Total cellular protein samples were prepared and Western blotting was carried out using anti‐6His (C‐terminal) HRP linked antibody (Invitrogen) (Browning et al., 2013).

Purified proteins

Escherichia coli RNA polymerase holoenzyme containing σ⁷⁰ was purchased from Epicentre Technologies (Madison, WI) and FNR DA154 and Fis proteins were purified as described previously (Wing et al., 2000; Grainger et al., 2008). Note that FNR carries the DA154 substitution, which renders FNR active under aerobic conditions (Wing et al., 2000).

Gel retardation assays

Gel retardation assays using purified FNR, Fis and RNA polymerase were carried out as detailed by Browning et al. (2008). Purified promoter fragments were end labelled with [γ‐³²P]‐ATP and approximately 0.5 ng of each fragment was incubated with varying amounts of each protein. The reaction buffer contained l0 mM potassium phosphate (pH 7.5), 100 mM potassium glutamate, 1 mM EDTA, 50 μM DTT, 5% glycerol and 25 μg ml⁻¹ herring sperm DNA. The final reaction volume was 10μl. FNR and Fis proteins were incubated at 37°C for 15 minutes, after which RNA polymerase was added and samples were incubated at 37°C for a further 15 minutes. Samples were loaded directly onto a running 6% polyacrylamide gel (12 V cm⁻¹), containing 2% glycerol and 0.25 × TBE and analysed using a Bio‐Rad Molecular Imager FX and Quantity One software (Bio‐Rad).

In vitro transcription assays

The pnrf97 promoter fragment was cloned into plasmid pLSR, such that the divergent pnrf and pascP1 promoters are both cloned upstream of a lambda oop transcription terminator (Supporting Information Fig. S3) (El‐Robh and Busby, 2002). Purified FNR D154A protein was then incubated with pLSR/pnrf97 plasmid (8 nM final concentration) at 37°C for 20 min in a reaction mixture containing 40 mM Tris‐Cl (pH 7.9), 10 mM MgCl₂, 50 mM KCl, 0.1 mM dithiothreitol (DTT), 0.2 μg of bovine serum albumin (BSA) μl⁻¹, 0.5 mM ATP, 0.5 mM CTP, 0.5 mM GTP, 0.05 mM UTP and 5 μCi of [α‐³²P]UTP, with a final reaction volume of 20 μl. Purified E. coli RNAP was then added to a final concentration of 50 nM, and the mixture was incubated at 37°C for a further 20 min, after which it was stopped by the addition of 25 μl of formamide buffer (95% [vol/vol] deionized formamide, 20 mM EDTA, 0.05% [wt/vol] bromophenol blue, 0.05% [wt/vol] xylene cyanol FF). Samples were then loaded onto a 5.5% denaturing polyacrylamide gel, containing 1 × TBE, and were analysed using a Bio‐Rad Molecular Imager FX and Quantity One software (Bio‐Rad).

Conflict of Interest

The authors declare no conflict of interest.

Author contributions

DFB and SJWB conceived and designed the research programme. REG, DJL and DFB performed the experiments. DFB and SJWB wrote the manuscript with input from all authors.

Supporting information

Supporting Information