The closely related pathogenic Neisseria species N. meningitidis and N. gonorrhoeae are able to respire in the absence of oxygen, using nitrite as an alternative electron acceptor. aniA (copper-containing nitrite reductase) is tightly regulated by four transcriptional regulators: FNR (fumarate and nitrate reductase), NarP, FUR (Ferric uptake regulator) and NsrR. The four regulators control expression of aniA in N. meningitidis by binding to specific and distinct regions of the promoter. We show in the present study that FUR and NarP are both required for the induction of expression of aniA in N. meningitidis, and that they bind adjacent to one another in a non-co-operative manner. Activation via FUR/NarP is dependent on their topological arrangement relative to the RNA polymerase-binding site. Analysis of the sequence of the aniA promoters from multiple N. meningitidis and N. gonorrhoeae strains indicates that there are species-specific single nucleotide polymorphisms, in regions predicted to be important for regulator binding. These sequence differences alter both the in vitro DNA binding and the promoter activation in intact cells by key activators FNR (oxygen sensor) and NarP (which is activated by nitrite in N. meningitidis). The weak relative binding of FNR to the N. gonorrhoeae aniA promoter (compared to N. meningitidis) is compensated for by a higher affinity of the gonococcal aniA promoter for NarP. Despite containing nearly identical genes for catalysing and regulating denitrification, variations in the promoter for the aniA gene appear to have been selected to enable the two pathogens to tune differentially their responses to environmental variables during the aerobic–anaerobic switch.
- ferric uptake regulator (FUR)
- Neisseria gonorrhoea
- Neisseria meningitis
- transcriptional regulation
The genus Neisseria comprises Gram negative diplococcal species which generally inhabit the mucosal surfaces of epithelia, typically in humans or other animals. There are two pathogenic species, N. meningitidis and N. gonorrhoeae, that colonize only humans. N. gonorrhoeae, the cause of a major sexually transmitted infection, colonizes the mucosal surfaces in the male and female urogenital tract. N. meningitidis inhabits the nasopharyngeal mucosa and whilst it usually colonizes asymptomatically, it occasionally invades the bloodstream leading to septicaemia and/or meningitis. These pathogenic species and also the commensal Neisseria live in environments in which the availability of oxygen varies as a consequence of patchy perfusion of tissue by blood vessels and heterogeneous occupation by other oxygen-respiring micro-organisms that colonize these surfaces.
The absence of a consistent source of oxygen for respiration has led to the Neisseria species being adapted to use denitrification as an alternative respiratory pathway when oxygen is limited [1,2]. Different Neisseria species vary in their capacity for denitrification, but all species contain the genetic potential to reduce nitrite to nitrous oxide . Nitrite is reduced to the free radical gas nitric oxide (NO) via AniA (copper-containing nitrite reductase) . Utilization and detoxification of NO to nitrous oxide is catalysed by NorB (NO reductase) and is transcribed divergently upstream from the aniA gene . The control of expression of these denitrification genes has been studied in detail for the pathogenic Neisseria species, and involves multiple transcriptional regulators [5–10]. Work from our group has shown that expression of N. meningitidis nitrite reductase aniA is induced by FNR (fumarate and nitrate reductase) regulator in response to low oxygen [10,11]. The nitrite-sensitive two-component sensor regulator NarQ/NarP is required for nitrite-dependent induction of expression . The aniA gene in N. meningitidis is also repressed in an NO-dependent manner via the NsrR (NO-sensitive repressor) [6,9]. FUR (ferric uptake regulator) can also regulate aniA expression in N. meningitidis, where it appears to function as an activator . To date it has not been shown whether FUR is necessary for induction of aniA under conditions relevant to denitrification (i.e. limited oxygen availability, presence of nitrite). Regulation of aniA in N. gonorrhoeae follows a similar, but not identical, pattern to that seen in N. meningitidis. FNR and NarQ/NarP were identified as regulators of N. gonorrhoeae aniA . However, NarQ in N. gonorrhoeae appears to be insensitive to nitrite, remaining in a ‘locked-on’ conformation; thus, NarQ/NarP acts as a constitutive activator of aniA in N. gonorrhoeae . N. gonorrhoeae does not sense nitrite directly, but is able to induce aniA in response to NO (the product of AniA) via inhibition of the repressor NsrR . Isabella et al.  have purified N. gonorrhoeae NsrR and have shown it to bind to DNA in the absence, but not the presence, of NO. FUR has also been identified as a regulator of N. gonorrhoeae AniA . Putative binding sites for the four identified transcriptional regulators are shown in Figure 1. The FNR-binding site for N. meningitidis aniA promoter has been shown experimentally to bind FNR in vitro . N. gonorrhoeae NsrR has been shown to bind in vitro to the sequence predicted as the NsrR-binding site in the aniA promoter [13,15]. The FUR-binding region from the aniA promoter in N. meningitidis was determined experimentally by DNaseI footprinting . Two potential NarP-binding sites were strongly predicted on the basis of the NarP consensus sequence defined by Ravcheev et al. .
In the present study, we demonstrate the binding of FUR and NarP to the N. meningitidis aniA promoter in vitro, and show that these two regulators work as co-activators of aniA. Furthermore, we report that there are species-specific single nucleotide polymorphisms between the aniA promoter sequences in N. meningitidis and N. gonorrhoeae. These sequence differences affect binding of transcriptional activators (FNR and NarP) in vitro, which in turn affect the activation of gene expression in vivo. We argue that this allows N. meningitidis and N. gonorrhoeae to differentially tune their response to anoxia and other environmental cues optimally for their particular lifestyles.
Cloning of the fur and narP genes from N. meningitidis MC58 for overexpression
Escherichia coli DH5α cells were used as a general cloning vehicle. E. coli BL21 (DE3) cells (Novagen) were used to overexpress N. meningitidis FNR and FUR. E. coli Origami 2 cells (Novagen) was used to overexpress N. meningitidis NarP. Cultures were routinely grown in LB (Luria–Bertani) medium with kanamycin at a final concentration of 100 μg/ml.
Primers for the genes KEGG entry NMB0205 (fur) and NMB1250 (narP) were designed to incorporate ligation-independent cloning linker regions suitable for cloning into the plasmid pET-YSBLIC3C . Sequences of these primers can be found in Table 1. The resulting expression constructs (named pETFUR and pETNarP) were transformed into E. coli BL21 (DE3) and E. coli Origami 2 (for NarP only) cells before freezing in aliquots at −80°C.
Recombinant expression and purification of NarP
Recombinant expression of NarP was achieved with E. coli Origami 2 cells grown in 650 ml of autoinduction medium using a previously described method . Cells were grown overnight at 30°C at 200 rev./min until the attenuance of cultures reached a D600 of ~10 before being harvested at 3500 g in a Sorvall Evolution centrifuge and resuspended in Buffer A (50 mM Tris/HCl, 500 mM NaCl and 40 mM imidazole, pH 7.5). The cells were lysed using a Misonix 3000 sonicator for 2 min with a 10 s on/off cycle with amplitude 7 before centrifugation at 40000 g at 4°C to clarify the cell lysate before nickel-affinity purification.
Purification was performed at room temperature (20°C) using an Akta Prime (Amersham Pharmacia) semi-automated FPLC system running buffer at 2 ml/min through a His-Trap column charged with 0.1 M nickel sulphate. The column was equilibrated with 5 column vol. of buffer A before 30 ml of soluble cell lysate was applied and then washed with 5 column vol. of buffer B (as buffer A, but with 80 mM imidazole). His-tagged NarP was eluted from the column by applying a gradient of imidazole from 80 to 500 mM with the protein eluting at approximately 300 mM imidazole. Purity of samples was judged by SDS/PAGE (12% gel) and the fractions showing the greatest purity were pooled and desalted into 50 mM Tris/HCl (pH 7.5) and 350 mM NaCl using a PD-10 column (GE Life Sciences).
Recombinant expression and purification of FUR
Expression of FUR was achieved with E. coli BL21 (DE3) cells grown at 37°C with 120 rev./min shaking in 500 ml of LB medium. Expression of the protein was induced with the addition of IPTG (isopropyl β-D-thiogalactopyranoside) to a final concentration of 1 mM once the growth of the cells had reached a D600 of 0.4. Expression was allowed to continue for 3 h under the same conditions before harvesting the cells at 3500 g. The cell pellets were then resuspended in 30 ml of buffer A before being frozen overnight at −20°C. Frozen stocks were thawed at room temperature and lysed by sonication for 2 min with a 10 s on/off cycle with amplitude 7 before centrifugation at 40000 g at 4°C to clarify the cell lysate before nickel-affinity purification.
Purification was performed as for NarP, except that the protein was kept reduced by the inclusion of 2 mM DTT (dithiothreitol) in all of the buffers and in the soluble cell extract. His-tagged FUR eluted from the His-Trap column at approximately 250 mM imidazole. The purity of samples was judged by SDS/PAGE (12% gel) and the fractions showing the greatest purity were pooled and desalted into 50 mM Tris/HCl (pH 7.5), 350 mM NaCl and 2 mM DTT using a PD-10 column and kept on ice whilst the reconstitution took place. To reconstitute FUR with Mn2+, a stock solution of 10 ml of 50 mM Tris/HCl (pH 7.5), 350 mM NaCl and 2 mM DTT containing 100 mM MnCl2 was prepared and combined with FUR on ice to give a 15-fold molar excess of manganese over FUR. Once combined, a green/yellow colour was seen to develop that deepened over a 1 h period. FUR–Mn2+ was then desalted back into 50 mM Tris/HCl (pH 7.5), 350 mM NaCl and 2 mM DTT to remove excess MnCl2 and used immediately for DNA-binding reactions.
Recombinant expression and purification of FNR
Recombinant FNR was expressed and purified anaerobically from E. coli BL21 (DE3) cells according to a previously described method . Upon purification the protein was aliquoted into sealed glass bottles and stored at 4°C until ready for use.
Spectra were obtained using a Shimadzu 1601 spectrophotometer. Purified FNR samples were maintained under anaerobic conditions in cuvettes fitted with a silicone seal secured in place by a screw cap (Hellma). Spectra were obtained from 250 to 700 nm, or at fixed wavelengths over a time course.
Fluorescence anisotropy was used to measure binding of the purified proteins to complementary pairs of primers designed to encode predicted binding sites within the aniA promoter region (Table 1). All of the primers were ordered from MWG Eurofins to HPLC grade purity and resuspended in filtered dH2O to a final concentration of 100 pmol/μl (single stranded). One set of the complementary pair of primers was labelled at the 5′-end with HEX (hexachloro-6-carboxyfluorescein), foil wrapped and stored at −20°C. Samples were prepared either aerobically or anaerobically using sealed florescence cuvettes (Hellma) containing 50 mM Tris/HCl, 350 mM NaCl (pH 7.5) and 5 mM DTT, 2 nM annealed primers, 360 μg/ml acetylated BSA and 0.0026 μg/ml DiDC (polydeoxyinosinic-deoxycytidylic acid). Fluorescence anisotropy was measured at 25.4°C in a Fluoromax-4 Fluorimeter fitted with autopolarizers (Jobin Yvon Horiba) using an excitation wavelength of 530 nm and detecting emission at 555 nm. The slit widths used were typically 6 nm, but were adjusted per experiment to keep the signal close to 1×106 counts per s. NarP was titrated in to final concentrations up to 200 nM (monomers). FUR was titrated in aerobically to a final concentration of 960 nM (monomers). FNR was titrated in anaerobically to a final concentration of 300 nM (monomers). Following the addition of protein to DNA, samples were incubated for 5 min prior to measuring the fluorescence anisotropy. Fluorescence anisotropy was calculated in triplicate for each data point and for each experiment, a HEX-labelled control was also included. Data were plotted using SigmaPlot and curves fitted to a single rectangular hyperbola in order to calculate the Kd value.
EMSA (electrophoretic mobility gel-shift assay)
Gel-shift assays were performed to visualize a protein–DNA complex. Gels (10 ml) were poured containing 8% Bis-acrylamide (Bio-Rad Laboratories), 2.5% (v/v) glycerol and 2 ml of running buffer [50 mM Tris/HCl (pH 8.0) and 190 mM glycine]. The gels were pre-run for 1 h at 100 V. Once loaded, the gel was run at 100 V for 70 min at 20°C and the DNA subsequently visualized by staining with 5 μg of SybR-safe. The samples were prepared by mixing 10 μl of protein and 1 μl of duplex HEX-labelled DNA at a concentration of 50 μM and incubating for 5 min prior to loading on to the gel. The DNA used in the EMSA experiments were the same double-stranded DNA molecules used for the fluorescence anisotropy experiments.
Growth of N. meningitidis, production of recombinant strains and analysis of aniA promoter activity
N. meningitidis MC58  was used as the parental strain for all of the experiments using N. meningitidis. N. meningitidis strains deficient in narP and nsrR were described previously [9,10]. N. meningitidis deficient in fur was a gift from Professor Vincenzo Scarlato (University of Bologna, Bologna, Italy) . Double-mutant strains deficient in nsrR & fur, and nsrR & narP, were generated by transformation of chromosomal DNA from nsrR-deficient N. meningitidis into fur-deficient and narP-deficient N. meningitidis as described previously . N. gonorrhoeae MS11 was used as a template for amplification of the aniA promoter from that species. Neisseria strains were cultured on Columbia agar (Oxoid) supplemented with 5% horse blood (Oxoid) at 37°C in an atmosphere of 5% CO2. For the liquid cultures, N. meningitidis strains were grown in Mueller–Hinton broth supplemented with 10 mM NaHCO3 at 37°C. For aerobic growth, 5 ml of broth was shaken at 200 rev./min in 25 ml sterile plastic tubes (Sterilin). For microaerobic growth, 20 ml of broth was shaken in Sterilin tubes at 100 rev./min. For denitrifying conditions, microaerobic cultures were supplemented with 5 mM sodium nitrite. Nitrite was assayed as described previously . Cultures were supplemented with antibiotics (50 μg/ml spectinomycin, 2.5 μg/ml chloramphenicol and 100 μg/ml kanamycin) as appropriate.
The 384 bases upstream of the aniA translational start from N. meningitidis MC58 (and from N. gonorrhoeae MS11) were amplified with primers MFA7 and MFA8 as described previously . The product was cloned into pLES94  in order to generate a fusion of the aniA promoter with the lacZ (β-galactosidase) gene. Transformation of these vectors into N. meningitidis yields stable incorporation of the promoter lacZ fusion in single copy into the chromosome . Primers were designed to introduce site-directed mutations into the aniA promoter as described in Table 1 using the QuikChange method (Stratagene). In each case the primers were used to amplify the cloned aniA promoter by inverse PCR, thus introducing the desired changes to the promoter sequence. Products were checked by DNA sequencing. Recombinant plasmids were maintained in E. coli and transformed into N. meningitidis as described previously . β-Galactosidase activity was measured for N. meningitidis strains bearing the promoter–lacZ fusions as described previously .
Purification and characterization of FUR and NarP from N. meningitidis
The transcriptional regulators were purified following their heterologous overexpression in pET vectors in E. coli cells as described in the Experimental section. Some issues related to the protein purification are worthy of note here.
N. meningitidis FUR was highly expressed in E. coli and could be purified in a soluble form under aerobic conditions, but occurred in two isoforms as judged by SDS/PAGE. Despite appearing as two bands on the gels, a single peak of 16631±2 Da was detectable by MS, compared with a predicted mass of 16632 Da. The most probable explanation seems to be that the protein runs as two different redox forms; however, inclusion of reducing agents, DTT or 2-mercaptoethanol, had no effect. Aerobically purified FUR was not able bind to DNA. If purified anaerobically, the protein could be purified with an oxygen-sensitive cofactor present, although this protein was also not able to bind DNA. To achieve in vitro DNA binding, FUR was reconstituted aerobically by the addition of a 10-fold excess of MnCl2 to FUR in the presence of 2 mM DTT on ice. Over a 30-min period, the sample of purified FUR was seen to change colour which was retained even when the protein was desalted to remove excess MnCl2. The colour is due to an absorbance maximum at 440 nm as is typical of MN2+-containing proteins  (Supplementary Figure S1 at http://www.BiochemJ.org/bj/445/bj4450069add.htm). FUR reconstituted in this way ran as a single band on SDS/PAGE (Supplementary Figure S1).
NarP was purified aerobically and yielded large amounts of pure soluble protein. SDS/PAGE showed two bands of 24 kDa (NarP predicted mass of 23.8 kDa) and 48 kDa under non-reducing conditions, the larger band disappearing under reducing conditions, suggesting that NarP is able to form a covalent dimer due to a surface-exposed cysteine residue (Supplementary Figure S1). (Cys94 is the only cysteine in NarP, and is predicted to be surface-exposed on the basis of comparison with the structure of its homologue NarL [PDB code 1JE8] from E. coli ).
NarP and FUR bind independently to adjacent sites on the N. meningitidis aniA promoter
Both NarP and FUR have been implicated as positive regulators of aniA in N. meningitidis and N. gonorrhoeae. The predicted sites for the binding of the two proteins are adjacent to one another, further upstream from the aniA transcriptional start site than the site of FNR binding (Figure 1). In the present study we investigated the affinity of these two regulators for aniA promoter DNA, and any positive or negative co-operativity between the binding of the two proteins.
Quantitative analysis of DNA binding by FUR and NarP was carried out using fluorescence anisotropy. Figure 2 shows that FUR bound to a duplex DNA 70mer containing the predicted binding sites for FUR and NarP at Kd=174±74 nM, but showed no detectable binding to a negative control duplex DNA lacking the FUR consensus binding site. NarP bound to its predicted binding site with Kd=18±5 nM, but failed to bind to a negative control region lacking the predicted NarP consensus sequence sites. Sequential FUR and NarP binding to aniA promoter sequence containing both FUR- and NarP-binding sites was assessed. FUR binding followed by NarP binding gave Kd values of 174±74 nM and 22±4 nM respectively; and NarP binding followed by FUR binding gave Kd values of 18±5 nM and 190±65 nM respectively, indicating that the two proteins bind to DNA independently of one another.
As part of a two component sensor regulator pair, NarP is expected to require phosphorylation to activate it for DNA binding. However, freshly prepared NarP bound to its predicted binding site following aerobic purification, without the need for additional treatment to phosphorylate it. Presumably, NarP is fortuitously phosphorylated in E. coli and retains that phosphate group following purification. NarP was always used freshly on the day on which it was prepared, as older preparations lacked the ability to bind DNA. This is in keeping with observations that E. coli NarL has a half-life of dephosphorylation of a few hours , although dephosphorylation rates of this class of compounds can vary by several orders of magnitude . Treatment of purified NarP with calf intestinal phosphatase prevented DNA binding, supporting the suggestion that NarP is phosphorylated following heterologous expression in E. coli (Supplementary Figure S2 at http://www.BiochemJ.org/bj/445/bj4450069add.htm).
FUR and NarP are required for aniA expression under denitrifying conditions in wild-type and nsrR-deficient N. meningitidis
Mutants deficient in fur have been shown to have decreased expression of aniA messenger RNA indicating that this regulator is able to activate aniA expression . However, that work was carried out under aerobic conditions, under which aniA is not normally significantly induced. In order to test whether FUR is required for induction of nitrite reductase under the relevant microaerobic conditions suitable for denitrification, we analysed growth and nitrite utilization and aniA expression in N. meningitidis strains with single and double mutations in fur, narQP and nsrR. An N. meningitidis fur-deficient strain failed to utilize nitrite under microaerobic growth conditions in the presence of nitrite (Figure 3). Double mutants were constructed deficient in fur & nsrR and nsrR & narQP. Neither double mutant strain was able to remove nitrite after culturing under denitrifying conditions (Figure 3), providing further evidence that fur is required for aniA promoter activity under denitrifying conditions. The inability of fur-deficient strains to induce aniA expression was confirmed by analysing expression of a wild-type aniA promoter–lacZ fusion in wild-type N. meningitidis MC58 compared with the strains deficient in fur, narQP, nsrR, fur & nsrR, and nsrR & narQP. The absence of fur prevented N. meningitidis inducing aniA, even in the absence of nsrR (Figure 4) indicating that activation via FUR does not operate via repressing NsrR-dependent repression.
Spacing between NarP- and FUR-binding sites and FNR-binding site is important for activation of the aniA promoter
The influence of the topological arrangement of proteins on the promoter DNA on promoter activity was assessed by introducing extra bases in the region just upstream of the FNR-binding site. Insertions of 6 bases (CACATT) and 10 bases (CACACACATT) were made between bases 51 and 52 upstream from the transcriptional start (Figure 1). These recombinant aniA promoters were cloned upstream of a lacZ gene, and the promoter activity assessed by measuring β-galactosidase activity. Activity was measured in a wild-type N. meningitidis, and in a strain deficient in nsrR, since the site of insertion of extra bases may interfere with NsrR-dependent repression.
An introduction of 6 bases upstream of the FNR-binding site almost completely ablated aniA activation by microaerobic conditions and the presence of nitrite, even in the absence of the repressor NsrR (Figure 4). Lengthening the insertion to 10 bases (i.e. one full turn of B-DNA) led to a return of promoter activity and responsiveness to denitrifying conditions (even in the absence of NsrR), indicating that positively acting activators (FUR and NarP) that bind upstream of the FNR-binding site only work if the proteins are appropriately arranged topologically. The introduction of a 10-base spacer into the promoter sequence causes a loss of NsrR-dependent repression and a decrease in the overall extent of induction from aerobic to denitrifying conditions (from 50-fold to 6.5-fold). This may be due to the increased distance between FNR- and NsrR-binding sites (thus removing competition for an overlapping sites) or the loss of NsrR binding.
Sequence variation in aniA promoters and regulators from Neisseria species
Analysis of the aniA promoter from N. meningitidis and N. gonorrhoeae reveals that the promoter regions are highly conserved between the two species, with some notable exceptions. Particularly notable is that despite an overall identity of 93% between the promoter sequences from the two species the crucial FNR-binding sequence is not completely conserved between N. meningitidis and N. gonorrhoeae, suggesting some difference in the mode of control between the promoters from the different species. The polymorphisms are species-level sequence differences, as judged by the complete conservation within species of aniA promoter motifs among the 16 sequenced aniA promoter regions for N. gonorrhoeae and the 17 sequenced aniA promoter regions for N. meningitidis (based on BlastN analysis of all available microbial DNA sequences deposited in GenBank®). Single nucleotide polymorphisms were identified in the putative FNR-binding site (38 bases upstream from the transcriptional start site; C in N. meningitidis and T in N. gonorrhoeae), in the putative FUR-binding region (79 and 80 bases upstream from the transcriptional start; TA in N. meningitidis and GC in N. gonorrhoeae) and the putative NarP-binding region (97 bases upstream from the transcriptional start site; C in N. meningitidis and T in N. gonorrhoeae) (Figure 1). The aniA promoter sequences were compared with those predicted for other Neisseria species. Among the commensal Neisseria species, the FNR-binding site polymorphism was always the same as that found in N. meningitidis aniA promoter, and for the NarP-binding site the polymorphism was always as that found in the N. gonorrhoeae aniA promoter (Supplementary Figure S3 at http://www.BiochemJ.org/bj/445/bj4450069add.htm), indicating that both these differences represent specific polymorphisms in each of the pathogenic species from an ancestral common commensal sequence motif. The regulators themselves are highly conserved between N. meningitidis and N. gonorrhoeae. FNR, FUR and NarP are 97%, 98% and 99% identical between the two species respectively, and are 100% identical within the DNA-binding regions. The repressor NsrR in N. gonorrhoeae is 94% identical to that from N. meningitidis.
Differences in the aniA promoter FNR-recognition sequence between N. gonorrhoeae and N. meningitidis affect FNR binding in vitro, and promoter activity in vivo
FNR recognizes an inverted palindrome with the half-site consensus sequence being TTGAT/C. The guanine base is recognized by a glutamate residue within FNR . In N. gonorrhoeae, the aniA promoter contains a deviation from the typical FNR consensus sequence, such that it has the sequence TTAAT (rather than the consensus TTGAT found in N. meningitidis at this position). To test whether this sequence difference affects FNR binding in vitro, we made 40mer DNA duplexes centred around the FNR box from the wild-type N. meningitidis aniA promoter, and 40mers containing the mutation found in the FNR box in N. gonorrhoeae. N. meningitidis FNR was purified as described previously , and binding to these sequences was quantified using fluorescence anisotropy (Figure 5). FNR binding to the sequence containing the deviant FNR consensus sequence found in N. gonorrhoeae was weak, and barely distinguishable from binding to a sequence lacking an FNR consensus sequence.
To determine whether the N. gonorrhoeae aniA promoter is able to be activated in a N. meningitidis background, we constructed a N. gonorrhoeae aniA promoter lacZ fusion and inserted this into the N. meningitidis genome. This was compared with equivalent constructs with the N. meningitidis aniA promoter, and to a site-directed mutant version of the N. meningitidis aniA promoter in which the single base change in the FNR consensus site, as found in N. gonorrhoeae, was introduced. Although the wild-type N. gonorrhoeae aniA promoter displayed a similar level of induction under microaerobic conditions with the N. meningitidis wild-type aniA promoter, the promoter containing the single base substitution in the FNR site showed a low level of activity compared with either wild-type promoter (Figure 6). Nitrite led to a further activation of the N. meningitidis aniA promoter, but not the N. gonorrhoeae aniA promoter, when contained within the N. meningitidis cell.
N. gonorrhoeae aniA promoter binds NarP more tightly than does the equivalent region in N. meningitidis, offsetting the effect of weak FNR binding
The finding that the N. gonorrhoeae aniA promoter retained activity under denitrifying conditions, when contained in a N. meningitidis background, despite having a weak FNR-binding site, indicated that other mutations in the aniA promoter region offset the poor FNR-binding site and allow the promoter to be activated. We therefore constructed versions of the N. meningitidis aniA promoter containing both the mutation found in the N. gonorrhoeae FNR box, and each of the other polymorphisms, in the FUR- and the NarP-binding sequences. A single base change in the NarP-binding region (C→T at position −97) is sufficient to return activity to the N. meningitidis aniA promoter containing the N. gonorrhoeae FNR box sequence (Figure 6, iv), whereas mutating the sequence in the FUR-binding region (bases −79 and −80) of the N. meningitidis aniA promoter to the N. gonorrhoeae sequence has no noticeable effect (Figure 6, v). It is notable that the N. gonorrhoeae aniA promoter is unresponsive to nitrite when expressed in N. meningitidis, as is the N. meningitidis aniA promoter containing NM→NG mutations in both the FNR- and NarP-binding promoter sequences.
The sequence difference between the NarP-binding regions from the aniA promoters from N. gonorrhoeae and N. meningitidis alters the affinity for NarP (Figure 7). We constructed sets of duplex DNA on the basis of two strongly predicted NarP-binding heptamer repeats derived from the consensus NarP-binding sequence of E. coli  (Figure 1), using the promoter sequences from N. meningitidis and N. gonorrhoeae (Table 1). The two duplexes are referred to as NarP site 1 (the predicted NarP site furthest upstream from the aniA start) and NarP site 2. We used fluorescence anisotropy to assess binding of NarP to these DNA sequences. NarP binds to both NarP sites for both N. meningitidis and N. gonorrhoeae sequences, but the binding to site 2 is significantly tighter for the N. gonorrhoeae promoter than the N. meningitidis promoter (Kd=3.0±0.8 nM compared with 12.1±4.0 nM), consistent with the increased activity in vivo of aniA promoters containing the N. gonorrhoeae sequence at this position (Figure 6). NarP also binds three times more tightly to a 70mer DNA sequence (as used for the data presented in Figure 2) containing the sequence from N. gonorrhoeae compared with the sequence from N. meningitidis (results not shown). The absolute values of Kd measured vary between NarP preparations, consistent with the purified NarP being partially phosphorylated as prepared.
The pathogens N. meningitidis and N. gonorrhoeae are highly similar in terms of gene content and sequence identity, yet occupy different sites within the human body. For example, both organisms contain the same set of genes for transcriptional regulators, and these are typically greater than 96% identical at the amino acid level between the species. It has been argued that, in addition to gene content, it will be crucial to understand regulatory differences between such closely related organisms in order to understand their niche specializations [26,27]. An environmental cue of particular importance for microbes in the human body is oxygen availability. Neisseria species can respond to oxygen limitation by inducing the alternative respiratory pathway of denitrification, which is initiated by the nitrite reductase AniA. aniA expression is tightly regulated by at least four transcriptional regulators. In the present study we demonstrated that although both N. meningitidis and N. gonorrhoeae use essentially the same regulatory proteins to control this expression, single nucleotide polymorphic differences between the aniA promoters from the two pathogens (and from the commensal Neisseria) govern the response to anoxia and nitrite.
Mechanism of action of multiple regulators on the aniA promoter
The results of the present study add to our understanding of how multiple transcription factors are able to govern regulation of the aniA promoter in a number of ways. (i) We have demonstrated that FUR is required for induction of the aniA promoter under denitrifying conditions. (ii) FUR and NarP bind to adjacent sites, but there is no co-operativity between them. (iii) FUR and NarP are not repressors of the repressor NsrR (because double mutants deficient in fur & nsrR, or narQP & nsrR, are inactive). (iv) Activation via FUR and NarP is sensitive to the DNA topology. That NarP needs to be located on a particular side of the DNA in order to activate gene expression is consistent with it acting as a co-activator by interacting directly with RNA polymerase. This is different from the situation in E. coli where there is also co-dependence of the nitrite reductase (nir) promoter by FNR and NarP (or NarL). In E. coli, NarP (or NarL) acts by counteracting the repression by IHF (integration host factor) . NarL in E. coli is, however, also able to act as an activator directly , and we presume that in N. meningitidis NarP (or FUR) is also able to make productive contacts with RNA polymerase to either increase RNA polymerase binding affinity or stimulate the isomerization of RNA polymerase to the open complex. It is not yet clear how activation via FUR and NarP operates precisely, but both these proteins are required and involved in enabling the promoter to make an active conformation. It may be that FUR is a co-activator that allows NarP to achieve the correct conformation on DNA to allow it to act as an activator (or vice versa).
NarP recognizes two partially overlapping sequence regions containing pairs of inverted heptamers with tight affinity. It is worth noting that the binding affinities to these sites are not completely consistent with the NarP recognition sequence consensus for E. coli NarP, i.e. site 2 is more weakly predicted to bind NarP in N. gonorrhoeae than in N. meningitidis according to the position weight matrix of Ravcheev et al. , yet the data show that NarP in fact binds tighter to the N. gonorrhoeae sequence in this region. This presumably reflects minor differences in the DNA recognition in Neisseria NarP compared with that from E. coli.
Different adaptations for regulation of anaerobic respiration between Neisseria species
N. gonorrhoeae and N. meningitidis are strictly dependent on the anoxia-sensing regulator FNR for activation of the aniA promoter [7,10]. So, it was surprising to see that, despite a high level of general sequence conservation amongst these two organisms including within their aniA promoters, the FNR-binding consensus site was not conserved. In fact, the deviation from the consensus sequence found in the N. gonorrhoeae aniA promoter (TTGAT→TTAAT) (this is the reverse complement of the sequence shown in Figure 1, i.e. ATCAA→ATTAA) is particularly notable since the guanine residue of the FNR half-site has been shown to be the most important residue in providing specificity for FNR, being recognized directly by a key glutamate in the FNR helix-turn-helix motif . Mutating the N. meningitidis FNR-binding sequence to that of N. gonorrhoeae ablates FNR binding in vitro and decreases the anaerobic inducibility of aniA in vivo 10-fold. This loss of promoter activation can be complemented by introducing a mutation in the NarP-binding region of the aniA promoter, an effect that can be correlated with increased affinity for NarP, as determined by in vitro analysis of NarP–DNA interaction (Figure 7). To gain a more complete understanding of this relationship in the cell we would need to know the concentration of the active forms of these regulators in vivo, and crucially also their binding affinities to other competing DNA sites within the genome. However, from what we have observed it would appear that single nucleotide polymorphisms change the affinities of FNR and NarP for their cognate recognition sequences in the aniA promoter in vitro and that this leads to alterations in promoter activity in vivo.
The native N. gonorrhoeae promoter, and the N. meningitidis promoter with the sequence altered in the FNR- and NarP-binding sites to equivalence with N. gonorrhoeae, are oxygen sensitive, but nitrite insensitive. The nitrite insensitivity of the N. gonorrhoeae aniA promoter has been noted previously , but those authors interpreted this as being due to nitrite insensitivity of the NarQ sensor, rather than a cis-effect from the aniA promoter itself. Support for this interpretation was obtained from experiments in which N. gonorrhoeae NarQ was expressed in E. coli and found to be nitrite insensitive in a bioassay with a NarL-dependent repressible promoter . Whether or not N. gonorrhoeae NarQ is nitrite insensitive, the results of the present study offer an alternative explanation for the nitrite insensitivity of the N. gonorrhoeae aniA promoter. That is, cis-acting promoter factors are sufficient to render N. gonorrhoeae aniA expression nitrite insensitive. NarQP is nitrite sensitive in N. meningitidis [9,10], but it may be that for promoter sites with a very high affinity for NarP (such as the N. gonorrhoeae aniA promoter) there is sufficient phosphorylated NarP available to activate gene expression even in the absence of exogenous nitrite to activate the NarQ sensor kinase. This is supported by our finding that on expression in E. coli, N. meningitidis NarP is fortuitously phosphorylated and competent for DNA binding in the absence of its cognate sensor kinase.
Both FNR and NarP binding are required for aniA promoter activation in vivo in N. meningitidis. Presumably, even though FNR binding cannot be detected in vitro with the N. gonorrhoeae FNR-binding site, it can bind in vivo sufficiently well to induce expression anaerobically. It should be noted that the in vitro analysis of FNR–DNA interaction was conducted under a higher-than-physiological salt concentration (required for stabilizing the purified protein), and this may accentuate the relatively poor binding of FNR to the FNR-binding sequence in the N. gonorrhoeae promoter. It is notable that other experimentally determined FNR-dependent, anaerobically inducible, promoters in N. meningitidis  also lack well conserved FNR consensus sequences, indicating that, in vivo, deviation from the consensus is tolerated. Differences in regulation of nitrite reduction between N. meningitidis and N. gonorrhoeae need not be entirely due to differences in aniA promoter sequences. Although the regulators themselves are well conserved, and completely conserved in the DNA-binding regions, subtle differences between the sequences of the regulators may impact in trans on the regulation of aniA in Neisseria species. Also, the expression levels of these regulators may vary between species, thus altering target gene regulation. It is worth noting that although in the present study we show that NarP and FUR are required for activation of aniA in N. meningitidis, in N. gonorrhoeae deletion of the NarP- or FUR-binding site does not completely ablate activity of the aniA promoter in N. gonorrhoeae , indicating that there are further differences in regulation between these two closely related organisms.
In addition to the differences in the regulation of expression of nitrite reductase between N. meningitidis and N. gonorrhoeae, a number of other differences in anaerobic respiration have been identified between these pathogens. Several independent studies have found that there are evolutionary selection pressures acting against the maintenance of aniA and/or AniA activity in N. meningitidis. Numerous isolates of N. meningitidis contain pseudogenized aniA or lack the gene altogether [31–33], and all of the N. meningitidis strains lack a particular haem domain of the cytochrome oxidase that has been shown to be one of two pathways for electrons to AniA from the inner membrane in N. gonorrhoeae and all of the commensal Neisseria . The differences in the adaptations of N. gonorrhoeae and N. meningitidis to oxygen availability may be related to their underlying physiological niches. The urogenital tract occupied by N. gonorrhoeae is frequently anoxic , whereas N. meningitidis occupies a more aerobic environment and appears to be in the process of evolving to lose the ability to respire nitrite . All of the N. gonorrhoeae strains sequenced so far maintain aniA. The weak FNR-binding site of the aniA promoter does not prevent high expression levels of aniA anaerobically in N. gonorrhoeae. The deviation from FNR consensus in N. gonorrhoeae may have the role of preventing leaky expression of aniA under aerobic conditions. In N. gonorrhoeae, aniA (and norB) are up-regulated in biofilms, and play an important role in providing a respiratory route under anaerobic conditions in these structures [37,38]. For N. gonorrhoeae, the nitrite insensitivity (caused potentially by both the trans-acting ‘locked-on’ NarQP  and the cis-acting high-NarP-affinity site in the aniA promoter) may allow for tighter co-regulation of aniA and norB, since for both genes induction in response to denitrification substrates will be at the level of NO, via NsrR, rather than nitrite via NarQP . This in turn is likely to mean less accumulated NO during the aerobic–anaerobic transition, limiting the damage due to internally generated NO.
To conclude, the transition between aerobic and anaerobic respiration is tightly regulated by multiple transcriptional regulator proteins in the genus Neisseria. In the two pathogenic species, specific mutations in the aniA promoter region have become fixed in the genomes of these species compared with the other Neisseria species. This shows how even in the absence of different gene content, related organisms can be differentially adapted to their differing environmental conditions through point mutations affecting key regulatory processes.
James Edwards developed the protocols for protein purification, analysed protein–nucleic acid interactions, generated genetic constructs, analysed gene expression in vivo, analysed data and helped write the paper; Diana Quinn generated the genetic constructs and analysed gene expression in vivo; Karyn-Anne Rowbottom developed the protocols for protein purification; Jean Whittingham made the constructs for protein overexpression, initiated development of the protein purification protocols and analysed data; Melanie Thomson generated the genetic constructs, analysed gene expression in vivo and analysed protein–nucleic acid interactions; and James Moir developed the hypotheses, obtained funding, analysed data and wrote the paper.
This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC) [grant number BB/F000952/1 (to J.W.B.M.)]. M.J.T. was a recipient of a BBSRC quota studentship.
Abbreviations: AniA, copper-containing nitrite reductase; DTT, dithiothreitol; EMSA, electrophoretic mobility gel-shift assay; FNR, fumarate and nitrate reductase; HEX, hexachloro-6-carboxyfluorescein; lacZ, β-galactosidase; LB, Luria–Bertani; NorB, nitric oxide reductase; NsrR, NO-sensitive repressor
- © The Authors Journal compilation © 2012 Biochemical Society