The cytochrome c nitrite reductase NrfA is a 53 kDa pentahaem enzyme that crystallizes as a decahaem homodimer. NrfA catalyses the reduction of NO2− to NH4+ through a six electron reduction pathway that is of major physiological significance to the anaerobic metabolism of enteric and sulfate reducing bacteria. NrfA receives electrons from the 21 kDa pentahaem NrfB donor protein. This requires that redox complexes form between the NrfA and NrfB pentahaem cytochromes. The formation of these complexes can be monitored using a range of methodologies for studying protein–protein interactions, including dynamic light scattering, gel filtration, analytical ultracentrifugation and visible spectroscopy. These methods have been used to show that oxidized NrfA exists in dynamic monomer–dimer equilibrium with a Kd (dissociation constant) of 4 μM. Significantly, the monomeric and dimeric forms of NrfA are equally active for either the six electron reduction of NO2− or HSO3−. When mixed together, NrfA and NrfB exist in equilibrium with NrfAB, which is described by a Kd of 50 nM. Thus, since NrfA and NrfB are present in micromolar concentrations in the periplasmic compartment, it is likely that NrfB remains tightly associated with its NrfA redox partner under physiological conditions.
A model for the assembly and disassembly of decahaem and icosahaem complexes of NrfA and NrfB
The textbook description of electron transfer in respiratory systems using a c-type cytochrome commonly involves transient complexes formed by the monohaem cytochrome c of mitochondria that shuttles electrons between the cytochrome bc1 complex and cytochrome aa3 oxidase. Over the last 10 years in the bacterial world, however, it has been recognized that many of the important electron transfer reactions that underpin key biogeochemical cycles, such as that of nitrogen, involve electron exchange between complex multihaem c-type cytochromes. Examples include the electron exchange between the octahaem hydroxylamine oxidoreductase and the tetrahaem electron donor cytochrome c554 of Nitrosomonas species that is critical to bacterial nitrification (the conversion of ammonium to nitrite and nitrate) in the environment [1,2]. Another example is the interaction between a number of tetrahaem and decahaem cytochromes involved in electron storage and the reduction of mineral iron(III) and fumarate by Shewanella species [3–5]. Crystal structures have been determined for some of the multihaem cytochromes involved in these processes and this has yielded much insight into the arrangement of the haems within the proteins, possible routes of intermolecular electron transfer and the mechanisms of catalysis at the active sites [1–7]. However, it is clear that many of these multihaem c-type cytochromes must interact as homomeric or heteromeric complexes to enable intersubunit electron transfer to take place, ultimately driving catalysis and/or electron storage processes. By contrast with the large amount of literature dealing with the electron transfer complexes formed by the monohaem cytochrome c of mitochondria there have been few studies of the assembly and disassembly of the large electron storage/transfer complexes that can potentially form through association of multihaem cytochromes. In the present study, this is addressed for the pentahaem cytochrome c nitrite reductase NrfA from Escherichia coli and its redox partner, the pentahaem protein NrfB (Figure 1A).
Sites of protein–protein interactions in the NrfAB system
During the process of respiratory nitrate-dependent ammonification by enteric bacteria, such as E. coli, and sulfate-reducing bacteria such as Desulfovibrio desulfuricans, nitrate is reduced to nitrite by nitrate reductase, and the reduction of nitrite into ammonia is then catalysed by the cytochrome c nitrite reductase (NrfA). In addition to the six e− reduction of nitrite, NrfA can catalyse the five e− and two e− reduction of the potential intermediates NO and NH2OH respectively and is capable of the six e− reduction of sulfite to sulfide at higher rates than the membrane-bound sulfite reductases . A physiological detoxification role for the NrfA-catalysed reduction of NO has been suggested .
The crystal structures of NrfA have been solved for the enzyme from Sulfurospirillum deleyianum , E. coli , Wolinella succinogenes  and D. desulfuricans . NrfA crystallizes as a 106 kDa homodimer with a total of 10 closely packed haems (Figure 1A). Four of the haems in each monomer are bis-His-ligated, while the active site haem is ligated by a distal lysine ligand and, depending on the crystallization conditions, a hydroxide on the proximal side. A pair of α-helices from each NrfA monomer forms the dimer interface with a surface contact area of 1548 Å2 (1 Å=0.1 nm) in E. coli . The amino acid sequence of the interface helices is not highly conserved (Figure 1B) and may alter the stability of the dimer in different species. At one end of each interface helix pair, partially exposed to solvent, is Haem 5. The distance between the Haems 5 in the dimer is only 11.8 Å [Fe–Fe] and could potentially allow electrons to travel from one monomer to the other (Figure 1).
There are two genetically distinct systems by which electrons are transferred from the quinol pool to NrfA . Bacteria including W. succinogenes and S. deleyianum use System 1, where the electron donor to NrfA is the membrane-bound tetrahaem NrfH. NrfA forms a stable isolatable complex with NrfH  on the periplasmic membrane that has been shown to mediate e− transfer between menaquinol and NrfA . System 2 is used by enteric bacteria such as E. coli and Salmonella Typhimurium, which have three proteins, NrfB, NrfC and NrfD, implicated in electron transfer to NrfA . NrfB is a soluble 21 kDa pentahaem protein capable of electron transfer to NrfA . NrfC and NrfD are proposed to form a membrane-bound complex that can transfer electrons from the quinol pool to NrfB. The crystal structure of NrfB has recently been solved, showing it to be structurally distinct from NrfH . Complex formation between NrfA and NrfB would provide a method for electron transfer to NrfA from the quinol pool and allow NrfA to remain soluble in the periplasm. A putative protein-binding site on NrfA has been identified in the crystal structure near Haem 2 [7,18] (Figure 1A), where a conserved positively charged electrostatic patch has been identified in all crystallized NrfA structures. A four-residue loop is also conserved in NrfA from bacteria that also contain NrfB.
A key difficulty in studying protein–protein interactions between multihaem c-type cytochromes is that of the overlap of their UV–visible spectra, and fluorescence-quenching properties of the haems that can preclude measurements of interaction through monitoring changes in endogenous protein absorbance and fluorescence. Thus to engage in a study on the assembly of complexes formed by multihaem c-type cytochromes, it is advantageous to employ a combination of analytical methodologies. In the present paper, we illustrate the use of ultracentrifugation, dynamic light scattering, gel-filtration and UV–visible spectroscopy to explore the formation of a range of homologous and heterologous complexes between the pentahaem NrfA and NrfB cytochromes and relate this to possible physiological oligomeric states in the periplasmic compartment of E. coli. The formation of redox complexes from individual NrfA and NrfB monomers could occur through a number of different routes. A simple model allows for the assembly of either (NrfA)2, (NrfB)2 or NrfAB decahaem dimers or a (NrfAB)2 icosahaem complex (Figure 2A).
Study of a dynamic model of protein–protein interactions in the NrfA system
Characterization of the self-association of NrfA
Gel filtration is a commonly used and easily accessible method to study the self-association of proteins. Although gel filtration is not an effective way to accurately measure the molecular mass of proteins, owing to the effects of both size and shape on the retention volume, changes caused by either complex formation or large-scale structural changes can be observed by slight changes in the retention volume. This is illustrated when NrfA is passed through a gel-filtration column at increasing initial concentrations (3–250 μM). As the concentration of NrfA increases, the retention volume of the peak corresponding to NrfA increases (Table 1). This suggests that NrfA and a NrfA multimer, probably (NrfA)2, are in dynamic equilibrium. The self-association of air-oxidized NrfA can be assessed by measuring the hydrodynamic radius and polydispersity using DLS (dynamic light scattering) . The percentage polydispersity increases as the concentration of NrfA decreases, for example 29±11% at 38 μM NrfA and 91±11% at 5 μM NrfA . Since a low polydispersity value indicates a homogenous sample and a high value indicates a sample containing mixed species, this suggests that oxidized NrfA is a mixture of monomer and dimer at 5 μM and that as the concentration of NrfA increased, the proportion of dimeric NrfA increased. The Mw of the protein in solution can be calculated from the hydrodynamic radius measured by DLS and yielded a value of 102 kDa at NrfA concentrations ≥9 μM, which is similar to the calculated 106 kDa molecular mass of the (NrfA)2 complex. However, as the concentration of NrfA was decreased to below 9 μM, the Mw decreased to 62 kDa, close to the molecular mass of 53 kDa calculated for the NrfA monomer . Thus, taken together, the DLS data suggested that the 2(NrfA)↔(NrfA)2 assembly/disassembly takes place within the 5–9 μM concentration range.
|Sample||Concentration (μM)||Retention volume (ml)|
|Sample||Concentration (μM)||Retention volume (ml)|
SE (sedimentation equilibrium) experiments, using analytical ultracentrifugation, can complement DLS data to give information on oligomeric states . The log of a sedimentation gradient is directly dependent on the averaged molecular mass [Mw(av)] of a protein in solution. This is illustrated in Figure 2(B) where the gradient is shallower for a 2 μM solution of NrfA than for a 20 μM. The Mw(av) calculated from the data increases from 72 at 2 μM to 98 at 20 μM, in a similar manner to that observed in DLS experiments. This confirms that dissociation of the (NrfA)2 complex occurs over the 2–20 μM concentration range and, from the SE data, a KdAA of 4.0 μM for the dissociation of the (NrfA)2 decahaem complex into pentahaem monomers has been determined .
The spectroscopic and turnover properties of NrfA in the monomeric and dimeric states
In the (NrfA)2 structure, two Haems 5 come within an edge-to-edge distance of ~4 Å  and so in principle the absorbance spectra of these haems may be different in the monomeric and dimeric states. However, after normalization of spectra to account for the difference in concentration there are no significant observable differences over the concentration range 1–30 μM NrfA, where NrfA will shift from being predominantly monomeric to over 70% dimeric. This suggests that the environment of the NrfA haems is not significantly affected by (NrfA)2 complex formation. To investigate the possibility that the α-helices at the interface of the dimer axis are changed during complex association/dissociation, CD spectra between 190 and 300 nm were collected at NrfA concentrations in the 1–30 μM range. No significant changes were observed in the region at approx. 220 nm, indicating that no change in the content of α-helix was occurring between the monomeric and dimeric state. Thus, from the measurements, it appears that the structure of the NrfA does not differ significantly between the monomeric and dimeric forms.
The activity of the monomeric form of NrfA was measured using nitrite as a substrate and monitoring the rate of ammonium formation. The rate of nitrite reduction at saturating substrate concentrations (5 mM) remained constant within the concentration range 0.02–4 nM NrfA (502±69/s). This shows that the monomeric form of NrfA is capable of the six e− reduction of nitrite to ammonium, and that formation of the dimeric complex is not required for this reaction. The activity of NrfA measured over this nanomolar concentration range is similar to that measured previously . However, at higher concentrations, the rate of nitrite reduction during the time-course of the assay becomes too rapid to measure accurately. Fortunately, over the concentration range 2–15 μM, where NrfA changes from a predominantly monomeric to predominantly dimeric form, the activity towards an alternative substrate, sulfite , can be monitored instead. Measuring the rate of sulfide formation at different concentrations of NrfA in the presence of 0.5 mM sodium metabisulfite and 1 mM sodium dithionite gave a rate of 0.07±0.01/s and this rate remained constant over the NrfA concentration range 1–15 μM, indicating that the dimeric form of NrfA was not more effective than the monomeric form at reducing sulfite to sulfide. The low rate of NrfA activity in the presence of sulfite also allows the properties of NrfA to be studied under turnover conditions during the 3 h timescale required for a sedimentation velocity experiment. Determination of the molecular mass from the sedimentation and diffusion coefficients of 1 μM NrfA in the presence of sodium dithionite as electron donor and sulfite as electron acceptor indicated that NrfA was still predominantly monomeric under turnover conditions at 1 μM.
Characterization of the assembly of the heterodimeric decahaem (NrfAB) complex
Analysis of NrfB on a gel-filtration column reveals that the retention volume remains constant over the concentration range 3–70 μM (Table 1). Using DLS, it has also been demonstrated that the measured hydrodynamic radius remains approximately the same (2.5±0.1 nm) in the concentration range 10–45 μM . The percentage polydispersity was <30% in all cases, which indicated homogenous samples at the concentrations studied, for which an Mw of 27–30 kDa can be calculated . Sedimentation velocity experiments also indicated there was no change in the properties of NrfB in the range 1–9 μM and no change in the visible absorbance spectrum of NrfB was detected over this concentration range. Thus, taken together, this combination of techniques shows that there is no evidence of dimerization of NrfB over the concentration range 1–70 μM, indicating that the Kd of 2(NrfB)↔(NrfB)2 (KdBB, Figure 2A) is greater than 100 μM. E. coli NrfB contains only one tryptophan residue. This is conserved in NrfB from other bacteria, and it is positioned on the surface near Haem 5. The fluorescence of this tryptophan residue has been used to reveal changes in the vicinity of Haem 5 and suggested that the two proteins form a complex that leads to a decrease in the fluorescence of the NrfB tryptophan residue and indicated the formation of a 1:1 complex . When applied to a gel-filtration column at equimolar concentrations of 1 μM, NrfA and NrfB elute as separate peaks with retention volumes corresponding to those obtained when NrfA and NrfB were applied alone at this concentration (Table 1). However, at an equimolar NrfA and NrfB concentration of 133 μM, the proteins eluted together as a complex of NrfAB with a slower retention time than for the NrfA dimer complex (Table 1). These results show that, at higher concentrations, NrfA and NrfB are capable of forming a stable complex that survives passage through the gel-filtration column. UV–visible spectroscopy has been used to study this complex formation in more detail and indicated that NrfB forms a stable complex with NrfA with a stoichiometry of 1 NrfB monomer to 1 NrfA monomer . The results have been used to obtain a dissociation constant of 50 nM for the NrfAB complex (KdAB, Figure 2A) .
Characterization of the assembly of the heterotetrameric icosahaem (NrfAB)2 complex
SE experiments on solutions containing equal amounts of NrfA and NrfB at 20 μM show that the Mw(av) of the NrfA–NrfB sample was significantly greater than when compared with 20 μM NrfA only, as indicated from the steeper slope of the semi-log plot shown in Figure 2(B). This increase in molecular mass is consistent with the results obtained from gel filtration and the spectroscopically determined dissociation constant. If the NrfAB complex is assumed to be a stable complex with a Mw of 74 kDa then the association of the icosahaem (NrfAB)2 complex from the decahaem NrfAB complex can be quantified through determination of the equilibrium dissociation constant. SE profiles of the NrfAB complex over the concentration range 2–20 μM could be simultaneously fitted to a monomer–dimer equilibrium with an NrfAB heterodimer molecular mass of 74 kDa to yield a dissociation constant of 6.0 μM for 2(NrfAB)↔(NrfAB)2 equilibrium (KdABAB, Figure 2A). This is similar to the Kd of 4.0 μM obtained for the 2(NrfA)↔(NrfA)2 equilibrium (KdAA, Figure 2A). As KdAA, KdAB and KdABAB have been experimentally determined, it is then possible to calculate the value of KdAABB (Figure 2A) as 75 μM. Taken together the Kd values for the potential complexes formed between the pentahaem NrfA and NrfB proteins (Figure 2B) imply that formation of a decahaem NrfAB complex does not affect the dissociation of the NrfA at the (NrfA)2 dimer interface in the (NrfAB)2 complex. This is consistent with the view that NrfB binds at a site that is remote from the NrfA dimer interface .
The cytochrome c nitrite reductase is critical to ammonification in many enteric- and sulfate-reducing bacteria . The crystal structures of NrfA from four different bacterial species have been determined and in all cases show it to be a homodimeric decahaem complex [7,10–12]. Since each monomer binds five haems and the reduction of nitrite to ammonium consumes six e−, the possibility has been raised that electron transfer across the NrfA dimer interface might be required for catalysis. In this scenario, rapid six e− reduction of nitrite will take place by virtue of the presence of a pool of ten electrons in the decahaem homodimeric (NrfA)2 complex. In support of this view Stach et al.  observed a decrease in nitrite reductase activity when S. deleyianum NrfA was diluted to less than 3 nM and suggested that this might be due to dissociation of the dimer to yield inactive monomer. If correct this would have given a Kd~1 nM for this monomer–dimer equilibrium. We have shown, by assaying for ammonium formation, that the E. coli NrfA is fully active in the concentration range 0.02–5 nM and thus that the pentahaem NrfA monomer is catalytically competent for the six e− reduction of nitrite to ammonium. Our extensive examination of the NrfA protein from E. coli using a range of complementary techniques indicates a KdAA (Figure 2A) three orders of magnitude higher than that inferred for the S. deleyainum enzyme and places it in a range where the dissociation of the oxidized NrfA dimer can be envisaged to occur at the physiological concentrations of the enzyme measured in the bacterial periplasmic compartment. It is possible that the differences in KdAA inferred for S. deleyianum (NrfA)2 and measured directly for E. coli (NrfA)2 arise from differences in the interfacial helices (Figure 2A). However, since our work suggests that in E. coli the NrfA monomer is equally active as both monomer and dimer, it raises the possibility that factors other than dissociation may have contributed to the loss of activity of the S. deleyianum NrfA at very low concentration.
The observation that the KdAA [NrfA↔(NrfA)2] is similar to KdABAB [NrfAB↔(NrfAB)2] suggests that NrfB does not sterically hinder the formation of the NrfA dimer when it is pre-bound to NrfA. This then suggests that NrfB binds to NrfA at a site that is remote from the NrfA dimerization interface. The best candidate for such a binding site is the positively charged patch around the solvent exposed Haem 2 that has been previously postulated as an NrfB-binding site [7,18] (Figure 1A). The role of Haem 5 at the NrfA-NrfA interface remains unclear since the results have demonstrated that intramolecular electron transfer across this interface is not required for the six e− reduction of nitrite. The dissociation constant of the oxidized (NrfAB) complex (KdAB) of 50 nM is 100-fold lower that for (NrfA)2 or (NrfAB)2. Previously, the concentration of NrfA within the periplasm of E. coli strain JCB387 was determined as 13 pmol/mg of periplasm giving a periplasmic NrfA protein concentration of ~5 μM . This value is close to the dissociation constant of the (NrfA)2 and (NrfAB)2 dimers. Haem stained SDS/PAGE gels of periplasmic fractions of E. coli showed that NrfA and NrfB were present in approximately equal amounts . Thus, given that KdAB=50 nM, the predominant species in a periplasm containing NrfA and NrfB at a concentration of ~4 μM would be the heterodimeric NrfAB complex. This suggests that at physiological concentrations in the bacterial periplasmic compartment rapid electron transfer at the interface of a tightly bound decahaem NrfAB complex may be more important than electron transfer between NrfA molecules at the interface of the (NrfA)2 or (NrfAB)2 complexes.
Enzymology and Ecology of the Nitrogen Cycle: A Biochemical Society Focused Meeting held at University of Birmingham, U.K., 15–17 September 2010. Organized and Edited by Jeff Cole (University of Birmingham, U.K.), Rosa María Martínez-Espinosa (University of Alicante, Spain), David Richardson (University of East Anglia, Norwich, U.K.) and Nick Watmough (University of East Anglia, Norwich, U.K.).
This work was supported by the Biotechnology and Biological Sciences Research Council [grant number B18695] to D.J.R. and J.N.B., a Research Councils UK Fellowship to T.A.C., a Royal Society Wolfson Foundation Merit award to D.J.R. and a University of East Anglia Dean's Studentship to C.L.