OUP user menu

Look on the positive side! The orientation, identification and bioenergetics of ‘Archaeal’ membrane-bound nitrate reductases

Rosa Maria Martinez-Espinosa, Elizabeth J. Dridge, Maria J. Bonete, Julea N. Butt, Clive S. Butler, Frank Sargent, David J. Richardson
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00887.x 129-139 First published online: 1 November 2007


Many species of Bacteria and Archaea respire nitrate using a molybdenum-dependent membrane-bound respiratory system called Nar. Classically, the ‘Bacterial’ Nar system is oriented such that nitrate reduction takes place on the inside of this membrane. However, the active site subunit of the ‘Archaeal’ Nar systems has a twin arginine (‘RR’) motif, which is a suggestion of translocation to the outside of the cytoplasmic membrane. These ‘Archaeal’ type of nitrate reductases are part of a group of molybdoenzymes with an ‘RR’ motif that are predicted to have an aspartate ligand to the molybdenum ion. This group includes selenate reductases and possible sequence signatures are described that serve to distinguish the Nar nitrate reductases from the selenate reductases. The ‘RR’ sequences of nitrate reductases of Archaea and some that have recently emerged in Bacteria are also considered and it is concluded that there is good evidence for there being both Archaeal and Bacterial examples of Nar-type nitrate reductases with an active site on the outside of the cytoplasmic membrane. Finally, the bioenergetic consequences of nitrate reduction on the outside of the cytoplasmic membrane have been explored.

  • nitrate reductase
  • Archaea
  • nitrogen cycle
  • Rieske protein
  • molybdoenzyme
  • Q-cycle


Coupling the reduction of nitrate to energy-conserving electron transport pathways is a widespread means of sustaining growth and cell maintenance in anoxic environments (Richardson, 2000). In the process of nitrate respiration, the membrane-bound nitrate reductases of the γ proteobacterium Escherichia coli and the α proteobacterial Paracoccus species have for many years been paradigm enzymes for bioenergetic and biochemical elucidation of the process. These so-called Nar enzymes comprise three subunits: NarG, the ‘α-subunit’ of about 140 kDa, which contains the molybdenum bis molybdopterin guanine dinucleotide (Mo-bis-MGD) cofactor at its catalytic site and an [4Fe–4S] cluster; NarH, the β-subunit of about 60 kDa, which contains one [3Fe–4S] and three [4Fe–4S] clusters; and NarI, the integral membrane γ-subunit of about 25 kDa with five transmembrane helices that bind two haems b: one low potential (bL) located at the periplasmic side, and one high potential (bH) located at the cytoplasmic side (Bertero et al., 2003) (Fig. 1). NarG and NarH are located in the cytoplasm and associate with NarI at the membrane potential-negative cytoplasmic face (Δψ−) of the cytoplasmic membrane. Most published bioinformatic analyses of membrane-bound nitrate reductase sequences currently in the databases suggest that this arrangement is conserved among Gram-negative bacteria and indeed, for many years, it was assumed that this orientation of membrane-bound nitrate reductases would be conserved among prokaryotes in general. However, recent evidence suggests that this is not the case. Where respiratory nitrate reduction has been identified in Archaea, it is predicted to take place in a catalytic subunit that has a signal sequence that is characteristic of twin-arginine signal peptides, which serve to export folded redox proteins across the cytoplasmic membrane (Cabello et al., 2004). Examples of Archaea that probably have this type of nitrate reductase at the time of writing include Haloferax mediterranei (Lledo et al., 2004), Haloarcula marismortui (Yoshimatsu et al., 2002), Pyrobaculum aerophilum (Afshar et al., 2001) and Archaeoglobus fulgidus (Richardson et al., 2001; Dridge et al., 2006). In this minireview, the ‘RR’ sequences of nitrate reductases of Archaea and some that have recently emerged in Bacteria will be considered and it will be assessed whether they can be used to argue for a Δψ+ location of the active site subunit by also considering the genetic context of the narG gene and available biochemical evidence on the limited number of systems that have been characterized.

Figure 1

The different classes of prokaryotic Mo-bis-MGD nitrate reductases. Note that: (i) the Nap system illustrated is the rather simple system from Paracoccus pantotrophus (Berks et al., 1995), some Nap systems do not use a NapB as electron donor and that NapC may be substituted by a different quinol dehydrogenase (Jepson et al., 2006; Marietou et al., 2005) and (ii) the Nas enzymes characteristically have one [4Fe–4S] cluster, but some are predicted to bind an additional [2Fe–2S] cluster (Richardson et al., 2001).

The ‘Archaeal’ type of nitrate reductase is part of a group of Mo-bis-MGD enzymes with an ‘RR’ motif that are predicted to have an aspartate ligand in the molybdenum ion co-ordination sphere provided by the polypeptide chain. This group includes selenate reductase and there is a high level of sequence identity between enzymes designated selenate reductases in databases and enzymes of the Archaeal nitrate reductase group. Possible sequence signatures are sought that can aid in distinguishing a nitrate reductase from a selenate reductase with the aim of removing confusion over the annotation of these closely related enzymes when sequences emerge in those databases.

There is a high percentage of sequence similarity between the putative Tat-dependent Archaeal nitrate reductase NarG subunits and the cytoplasmically active NarG subunits of Gram-negative bacteria and thus they are often given the same gene nomenclature ‘narG’ in the literature. However, this can serve to obscure important bioenergetic differences in the two types of nitrate reductase system. Thus, the bioenergetic consequences of nitrate reduction on the Δψ+ of the cytoplasmic membrane and the possibility that a protonmotive Q-cycle might operate in at least one group of Archaeal Nar enzymes that would be bioenergetically equivalent to the Q-loop mechanism in operation in the paradigm bacterial Nar enzymes are explored.

Cellular location of the active site ‘Archaeal’ NarG

Archaeal membrane-bound nitrate reductase systems have attracted some interest because of the extreme conditions under which they can operate, for example the hyperthermophilic systems of Pyrobaculum aerophilum (Afshar et al., 2001) or halophilic systems of Haloferax mediterranei (Lledo et al., 2004) and Haloarcula marismortui (Yoshimatsu et al., 2000). Despite this, there are still only a few Archaeal respiratory nitrate reductase sequences available. In all cases, though, analysis of the N-terminal region of the nitrate reductases reveals the conservation of a twin arginine (‘RR’) motif that has similarities to a Tat signal peptide n-region consensus sequence: S/T-RR-X-FLK (Berks et al., 2000; Sargent, 2007). The Tat translocase system is a protein transport apparatus present in Bacteria and Archaea that can transport folded proteins across the cytoplasmic membrane. In the cases of the Archaeal Nar systems, the ‘RR’ motifs are followed by characteristic h- and c-regions and are strongly predicted by the authors' analysis, using the modelling programme TatP (Bendtsen et al., 2005), to indeed be signal peptides for protein export (Fig. 2). Is there biochemical evidence to support this assertion?

Figure 2

N-terminal Tat-like signal peptides of pNar and related enzymes. The ‘RR’ sequences or remnant ‘RR’ are underlined. The Tat consensus sequence is indicated at the top of the alignment.

Where data are available, cell fractionation studies with nitrate-reducing Archaea suggest that the NarG protein is strongly associated with the membrane fraction and requires detergent solublization to release it (Yoshimatsu et al., 2000, 2002; Afshar et al., 2001; Lledo et al., 2004). This raises the question of whether the subunit is located on the inside or outside of the cytoplasmic membrane. In nitrate-reducing Gram-negative bacteria, such as Paracoccus pantotrophus and E. coli, the widely accepted diagnostic method for assessing the cellular location of the active site of a nitrate reductase in intact cells has been to assay for activity with the nonphysiological electron donors methylviologen (MV) and benzylviologen (BV) (Bell et al., 1990) The membrane-permeant benzylviologen is a more effective electron donor to enzymes with their active site located on the inside of the cytoplasmic membrane than the relatively more membrane-impermeant methylviologen, despite the latter redox dye being more reducing. Thus, for example, in intact cells of E. coli (Yoshimatsu et al., 2000) expressing nar, an MV/BV activity ratio of 0.12 was measured, but a much higher ratio of 1.2 was measured for the halophilic Haloarcula marismortui by Yoshimatsu (2000) and this was confirmed in studies for the present minireview where a ratio of MV/BV-dependent Nar c. 3.0 has been measured in whole cells of halophilic Haloferax mediterranei Nar. These results are diagnostic for electron donation to the active site of an enzyme that is on the outside, rather than inside, of the cytoplasmic membrane (Table 1). Such experiments have not yet been reported for the other putative Archaeal Nars with Tat sequences thus far identified. However, the data show that where experimental evidence is available, there is compelling evidence that the active site of these Archaeal Nar systems is indeed on the outside of the cytoplasmic membrane. This gives confidence that this will be true of those that have not been experimentally tested but have predicted Tat signal sequences. However, in the absence of whole-cell MV/BV Nar activities, can other bioinformatic analysis be brought in to support the assertion that the ‘RR’ sequence directs a Nar for export?

View this table:
Table 1

Ratios of the methylviologen-dependent and benzylviologen-dependent nitrate and selenate reductase activities in intact cells

Species/enzymeMV : BV RatioPredicted locationReferences
Escherichia coli Nar0.12Δψ−Yoshimatsu (2000)
Haloarcula marismortui Nar1.2Δψ+Yoshimatsu (2000)
Haloferax mediterranei Nar3.0Δψ+This work
Enterobacter cloacae Nar0.5Δψ−Watts (2003)
Enterobacter cloacae Ser2.0Δψ+Watts (2003)
Thauera selantis Nar0.17Δψ−This work
Thauera selantis Ser10Δψ+This work

In Nar systems in which nitrate is reduced on the cytoplasmic (Δψ−) face of the membrane, the quinol is oxidized at the membrane potential-positive face (Δψ+) (Fig. 1). The consequence of this is that electrons have to be moved some 40 Å across the membrane from the site of quinol oxidation to the iron–sulphur clusters of the NarH subunit. This electron transfer is catalysed by NarI, which binds two b-haems that are stacked in a five helix bundle. By contrast, in a Nar system in which the active site subunit is at the Δψ+ side of the membrane, the electrons from quinol oxidation at this face do not need to pass back across the membrane. Thus, a di-haem NarI type of subunit is not needed. As the narI gene characteristically clusters with the other nar genes, then its absence in a nar cluster in conjuction with the NarG gene having an ‘RR’ motif will add weight to the assignment of such a subunit to the Δψ+ of the membrane.

An example of the worth of bringing the ‘RR’ and NarI analyses together comes when the Bacillus subtilis Nar system is considered (Fig. 2). In this case, the NarG subunit has an ‘RR’ motif, albeit a very poor one. However, analysis of the genetic context of the narG gene reveals that it is located in a classical narGHJI cluster with the presence of the narI gene (Hoffmann et al., 1995), suggesting that the NarG protein is located at the Δψ− face of the membrane. In fact, the remnants of a Tat signal sequence towards the N-terminus can be seen in many bacterial NarG proteins where the presence of a single R in the position of the ‘RR’ motif is quite common (Sargent, 2007). But can any Bacterial Nar systems be identified where the combined application of ‘RR’ and ‘NarI’ bioinformatics analysis strongly suggests an Archaeal-like Δψ+ orientation? The evolutionary question of ‘what came first, pNar or nNar?’ is not one that can be answered with any degree of certainty. However, there seems to be no biochemical reason why each should not be present in both Archaea and Bacteria. Evidence for pNar enzymes in Archaea has not been found (note, though, that the number of genome sequences is relatively small). However, searching Bacterial genome databases for pNar enzymes using the defining characters discussed above produced three hits that could be pNarG enzymes. This first is in Carboxydothermus hydrogenoformans Z-2901. This thermophilic (optimum growth temperature=78 °C), Gram-positive Firmicutes is a bacterial hydrogenogen that can grow anaerobically utilizing carbon monoxide as the sole carbon source and water as an electron acceptor, producing carbon dioxide and hydrogen as waste products. The organism has been shown to grow slowly under heterotrophic conditions with lactate as an electron donor and nitrate as an electron acceptor (Wu et al., 2005). The second hit is in Moorella thermoacetica ATCC 39073. This strict anaerobe is also a Gram-positive Firmicutes and a moderately thermophilic acetogen (optimum growth temperature=58 °C). A gene locus (ABC20208) is strongly predicted by the present analysis to be a pNarG and, accordingly, the organism has been reported to utilize nitrate as an electron acceptor under some conditions (Seifritz et al., 2002). The third example of a bacterial pNar can be found in Candidatus Kuenia stuttgartensis (Strous et al., 2006). This strictly anaerobic ammonia-oxidizing planktomycete has internal membranes, in addition to the cytoplasmic membrane, that compartmentalize anammoxosomes. It is known to be able to reduce nitrate and is strongly predicted to have a pNar, but it is not possible at present to predict which membrane pNar is transported across.

It seems certain that as more genome sequences emerge, other examples will arise of Archaeal-type Nar systems being present in Bacteria. Thus, rather than a kingdom-based subclassification of Nar systems, it is now proposed to adopt a location-based classification of nNar for a system in which the NarG subunit is located on the membrane potential-negative (Δψ−) side and pNarG for a system in which the NarG subunit is located on the membrane potential-positive (Δψ+) side (Fig. 1). It is noted that in all pNar systems, the NarH proteins encoded in the pnarG gene clusters do not have Tat signal peptides. This is also true of the iron–sulphur subunits of structurally defined nitrate-inducible formate dehydrogenase of E. coli where it is proposed that the iron–sulphur-containing β-subunit is exported as a passenger with the Tat-dependent Mo-bis-MGD-containing α subunit (Sargent, 2007). The same mechanism is thus likely to apply to the Nar systems so that a pNarH that will be associated will be coexported by Tat with a pNarG.

Is there a sequence signature that allows a putative pNar sequence to be assigned as a nitrate reductase?

The recent X-ray crystal structures of the nNarG subunit of E. coli have revealed that the Mo ion at the active site is co-ordinated by an aspartate residue, Asp222 (Bertero et al., 2003; Jormakka et al., 2004). This residue is conserved in all the pNarG subunits and so it is likely that it is also a Mo ligand in these enzymes (Fig. 3). However, some of the pNarG proteins [c. 900 amino acids (aa)] are considerably smaller than the nNar proteins (c. 1200 aa). In fact, they are very similar in size to the c. 900 aa catalytic subunits of selenate reductase (SerA) that are also Mo-bis-MGD enzymes. These molybdoenzymes also have Tat-like signal peptides (Fig. 2) and are located in the periplasm, on the membrane potential-positive side of the cytoplasmic membrane (Krafft et al., 2000). In addition, the Mo–Asp ligand is also conserved (Fig. 3); in fact, SerA is part of a group of enzymes (the D group of Mo-bis-MGD enzymes; Jormakka et al., 2004) in which this aspartate ligand is conserved (Fig. 3). This group also includes dimethylsulphide dehydrogenase (McDevitt et al., 2002), ethylbenzene dehydrogenase (Kniemeyer & Heider, 2001), chlorate reductase (Thorell et al., 2003) and perchlorate reductase (Bender et al., 2005) and all have Tat signal peptides (Fig. 2). This raises the question of when, from the sort of bioinformatic analyses described in the previous section, can one predict whether a putative pNarG is actually a nitrate reductase and not a selenate reductase? Can one be confidently distinguished from the other clearly on the basis of bioinformatics?

Figure 3

Signature sequences of pNar enzymes. Residues discussed in the text are underlined in bold. The residues that bind the predicted 4Fe4S cluster are indicated by an asterisk.

To illustrate the problem, initially, two enzymes will be considered on which the authors work in their laboratories: Haloferax mediterranei pNarG and Thauera selenatis SerA (Fig. 4). Haloferax mediterranei pNarG shows 31% identity and 63% similarity to T. selenatis SerA and the processed proteins are both of a similar size (c. 920 aa). It also shows 29% identity and 53% similarity to E. coli nNarG, but is c. 300 aa smaller. On this basis alone, one could not conclude that Haloferax mediterranei pNarG should be classified as a nitrate reductase, rather than a selenate reductase. In the light of this, for this minireview purified Haloferax mediterranei pNarG has been assessed for nitrate and selenate reductase activity and it has been established that the enzyme is catalytically similar to nNarGs in that it is highly reactive towards nitrate and chlorate as substrates, but shows no reactivity towards selenate (Fig. 4). Likewise, the activity of T. selenatis SerA has been analysed and it has been established that it is highly active towards selenate, but essentially inactive towards nitrate (Fig. 4). There have been some suggestions in the literature that nNar reduces selenate, for example NarG and NarZ have been reported to confer a low selenate reductase activity to E. coli membranes (Avazeri et al., 1997). However, these experiments were carried out at comparatively high selenate concentrations of 160 mM and significant selenate reduction by Paracoccus (Fig. 4; Watts et al., 2003) or Enterobacter cloacae pNar has not been detected (Watts et al., 2005; Ridley et al., 2006). pNar, nNar and Ser, though, do have the common ability to reduce chlorate at high rates and this can serve as a reference point for normalizing activities across different types of enzymes (Fig. 4).

Figure 4

Methylviologen-dependent selenate, chlorate and nitrate reductase activities of (a) Haloferax meditteranei pNar, (b) Thauera selenatis Ser and (c) Paracoccus denitrificans nNar. All activities were determined at 25 mM substrate. The activities are presented as the percentage of the maximum rate detected with chlorate. Methylviologen activities were measured as described in Anderson (2001). All data have been newly collected for this work. Haloferax meditteranei pNar 100% activity=250 nmol MV oxidized mg−1 min−1; Thauera selenatis; Ser 100% activity=500 nmol MV oxidized mg−1 min−1; Paracoccus denitrificans nNar 1000 nmol MV oxidized mg−1 min−1.

Thus, the biochemical analysis provides clear information on a Nar or Ser enzyme's catalytic character that the overall sequence identities are ambiguous about. With this information, one can then look further into the pNarG, nNarG and SerA sequences using the structurally defined E. coli nNarG as a platform from which to identify pNars correctly in emerging genome databases. Using molecular modelling (Dridge et al., 2006) based on the three-dimensional X-ray structure of nNarG from E. coli (Bertero et al., 2003; Jormakka et al., 2004) and also examining the recent X-ray structure of Aromatoleum aromaticum EbdA (Kloer et al., 2006) and the amino acid sequences of other D-group Mo-bis-MGD enzymes, a number of key signatures of a pNarG begin to emerge. The first is Asn52, which is located within a cysteine-rich motif H/CX3CX3−5CX35−46C located close to the N-terminal of all the D-group molybdoenzymes. This Cys cluster is thought to be involved in the co-ordination of an iron–sulphur cluster of the [4Fe-4S] type (Fig. 3a). The involvement of this motif in the co-ordination of such a cluster has been shown in the crystal structures of E. coli nNarG (Bertero et al., 2003; Jormakka et al., 2004) and Aromatoleum aromaticum EbdA (Kloer et al., 2006). Among all the nNarGs, the first cluster-co-ordinating residue is His, but this is not the case in the putative pNarG proteins where it can be a Cys residue (Archaeoglobus fulgidus). Asn52 is located in the cluster (Fig. 3a). It is positioned c. 3.9 Å away from the Mo atom and serves a structural role, positioning the [4Fe–4S] cluster and could form a hydrogen bond to a bound substrate. This asparagine residue is also conserved in the pNarGs, but is replaced with glycine in the active site of EbdA and SerA (Fig. 3a), which could affect enzyme specificity. The second signature relates to the substrate entry channel. In the E. coli nNarG, the conformation of the putative substrate entry channel is dictated by Thr54, Gln235 and Thr236. Gln235 and Thr236 are conserved in the Haloferax mediterranei pNarG sequence and, indeed, all the nNar and putative pNar sequences. This could therefore be a fingerprint for a Nar as they are not conserved in the other D-group members (Fig. 3b), for example they are replaced by Ala221 and Arg222 in T. selenatis SerA. This structure-based bioinformatics analyses suggests that there are signatures that may enable a pNar to be distinguished from a SerA among the D-group branch of molybdoenzymes.

Maintaining the bioenergetic equivalence of nitrate reduction on the membrane potential-positive and -negative sides of the membrane

During respiratory electron transfer, Nar enzymes receive electrons from quinols located within the lipid phase of the cytoplasmic membrane. As discussed earlier, in the case of the nNar system the oxidation of quinol takes place at the periplasmic side of NarI where the protons are released and the two electrons are moved from the periplasmic-face haem bL to cytoplasmic-face haem bH. This charge separation makes the enzyme electrogenic in that it contributes to the generation of a proton electrochemical gradient across the membrane (two charge separations, q+, during transfer of two electrons from quinol to nitrate; 2q+/2e). Electrons from haem bH of NarI are donated to the [3Fe–4S] cluster of NarH. From there, they flow via the iron–sulphur clusters in NarH to the one in NarG, which is the direct electron donor to the Mo-bis-MGD cofactor-containing catalytic site in NarG where nitrate is reduced to nitrite (Richardson & Sawers, 2002). In many species of bacteria that have nNarG, the nitrite produced is further reduced to nitric oxide, nitrous oxide and dinitrogen in a series of enzyme-catalysed reactions of denitrification. In each case, these reactions take place in the periplasm or towards the periplasmic face of the membrane and can be coupled to energy conservation through the protonmotive activity of the cytochrome bc1 complex (Q-cytochrome c oxidoreductase) (Berks et al., 1995). Like NarI, this complex binds two haems: one at either side of the membrane. However, rather than oxidizing one quinol per turnover (like NarI), it oxidizes two quinols at the periplasmic face and moves two electrons to the cytoplasmic face where it reduces one quinone. Overall, this so-called Q-cycle effectively translocates two positive charges per quinol oxidized (2q+/2e) and so it is bioenergetically equivalent to nNarI. This then makes the reduction of nitrate, nitrite, nitric oxide and nitrous oxide bioenergetically equivalent, despite the active site of the nNar system being on the Δψ− side of the membrane and the active sites of the nitrite, nitric oxide and nitrous oxide reductases being on the Δψ+ side.

The importance of the cytochrome bc1 complex to this bioenergetic equivalence can be illustrated when another kind of respiratory nitrate reductase is considered — the periplasmic nitrate reductase or NapA (Berks et al., 1995). This enzyme, like nNarG, is widely spread among Gram-negative proteobacteria and indeed is found in many bacteria, like Paracoccus denitrificans and E. coli, that can also express an nNar system. However, NapA is not closely related to the Nar enzymes. The molybdenum is co-ordinated by a Cys rather than Asp ligand (Jormakka et al., 2004) and there is a rather poor identity overall (e.g. only 12% identity between E. coli nNarG and NapA). Nap does reduce selenate although this enzyme is catalytically quite distinct from Nar enzymes (Butler et al., 1999; Sabaty et al., 2001). NapA is commonly coupled to quinol oxidation via a tetrahaeme cytochrome called NapC (Roldan et al., 1998; Cartron et al., 2002) that is not electrogenic and so there is no net charge translocation associated with quinol oxidation (Fig. 1); consequently, electron transport to nitrate via the Nap system can only be energy conserving if the electron input into the Q-pool is protonmotive. This can be the case if electrons enter by a proton-translocating enzyme (e.g. NADH dehydrogenase) or an electrogenic enzyme (e.g. formate dehydrogenase or hydrogenase).

Turning back to the pNarG and nNarG enzymes, one can now ask the question of whether being located at the Δψ+ face of the membrane consigns pNarG to being a poorly coupled enzyme, like NapA, or whether there is a mechanism by which pNarG can maintain bioenergetic equivalence with its nNarG cousin? At present, it is not known how electrons move from the Q-pool to pNarG. However, the genetic context of the Haloferax mediterranei pNar and Haloarcula marismortui pNar provides an interesting possibility. Analysis of the genes of the pnar cluster reveals that one of the genes encodes a protein (NarC) that has sequence similarity (>50%) with the dihaem subunits of quinol-cytochrome c reductases, such as the well-studied cytochrome bc1 complex of mitochondria and b6f complex of plants and cyanobacteria (Lledo et al., 2004; Yoshimatsu et al., 2006). This subunit is predicted to fold into nine transmembrane helices and bind two b-haemes, stacked across the membrane, with bis-histidinyl co-ordination between helices II and IV. It should be noted that the Cys residue of Chlamydomas reinhardtii that makes a thioether bond to a third haem (Stroebel et al., 2003) is not conserved in the Haloferax mediterranei pNar and Haloarcula marismortui pNar and so the cytochrome b6 is predicted to be a dihaem, rather than a trihaem protein. Recent experimental evidence that can be drawn on to support for this view comes from the recent purification of NarC that showed that it does indeed bind two b-haems (Yoshimatsu et al., 2006). Adjacent to narC is a gene (narB) that is predicted to encode a Rieske iron–sulphur protein also of the type found in the protonmotive Q-cycling cytochrome bc1 or b6f complexes (Lledo et al., 2004; Yoshimatsu et al., 2006) (Fig. 5). From the present analysis, NarB is predicted to bear an N-terminal signal anchor with the redox-active C-terminal domain located at the p-side of the cytoplasmic membrane. Integration and orientation of NarB in the membrane is likely to be conducted by the Tat system because the signal anchor contains all the features of a Tat signal peptide and assembly of the Rieske protein has recently been shown to be Tat-dependent in bacteria (Bachmann et al., 2006; De Buck et al., 2007). In addition to having the residues (two Cys and two His) that bind a 2Fe2S cluster, we note that NarB also has two additional conserved Cys residues. These characteristically form a disulphide bond in Rieske proteins that modulates the redox properties of the iron–sulphur cluster so that it operates at a high potential (Emc. 250 mV) (Leggate & Hirst, 2005). This suggests that NarB is a true Rieske protein, rather than the Rieske-type protein found in the bacterial aromatic ring dioxygenases and the assimilatory nitrate reductases that lack the disulphide and operate at much lower potentials (Em<0 mV) (Butler & Mason, 1997). This then raises the possibility that the Haloferax mediterranei and Haloarcula marismortui pNar systems maintain bioenergetic parity with the nNar systems by being coupled to a protonmotive Q-cycle mechanism. Such a coupling mechanism has thus far been unprecedented in any other respiratory nitrate reductase system so far studied. As a Q-cycle activity is sensitive to the classical inhibitor antimycin A that binds to the cytochrome b subunit of plant and bacterial cytochrome bc1/b6f complexes, but not the structurally distinct NarI cytochrome b subunit of the pNar systems the sensitivity of nitrate reduction by the Haloferax mediterranei pNar system to this inhibitor has been assessed. The results show a high level of sensitivity (Fig. 6) to this inhibitor at concentrations that inhibit nitrite, nitric oxide and nitrous oxide reduction in Paracoccus denitrificans, all of which are dependent on the cytochrome bc1 complex, but do not inhibit nitrate reduction by the cytochrome bc1 complex-independent nNarG system (Berks et al., 1995).

Figure 5

A Q-cycle coupling mechanism for the pNar enzyme of Haloferax mediterranei.

Figure 6

Effect of Antimycin A on anaerobic growth and nitrate reduction by Haloferax meditteranei. Cells were grown under anaerobic conditions in maximal culture media in the presence of 100 mM KNO3 as described in Lledo (2004). Antimycin A was added to a final concentration of 100 µM at the point indicated.

In the cytochrome bc1 or b6f complexes, electrons flow from cytochrome b to the Rieske iron–sulphur centre and then to cytochrome c1 or cytochrome f. In the Haloferax mediterranei and Haloarcula marismortui nar gene clusters, there is no gene encoding a homologue of cytochromes c1 or f. There is, though, a gene (Orf7) that we note encodes another putative cytochrome subunit that has some homology with a group of soluble or membrane-anchored b-type cytochromes encoded in the operons of many of the other D-group molybdoenzyme systems, for example SerC in the T. selenatis selenate reductase system (Krafft et al., 2000) and EbdC in the ethylbenzene dehydrogenase system (Kniemeyer & Heider, 2001) (Fig. 5). This once again draws similarities between pNar systems and other D-type Mo-bis-MGD systems that are orientated towards the membrane potential-positive side of the membrane. Importantly, the EdbC structure has recently been resolved (Kloer et al., 2006) and shows it to be a direct redox partner with the EdbAB subunits that are equivalent to pNarGH. Significantly, the key lysine and methionine residues that provide a novel haeme co-ordination in EbdC are conserved in Haloferax mediterranei Orf7. This then strongly suggests that Orf7 is a functional homologue of EbdC and the direct electron donor to pNarGH, mediating electron transfer that is coupled from the Q-pool via the cytochrome b/Rieske protein complex.

If there is a Q-cycle mechanism for energy conservation for the Haloferax mediterranei and Haloarcula marismortui pNar systems, how widespread is it among other pNars? Examination of the pnar gene clusters of Pyrobaculum aerophilum, Archaeoglobus fulgidus and Aeropyrum pernix reveals that they do not encode NarC homologues. This does not exclude the possibility that the pNar system is coupled to a cytochrome b6–Rieske-type system encoded elsewhere on the chromosome. Genome analysis reveals cytochrome b6 homologues in Pyrobaculum aerophilum and Aeropyrum pernix. This leaves open the possibility that some, but not all pNars are coupled at the level of the Q-pool. If there is no cytochrome b–Rieske system and so no protonmotive Q-cycle, electron transfer will simply be bioenergetically equivalent to the periplasmic nitrate reductase (Nap) system, further highlighting the observation that one cannot assume that all pNar systems are bioenergetically equivalent.

Concluding remarks

In this minireview, the evidence has been discussed for the active subunits of ‘Archaeal’ or ‘pNar’ membrane-bound nitrate reductases being secreted by Tat to the outside of the cytoplasmic membrane, it has been proposed how to confidently identify a pNar from amino acid sequence analysis and the bioenergetic consequences of pNar being active on the membrane potential-positive side of a biological membrane have been explored. When prokaryotic nitrate reduction is considered as a whole, it is apparent that there are two broad subclasses of Mo-bis-MGD nitrate reductases, the ‘Nar group’ and the ‘Nap group’, which are structurally distinguished by having aspartate and cysteine Mo-bis-MGD ligands, respectively (Jormakka et al., 2004). Like the ‘Nar’ group, the ‘Nap’ group also includes nitrate reductases that are exported by the TAT system (the catabolic periplasmic NapA) and nitrate reductases that are active in the cytoplasm (the anabolic Nas enzymes associated with nitrogen assimilation) (Richardson et al., 2001). As yet, no Nap system has been reported to be coupled to a protonmotive Q-cycling cytochrome b–Rieske complex. The suggestion that at least some pNar systems may be coupled to such a complex highlights emerging bioenergetic, as well as structural, differences between these two groups of nitrate-reducing systems located on the membrane potential-positive side of the membrane.

Authors' contribution

R.M.M.E. and E.J.D. are joint first authors.


This work was funded in part by a research grants from the BBSRC (BB/D00781X/1, BB/D018986) and from the MEC-Spain (BIO2005-08991-C02-01). F.S. is a Royal Society University Research Fellow. The authors would like to thank Dr Rick Lewis (University of Newcastle) for help with the molecular modelling.


  • Editor: Rustam Aminov


View Abstract