Nitrate uptake is essential for various bacterial processes and combines with nitrite export to form the usual initial steps of denitrification, a process that reduces nitrate to dinitrogen gas. Although many bacterial species contain NarK-like transporters that are proposed to function as either nitrate/proton symporters or nitrate/nitrite antiporters based on sequence homology, these transporters remain, in general, poorly characterized. Several bacteria appear to contain a transporter that is a fusion of two NarK-like proteins, although the significance of this arrangement remains elusive. We demonstrate that NarK from Paracoccus denitrificans is expressed as a fusion of two NarK-like transporters. NarK1 and NarK2 are separately capable of supporting anaerobic denitrifying growth but with growth defects that are partially mitigated by coexpression of the two domains. NarK1 appears to be a nitrate/proton symporter with high affinity for nitrate and NarK2 a nitrate/nitrite antiporter with lower affinity for nitrate. Each transporter requires two conserved arginine residues for activity. A transporter consisting of inactivated NarK1 fused to active NarK2 has a dramatically increased affinity for nitrate compared with NarK2 alone, implying a functional interaction between the two domains. A potential model for nitrate and nitrite transport in P. denitrificans is proposed.
Uptake of nitrate to the cytoplasm is a fundamental step in a variety of bacterial processes, including nitrogen assimilation and denitrification. Denitrification is a respiratory process involving four sequential reductions of nitrogen oxides, eventually reducing nitrate to dinitrogen gas (Zumft, 1997). In Paracoccus denitrificans, the first of these steps, reduction of nitrate to nitrite, is catalysed by the membrane-bound nitrate reductase Nar, consisting of three subunits, NarG, NarH and NarI (Berks et al., 1995a; Wood et al., 2001). The active site of this enzyme is cytoplasmic (Ballard and Ferguson, 1988), whereas the enzymes for subsequent reduction steps, including cd1 nitrite reductase (Berks et al., 1995b), are located in the periplasm. Therefore, there must be a mechanism by which nitrate is transported to the cytoplasm and nitrite back to the periplasm. Nitrate is an ion at physiological pH (pKa HNO3 = −1.3) which, along with the fact that the cytoplasmic membrane potential is approximately 180 mV negative inside, would seem to eliminate inward nitrate transport via passive diffusion. Indeed, such a mechanism would limit the intracellular nitrate concentration to 0.1% of its external concentration, a value that is unrealistic given that cells can reduce nitrate at external concentrations as low as 1 μM (Parsonage et al., 1985). Contrastingly, the pKa of nitrous acid (HNO2) is 3.3, suggesting that nitrite may diffuse across biological membranes in its protonated state; whether this process could sustain an adequate rate of nitrite efflux is debatable.
It has been suggested that nitrate transport in assimilatory nitrate reductase pathways occurs via two types of transport, ABC-type transporters and secondary transporters (Moir and Wood, 2001). However, transporters associated with dissimilatory pathways appear to be exclusively secondary transporters. Recently, such proteins have been identified in a number of organisms, including Escherichia coli (Clegg et al., 2002), Bacillus subtilis (Cruz Ramos et al., 1995), Staphylococcus carnosus (Fast et al., 1996), Thermus thermophilus (Ramirez et al. 2000), Mycobacterium tuberculosis (Sohaskey and Wayne, 2003) and Paracoccus pantotrophus (Wood et al., 2002). All of these transporters are members of the major facilitator superfamily (MFS), forming a distinct subfamily known as NarK. Phylogenetic and functional studies have divided NarK family members into two distinct subfamilies. NarK1-like transporters are proposed to be nitrate/proton symporters whereas NarK2-like transporters are proposed to be nitrate/nitrite antiporters (Moir and Wood, 2001; Wood et al., 2002). Although genome sequencing and functional analysis have revealed many bacterial species to contain multiple NarK-like proteins, an unusual arrangement exists within certain organisms.
NarK family members are generally predicted to contain 12 transmembrane helices, being approximately 50 kDa in size. However, P. pantotrophus was found to contain what appeared to be a fused NarK protein consisting of 24 transmembrane helices, although there is no direct evidence as to whether this functions as a fusion of two domains or is post-translationally cleaved (Wood et al., 2002). The availability of sequence data for a wide variety of bacterial species has revealed that this arrangement is not confined to P. pantotrophus and its close relative P. denitrificans. In each case, the fusion consists of a NarK1-like transporter followed by a NarK2-like transporter. The existence of multiple apparently fused transporters suggests some physiological role for this arrangement. Additionally, it is still unclear as to whether NarK1-like proteins are confined to action as nitrate/proton symporters and, similarly, if NarK2-like proteins are obligate nitrate/nitrite antiporters.
In this paper we present evidence that NarK in P. denitrificans is assembled in the membrane as an uncleaved fusion of two NarK-like domains. Each domain can support anaerobic growth in the presence of nitrate but with associated growth defects. Coexpression of the two domains largely overcomes these defects. Each transporter can be inactivated by mutation of conserved arginine residues to leucine. Interestingly, fusion of inactive NarK1 to active NarK2 results in a significant decrease in the estimated Km for nitrate of the fused transporter compared with that of NarK2 alone. Disruption of P. denitrificans NarK can also be overcome by expression of E. coli NarK and NarU. Implications for the transport of nitrate and nitrite in P. denitrificans are discussed.
Phlylogenetic analysis of fused NarKs
Before recent genome sequencing projects, the existence of a NarK protein consisting of two apparently fused domains appeared to be confined to certain Paracoccus species. Our analysis revealed this arrangement to be more prevalent. blast (Altschul et al., 1997) analysis using the P. denitrificans NarK sequence revealed apparently fused homologues in a number of Brucella species, Hyphomonas neptunium, Chromohalobacter salexigens, Alcanivorax borkumensis, Ochrobactrum anthropi, Oligotropha carboxidovorans, Methylocella silvestris and Azorhizobium caulinodans (Fig. 1A). This analysis revealed that such an arrangement is not confined to the α-proteobacteria, with both C. salexigens and A. borkumensis being members of the γ-proteobacteria. The transporters from these two species appear more closely related to each other than to any of the transporters from α-proteobacteria. It is also interesting to note that in each case, the fusion is arranged as a NarK1-like domain followed by a NarK2-like domain. A similar blast search using an artificial construct in which the two domains of P. denitrificans NarK have been reversed (NarK2–NarK1) revealed no additional fused proteins in this orientation. This suggests that there may be a functional consequence of the fusion, beyond tethering the two domains together at the membrane.
The region between predicted transmembrane domain 12 (the last of NarK1) and transmembrane domain 13 (the first of NarK2) is relatively poorly conserved (Fig. 1B). Both the protein sequence and the length of this region are variable, but with 10 perfectly conserved residues, four of which lie very close to the predicted start of transmembrane domain 13. This lack of sequence similarity suggests that if these proteins are cleaved, there is not a conserved recognition site for protease activity. We chose to focus on studying the NarK protein from P. denitrificans, as the latter has a completely sequenced genome and is a well-characterized model organism for denitrification (Baker et al., 1998).
Construction of a ΔnapD ΔnarK strain
Paracoccus denitrificans can synthesize three nitrate reductase enzymes. The first, NarGHI, is relevant to this study, being the membrane-bound nitrate reductase associated with the dissimilatory pathway. A second putative nitrate reductase, NasC, has recently been identified via genome sequencing and is predicted to form part of a complex with a putative nitrite reductase NasD and also NasE. NasC is likely to function as part of the assimilatory nitrate reductase and its expression should be repressed in the presence of ammonia, which is utilized as a nitrogen source throughout this study. However, the third, periplasmic, nitrate reductase, Nap, could complicate interpretation of NarK function as its activity is independent of transport of nitrate to the cytoplasm. Therefore, all strains involved in this study contain an unmarked deletion of the gene encoding NapD, a protein involved in cofactor delivery to the periplasmic nitrate reductase. ΔnapD strains do not produce active periplasmic nitrate reductase (Wood et al., 2002). Therefore, in a ΔnapD strain under anaerobic conditions in the presence of ammonia, only NarGHI will be functional.
Previous studies in P. pantotrophus, a close relative of P. denitrificans, demonstrated that narK is expressed as part of an operon with the structural genes for the Nar nitrate reductase (Wood et al., 2001). It was also demonstrated that insertional mutagenesis of narK abolished NarGHI nitrate reductase activity (Wood et al., 2002) presumably due to disruption of the regulatory control or transcription of the operon. Given the similarity in sequence and arrangement of nar genes between P. pantotrophus and P. denitrificans, it seems likely that these genes form an operon in both organisms. Therefore, this study utilizes an unmarked deletion in narK in which the entire narK open reading frame (ORF) has been deleted from the chromosome.
NarK is expressed as a fusion of two NarK-like domains
Sequence analysis of NarK from P. denitrificans revealed that, as in P. pantotrophus (Wood et al., 2002), it appears to be a fusion of two NarK-like domains generating a protein with 24 predicted transmembrane helices. A fundamental question to address was whether this protein is inserted into the membrane as a 24-helix protein or whether post-translational cleavage generates two ‘typical’ NarK proteins. In bacteria, post-translational modifications generating two functional proteins from a single precursor are rare but not unprecedented. Examples of such events include Bacillus polymyxaβ- and α-amylases (Uozumi et al., 1989), E. coli penicillin acylase (Sizmann et al., 1990) and Bradyrhizobium japonica cytochromes b and c1 (Thöny-Meyer et al., 1991).
For the purposes of this study, the full-length NarK is defined to contain two distinct NarK-like proteins, termed NarK1 and NarK2. NarK1 is comprised of all residues up to and including V441 (Fig. 1B). A methionine is present at position 442 that is approximately equidistant between the predicted end of helix 12 and the predicted start of helix 13 (Fig. 1B, starred). Therefore, given that a P. pantotrophus NarK2 protein was functional when the corresponding residue was used as the start codon (Wood et al., 2002), this methionine was used as the starting residue for NarK2. In order to determine whether the full-length NarK protein was post-translationally cleaved, C-terminal hexahistidine-tagged variants of NarK, NarK1 and NarK2 were generated (NarK-His, NarK1-His and NarK2-His respectively). These were cloned into the pEG276 expression vector (Gordon et al., 2003) that contains a constitutive promoter, and expressed in the ΔnapDΔnarK strain.
Each strain was grown aerobically in Luria–Bertani (LB) media; membrane fractions were prepared and immunoblotted using a commercial HRP-conjugated antipentahistidine antibody (Fig. 2A). This demonstrated that the full-length NarK (apparent molecular weight ∼52 kDa) had a higher molecular weight than either NarK1 (∼35 kDa) or NarK2 (∼36 kDa), with NarK2 being slightly larger than NarK1, consistent with the division of the two domains at M442. It should be noted that although each construct migrates with an aberrant molecular weight (NarK has a predicted Mr of 98.358 kDa, while those of NarK1 and NarK2 are 48.290 kDa and 50.085 kDa respectively, and in each case addition of a hexahistidine tag adds ∼841 Da), it is not unusual for membrane proteins to exhibit an apparent reduction in their molecular mass during SDS-PAGE analysis. This may be due to their hydrophobicity, high binding of SDS or secondary structure retention (Ward et al., 2000) and, in this case, is highly unlikely to be due to N-terminal cleavage of each protein. Unfortunately, due to apparent N-terminal blocking, we were unable to sequence the purified NarK-His construct to confirm the presence of the entire N-terminus. However, MALDI-TOF-TOF mass spectrometry after tryptic digestion of purified NarK-His revealed the presence of seven peptides corresponding to regions of NarK1 (including residues 42–66), NarK2 (including residues 811–820) and the linker region (residues 425–439) between the two transporters.
NarK1 and NarK2 exhibit growth defects when used to complement a NarK deletion
Phylogenetic and functional studies have proposed two families of NarK proteins (Moir and Wood, 2001; Wood et al., 2002). NarK1 family members are predicted to function as nitrate/xH+ symporters whereas NarK2 family members are predicted to be nitrate/nitrite antiporters. P. denitrificans NarK1 clusters with other NarK1 members and likewise NarK2 lies within the NarK2 family. To investigate the functions of each domain, NarK1 and NarK2 were expressed separately in a ΔnapDΔnarK strain and growth assayed in the presence of 20 mM nitrate (Fig. 2B).
A ΔnapD strain exhibited a typical growth curve, with a lag phase lasting approximately 6 h, followed by a period of exponential growth (μ = 0.44 ± 0.04 h−1), reaching a maximum OD600 of 2.47 ± 0.11 after ∼16 h. A ΔnapDΔnarK strain containing pEG276 exhibited very low levels of growth, presumably due to the small amount of dissolved oxygen within the medium. This growth was independent of the concentration of nitrate (Fig. 3C). Expression of NarK in a ΔnapDΔnarK strain complemented the growth defect, confirming that the unmarked deletion in narK is non-polar. In this case μ = 0.41 ± 0.03 h−1 and the maximum OD600 = 2.32 ± 0.13. Expression of NarK1 resulted in an increase in the duration of the lag phase compared with the full-length transporter, from ∼6 h to ∼14 h. After this period, exponential growth at μ = 0.41 ± 0.02 h−1 occurred to a maximum cell density of OD600 = 2.22 ± 0.24. A ΔnapDΔnarK strain expressing NarK2 had a lag phase of ∼8 h but then grew exponentially at a slower rate than either NarK or NarK1 (μ = 0.20 ± 0.02 h−1) and also reached a lower maximum cell density (OD600 = 1.38 ± 0.09).
Nitrate taken up by NarK is converted to nitrite by the NarGHI nitrate reductase. To continue denitrification, this nitrite must be transported to the periplasm where it can be further reduced by cd1 nitrite reductase. Therefore, the above strains were also assayed for the transient accumulation of extracellular nitrite before it is reduced (Fig. 2C). In a ΔnapD strain, a rapid efflux of nitrite occurred after ∼6 h, coinciding with the start of exponential growth. The extracellular nitrite concentration peaked at ∼700 μM and was then presumably reduced by cd1, resulting in a decrease in extracellular nitrite concentration. A second, smaller peak appeared to occur after ∼18 h, which will be discussed later. As would be expected, a ΔnapDΔnarK strain exhibited little extracellular nitrite accumulation, consistent with a lack of periplasmic nitrate reductase and an inability to transport nitrate to, and nitrite from, the cytoplasm. Complementation of the ΔnapDΔnarK strain with NarK resulted in a nitrite peak similar to that observed in a ΔnapD strain (reaching a maximum of ∼800 μM). A strain expressing NarK1 had a dramatically reduced nitrite peak (maximum ∼90 μM), which also occurred later (after ∼12 h), with little nitrite present during exponential growth. This, along with the fact that the growth rates and maximal cell densities of NarK- and NarK1-expressing strains are similar, suggests that NarK1-expressing strains may have a defect in nitrite excretion. A ΔnapDΔnarK strain expressing NarK2 had a lower nitrite peak than the full-length NarK (maximum ∼250 μM), which again occurred at the start of exponential growth. This lower peak, along with slower growth and a lower maximal OD600, is consistent with a defect in nitrate uptake in NarK2-expressing strains, resulting in a lower intracellular concentration of nitrite to be excreted.
NarK1 and NarK2 have different affinities for nitrate
To investigate if the growth defects observed were a result of different affinities for nitrate between the transporters, growth assays were conducted at various nitrate concentrations (Fig. 3). Using the Monod equation, Ks (the concentration at which growth occurs at half the maximal rate) can be calculated. In the following study it has been assumed that nitrate transport is the limiting factor in nitrate metabolism under denitrifying conditions (Kucera and Kaplan, 1996) and hence the apparent Km for nitrate (the concentration at which nitrate transport occurs at half the maximal rate) will be equal to Ks. For each strain, μmax and apparent Km were calculated using Hanes plots. These data are shown in Table 1 as averages of triplicate experiments ± SD. For clarity, mean values are stated in the text.
Table 1. Growth characteristics of NarK-expressing strains.
μmax and Km values for each strain were calculated using Hanes plots. Values are averages of triplicate experiments ± SD. The source data are shown in the indicated figures.
0.48 ± 0.01
0.20 ± 0.06
ΔnapDΔnarK + pEG276-NarK
0.40 ± 0.03
0.17 ± 0.03
ΔnapDΔnarK + pEG276-NarK1
0.39 ± 0.06
0.66 ± 0.14
ΔnapDΔnarK + pEG276-NarK2
0.41 ± 0.02
15.0 ± 1.8
ΔnapDΔnarK + pEG276-NarKR66L
0.35 ± 0.04
1.2 ± 0.1
ΔnapDΔnarK + pEG276-NarKR269L
0.38 ± 0.02
1.2 ± 0.1
ΔnapDΔnarK + pEG276-NarKR520L
0.40 ± 0.03
0.55 ± 0.14
ΔnapDΔnarK + pEG276-NarKR736L
0.40 ± 0.02
0.67 ± 0.18
ΔnapDΔnarK + pEG276-coliNarK
0.37 ± 0.01
0.18 ± 0.01
ΔnapDΔnarK + pEG276-coliNarU
0.34 ± 0.01
0.20 ± 0.06
A ΔnapD strain exhibited a very similar growth curve at 50 mM nitrate to that observed at 20 mM (Fig. 3A). This suggests that concentrations of nitrate ≥ 20 mM are non-limiting to growth under the conditions of this assay. Maximal growth of this strain at reduced nitrate concentrations (10, 5 or 1 mM) was concomitantly attenuated. The growth rate of the strain was largely unaffected at nitrate concentrations of 5 mM or above but was reduced slightly at 1 mM nitrate, although the duration of the exponential phase at this concentration was lessened compared with that at higher nitrate levels (Fig. 3A). The μmax of this strain was 0.48 h−1 with a Km of 200 μM. A ΔnapDΔnarK strain exhibited a very low level of growth that was independent of nitrate concentration (Fig. 3C). Complementation of this strain with NarK resulted in a similar pattern of growth to a ΔnapD strain with μmax = 0.40 h−1 and Km = 170 μM (Fig. 3E).
Complementation of a ΔnapDΔnarK strain with NarK1 (Fig. 3G) resulted in an increased lag phase at all concentrations of nitrate compared with the same strain expressing NarK. NarK1 had a similar μmax to the full-length protein (0.39 h−1) but an increased Km (660 μM). Expression of NarK2 resulted in a decrease in both growth rate and maximal cell density compared with NarK for all concentrations of nitrate < 50 mM (Fig. 3I). This strain exhibited no growth above that demonstrated by a strain containing vector alone at 1 mM nitrate (compare Fig. 3I and C). The μmax of 0.41 h−1 for NarK2 is comparable with both NarK and NarK1 and the Km is significantly higher, 15 mM. These data suggest that NarK1 is a relatively high-affinity nitrate transporter, whereas NarK2 is a low-affinity transporter.
It is more difficult to estimate the Km of each transporter for nitrite as the intracellular concentration of nitrite is inherently limited by the rate of transport of nitrate. However, the extracellular nitrite accumulations observed are consistent with the transport defects proposed earlier. Both a ΔnapD strain and a ΔnapD ΔnarK strain complemented with NarK exhibited similar levels of extracellular nitrite accumulation (maximal ∼800 μM, beginning at ∼6 h) at concentrations of nitrate equal to or above 5 mM, with a slight decrease at 1 mM nitrate (compare Fig. 3B and F). A ΔnapDΔnarK strain containing empty vector had no appreciable nitrite accumulation (Fig. 3D). The same strain expressing NarK1 had very low levels of extracellular nitrite accumulation (maximal ∼100 μM), which also occurred later than in the previously described strains (beginning at ∼8 h; Fig. 3H). At nitrate concentrations ≥ 5 mM, a NarK2-expressing strain exhibited an intermediate level of extracellular nitrite accumulation (maximal ∼350 μM), beginning after ∼6 h, but very little accumulation at 1 mM nitrate, consistent with the lack of growth under these conditions (Fig. 3J). These data again suggest that NarK1 has limited capacity for nitrite export whereas NarK2 has a lowered affinity for nitrate (which results in a lower level of nitrite export) compared with the full-length transporter.
It should be noted that in the case of a ΔnapD strain and a ΔnapDΔnarK strain expressing either NarK or NarK1, there was a second nitrite peak approximately 12 h after the first peak. This peak only occurred in the presence of 50 mM nitrate and in some cases a severely reduced peak can be seen in the presence of 20 mM nitrate. As concentrations above 20 mM are non-limiting to growth, there will still be appreciable extracellular nitrate even when the cultures have reached stationary phase. As this peak occurred well into stationary phase, it is possible that there are some cells that become ‘leaky’ such that nitrate can permeate the cell membrane, and this nitrate is subsequently converted to nitrite by the NarGHI nitrite reductase and is then released into the media. It is also possible that high intracellular levels of nitrite lead to expression of an as yet unidentified nitrite export protein.
In order to investigate expression of the NarGHI nitrate reductase in the complemented ΔnapDΔnarK strains, Western blot analysis was performed using an anti-NarGH antibody (Fig. S1). This demonstrated equivalent expression levels of the NarG and NarH subunits in the ΔnapD strain and in a ΔnapDΔnarK strain complemented with NarK, NarK1 or NarK2. This confirms that differences in growth are due to the different transporters expressed and not due to altered nitrate reductase levels.
Growth defects are partially rescued by expression of unfused NarK1 and NarK2
In order to investigate whether the growth defects associated with both NarK1 and NarK2 could be overcome by coexpression of the two transporters in the same strain, an unmarked deletion of narK2 was created. This strain lacks the coding region of narK2 and has an artificial stop codon introduced after the last codon (V441) of narK1 (ΔnapD, ΔnarK2). This strain was unable to grow anaerobically in the presence of nitrate (Fig. 4A, pEG276). Maximal expression of the narK operon requires NarR, a member of the FNR family of transcriptional regulators, which requires nitrate and/or nitrite for full activity (Wood et al., 2001). Therefore, mutations that perturb either nitrate entry into the cell, its reduction to nitrite, or the activity of NarR can affect narK operon regulation. This observation is consistent with work in P. pantotrophus in which an unmarked deletion in narK2 was also unable to grow anaerobically in the presence of nitrate but expression of the structural nitrate reductase genes could be induced by azide (Wood et al., 2002).
Complementation of the ΔnapDΔnarK2 strain with full-length NarK restored growth to a similar rate (μ = 0.41 ± 0.02 h−1) and density (OD600 = 2.08 ± 0.10) to a ΔnapD strain (μ = 0.39 ± 0.02 h−1, OD600 = 2.15 ± 0.10; Fig. 4A). However, there was a slightly increased lag phase in the complemented strain. When NarK1 was expressed in a ΔnapD, ΔnarK2 strain, a significantly increased lag phase was observed, although a similar cell density (OD600 = 2.26 ± 0.09) and growth rate (μ = 0.38 ± 0.05 h−1) to NarK were achieved. Complementation of the same strain with NarK2 restored the maximum cell density (OD600 = 2.16 ± 0.11) with a growth rate of 0.32 ± 0.02 h−1. However, there was again a slightly increased lag phase (∼8 h compared with the ΔnapD strain, ∼4 h). When compared with a strain lacking the entire narK ORF expressing NarK2 (μ = 0.20 ± 0.02 h−1, OD600 = 1.38 ± 0.09; Fig. 2B), this suggests that the two domains of NarK can largely complement the associated growth defects when expressed separately.
Quantification of extracellular nitrite (Fig. 4B) revealed a similar peak of ∼900 and ∼700 μM, respectively, for a ΔnapD strain and a ΔnapDΔnarK2 strain complemented with NarK. Complementation of the same strain with NarK1 resulted in a reduced peak of nitrite (∼100 μM), which occurred later, corresponding with the increased lag phase. Complementation of the ΔnapDΔnarK2 strain with NarK2 resulted in a peak of ∼350 μM, slightly higher than in a ΔnapDΔnarK strain complemented with NarK2 (∼250 μM; Fig. 2B).
Two conserved arginine residues are essential for activity of both NarK1 and NarK2
It has been demonstrated in a number of nitrate transporters (Unkles et al., 2004; Jia and Cole, 2005) that two conserved arginine residues (in transmembrane domains 2 and 8) are essential for substrate transport, with the positive side-chains proposed to interact with the negative substrate. These arginine residues are conserved within NarK1 (R66 and R269) and NarK2 (R79 – equivalent to R520 in NarK and R295 – equivalent to R736 in NarK).
In order to study the potential transport mechanism of NarK1 and NarK2, each of these arginine residues were mutated to leucine, generating NarK1R66L, NarK1R269L, NarK2R79L and NarK2R295L. Expression of these constructs in a ΔnapDΔnarK strain failed to recover growth to a level above that exhibited in the vector-alone control (data not shown). A C-terminally tagged hexahistidine variant of each mutant was produced and expression confirmed by Western blot analysis (Fig. 5). In each case the mutant transporter was expressed at an approximately equivalent or higher level than its wild-type counterpart. Presumably, as growth is dependent on the ability of the transporters to import nitrate, each of the above mutants has a defect in nitrate transport. However, it is also possible that there is a defect in nitrite export, leading to toxic accumulation of nitrite within the cytoplasm; the nitrite anion can inhibit the electron flow through electron carrier proteins by binding to heme groups (Rowe et al., 1979). It must also be considered that nitrite may diffuse across the plasma membrane in its protonated form. These results are consistent with NarK1 and NarK2 acting via a similar mechanism to other members of the NarK family.
Fusion of inactive NarK1 to active NarK2 alters transport activity
As neither NarK1 nor NarK2 alone have comparable Km values to the full length NarK, it is possible that an interaction occurs between the two domains which is mediated by the structural fusion. To investigate any such effects, fusions between one inactive and one active transporter were created. To do so, each arginine was mutated to a leucine in the full-length NarK, producing NarKR66L, NarKR269L, NarKR520L and NarKR736L, and μmax and apparent Km determined for each strain as described previously (Table 1).
When expressed in a ΔnapDΔnarK strain, NarKR66L (Fig. 6A) and NarKR269L (Fig. 6C) grew in a very similar manner at all concentrations of nitrate. The μmax for NarKR66L was 0.35 h−1 and that of NarKR269L 0.38 h−1 while the Km was 1.2 mM in each case. Although the μmax values are comparable with that of NarK2, a fusion protein in which the NarK1 domain is predicted not to transport nitrate has a significantly lower Km than NarK2 expressed alone. Consequently, the maximal cell density achieved by either inactive or active transporter is almost identical to that of the wild-type NarK, while growth of NarK2 is restricted at all nitrate concentrations < 50 mM. Contrastingly, fusion of an inactive NarK2 to active NarK1, either R520L (Fig. 6E) or R736L (Fig. 6G), had little effect upon transporter activity compared with NarK1 (compare with Fig. 3G). NarKR520L had a Km of 550 μM and μmax of 0.40 h−1 while NarKR736L had a Km of 670 μM and μmax of 0.40 h−1.
Both NarKR66L and NarKR269L had an extracellular nitrite peak (maximal ∼600 μM) occurring after ∼10 h, corresponding with the start of exponential growth. This peak is larger than that of NarK2 alone, correlating with the decreased Km for nitrate that presumably sustains an increased intracellular nitrate (and hence nitrite) concentration. Correspondingly, this peak is lower than that of NarK that has a Km of 166 μM. Strains expressing either NarKR520L (Fig. 6F) or NarKR736L (Fig. 6G) exhibited a very low level of extracellular nitrite accumulation (∼100 μM), which is comparable with that observed in a strain expressing NarK1 alone.
To examine expression levels, C-terminal hexahistidine-tagged variants of each mutant were produced, and Western blot analysis conducted as described previously. This confirmed that all mutants are expressed at an equivalent level to the wild-type protein (Fig. 7)
A ΔnarK strain can be complemented by E. coli NarK and NarU
NarK and NarU from E. coli represent two of the best characterized bacterial NarK transporters. Although E. coli NarK was initially proposed to be a nitrite extrusion protein (Rowe et al., 1994), both NarK and NarU have since been demonstrated to transport nitrate and nitrite (Clegg et al., 2002). In order to investigate whether a deletion in P. denitrificans NarK could be fully complemented by a single 12 transmembrane domain protein, E. coli NarK and NarU were expressed in a ΔnapDΔnarK strain; μmax and Km values were calculated as described previously and are shown in Table 1.
Both E. coli NarK (Fig. 8A) and NarU (Fig. 8C) were capable of restoring anaerobic growth in the presence of nitrate when expressed in a ΔnapDΔnarK strain with similar affinities for nitrate and μmax to P. denitrificans NarK (Table 1). In addition to the uptake of nitrate, these transporters must be capable of nitrite excretion, as evidenced by the large extracellular nitrite peak with a maximal of ∼800 μM after 10 h (Fig. 8B and D). Again, in the presence of 50 mM nitrate, there is a large extracellular nitrite peak that occurs after approximately 20 h, which has been discussed previously. These data support the proposal (Jia and Cole, 2005) that NarK and NarU from E. coli are capable of acting as nitrate/nitrite antiporters.
Analysis of available genome sequences has revealed a number of organisms to contain an apparently fused NarK transporter consisting of a NarK1-like domain followed by a NarK2-like domain. In this study, we present evidence that the NarK protein from P. denitrificans is expressed as a fusion of these two domains and is not post-traslationally cleaved. This is the first confirmation that such an arrangement exists in vivo. However, as each domain retains some activity when expressed separately, this suggests that the fusion consists of two 12-transmembrane domain transporters with differential function rather than a single transporter with a 24-transmembrane domain structure.
The two domains, NarK1 and NarK2, are capable of supporting anaerobic growth in the presence of nitrate when expressed individually. This suggests that both proteins are capable of transporting nitrate and potentially nitrite, although with different affinities. Each domain has an associated growth defect when expressed individually. NarK1 has an increased lag before exponential growth. This may be due to poor induction of the narGHJI genes required for formation of the membrane-bound nitrate reductase. Additionally, if nitrate uptake by NarK1 is indeed driven mainly by proton motive force, it may be necessary for induction and activity of other components of the denitrification pathway to occur as these contribute to the proton motive force. However, it is also possible, given that the maximal level of extracellular nitrite achieved by this transporter is significantly reduced, while the apparent Km and μmax for nitrate remain relatively unchanged compared with the full-length protein, that NarK1 has a greatly reduced capacity for nitrite export. This may lead to a build-up of intracellular nitrite that could affect NarGHI expression or activity. It has also been proposed (Moir and Wood, 2001) that NarK1 may be negatively regulated by nitrite. It may therefore be notable that in contrast to strain expressing either NarK or NarK2, there is little accumulation of extracellular nitrite during exponential growth of a NarK1-expressing strain. NarK2 grows at a slower rate than either the full-length NarK or NarK1 and also reaches a lower maximal cell density at concentrations ≤ 20 mM nitrate. This can be explained by the fact that the apparent Km for nitrate of NarK2 is significantly higher (15 mM) than the full-length NarK (170 μM).
These data suggest that NarK1 is a relatively high-affinity nitrate transporter whereas NarK2 is a low-affinity transporter. The Km of high-affinity nitrate transporters is generally in the mid micromolar range, e.g. 125 μM for CHL1 (Liu and Tsay, 2003) and 130 μM for NrtA (Unkles et al., 2004) whereas low-affinity transporters are generally in the low millimolar range, e.g. 2.2 mM for CHL1 (Liu and Tsay, 2003). A previous study by Kucera (2003) demonstrated that the Km of nitrate uptake in P. denitrificans was approximately 20 μM. This is lower than the value calculated in this study. However, it should be noted that Kucera (2003) demonstrated an inhibitory effect of nitrite upon the Km, increasing it to ∼150 μM in presence of 500 μM nitrite (half saturation ∼200 μM). The extracellular nitrite concentration present during exponential growth of a ΔnapDΔnarK strain in this study was found to be in excess of 200 μM. Therefore, the value of 170 μM calculated in this study agrees very closely with that determined by Kucera (2003).
It is also possible that each transporter has a different affinity for nitrite. The maximal extracellular concentration of nitrite reached in a strain expressing NarK1 is less than half of that achieved by a strain expressing NarK2. This is particularly striking as it can be reasoned that the intracellular nitrate pool for conversion to nitrite in a strain expressing NarK2 will be lower due to its much higher Km for nitrate. However, the potential inhibitory role of nitrite on transport by NarK1 may be also relevant. It is plausible that NarK2 has a higher affinity for nitrite than NarK1. All these data are consistent with the proposal that NarK1 is predominantly a nitrate/xH+ symporter, whereas NarK2 is predominantly a nitrate/nitrite antiporter. However, it also appears that when the appropriate concentrations of nitrate or nitrite are high enough, each domain can transport both substrates. Expression of separate NarK1 and NarK2 partially rescues the growth defects observed upon expression of either half alone. In this case it can be envisaged that NarK1 acts largely as a nitrate uptake protein, due to its lower Km, and NarK2 acts mainly to transport nitrite to the periplasm.
Analysis of NarK-like transporters in other organisms has generally revealed functional overlap, for example in E. coli (Clegg et al., 2002) and T. thermophilus (Ramirez et al., 2000). In contrast, a study of Pseudomons aeruginosa (Sharma et al., 2006) revealed that only NarK2 was capable of supporting nitrate-dependent anaerobic growth. However, the growth curves detailed by Sharma et al. (2006) bear striking similarities to our observations in P. denitrificans. A P. aeruginosa strain expressing only NarK2 was capable of growth with a similar profile to a wild-type strain but with a reduced growth rate and reaching an OD660∼80% of the wild-type after 12 h. Contrastingly, NarK1 alone was only able to support growth to ∼40% of the OD660 of the wild-type after 12 h. It would be interesting to see if a P. aeruginosa strain expressing NarK1 has an increased lag phase and would begin to grow exponentially if observed over an increased incubation period.
In order to examine the potential transport mechanisms employed by NarK1 and NarK2, two arginine residues were mutated to leucine. Mutation of the equivalent arginines in Aspergillus nidulans NrtA to lysine dramatically increased the Km for nitrate but did not abolish transport, suggesting a positively charged residue to be essential at these positions (Unkles et al., 2004). Although the exact role of these residues in anion transport remains elusive, it is likely that the mechanism of substrate transport by NarK1 and NarK2 is similar to that observed for other nitrate transporters (Unkles et al., 2004; Jia and Cole, 2005). As NarK2 is a member of the MFS family of 12-helix transporters, it is envisaged that a single re-orientating binding site favours the preferential binding of nitrate from the periplasmic side and nitrite from the cytoplasmic side.
A transporter that contains an active NarK1 fused to an inactive NarK2 has very similar properties to NarK1 alone. Contrastingly, when a transporter containing an inactive NarK1 fused to an active NarK2 is expressed, it has a greater affinity for nitrate than NarK2 alone. This observation may be a result of activation of the inactive NarK1 or an increase in affinity of the NarK2 domain in the fusion. It is possible that the inactive NarK1 is activated by the proximity of NarK2; fusion of the two domains may result in a structural change that enables transport even in the absence of the essential arginine residues. Activation of an inactive transporter as part of a fusion has been demonstrated in the LacS lactose transporter from Streptococcus thermophilus (a member of the MFS family of transporters; Geertsma et al., 2005); creation of fusions between inactive and active LacS transporters resulted in partial complementation of the inactive transporter.
NarK2 may become more active when fused to inactive NarK1 as the inactive transporter may stabilize NarK2 in a conformation in which it has a greater affinity for nitrate. This is consistent with a model proposed by Wood et al. (2002) in which NarK1 is responsible for initial nitrate uptake (dependent upon proton motive force) until a sufficient nitrite gradient has been established, when the nitrate/nitrite antiport mechanism of NarK2 predominates. It can be envisaged then in such a situation, NarK1 becomes locked in an ‘inactive’ conformation, which increases the affinity of NarK2 for nitrate and allows this transporter to dominate. The mechanism by which such a change in NarK2 affinity may arise is, as yet, unknown. However, it has been demonstrated that phosphorylation of T101 in Arabidopsis thaliana CHL1 (a nitrate transporter) dramatically increases the affinity of the transporter for nitrate (Liu and Tsay, 2003). It is therefore not unprecedented for relatively minor changes in a transporter to bring about dramatic changes in substrate affinity. Although at present it is impossible to conclude which of the hypotheses are correct (either inactive NarK1 becoming active or NarK2 having an increased affinity for nitrate), it is clear that the fusion of the two domains dramatically alters the ability of at least one of the component domains to transport nitrate. The fact that neither NarK1 nor NarK2 can fully complement a mutation in narK, even when fused to the reciprocal inactive transporter, suggests that the entire active transporter is required for maximal nitrate and/or nitrite transport.
Absence of the 24-transmembrane domain NarK protein from P. denitrificans can be overcome by expression of either 12-transmembrane domain transporter from E. coli, NarK or NarU. This demonstrates that both E. coli proteins are capable of transporting nitrate and nitrite. It also suggests that it is the combined transport activity and not the particular arrangement of two transporters that is essential in P. denitrificans. This does raise the obvious question of why such a fused transporter system is employed when it seems to have little benefit over a single transporter. It is possible that NarK1 and NarK2 have evolved to be specialized for a particular part of the nitrate/nitrite antiport process, with the fusion ensuring a stoichiometric ratio and a proximity that contributes to the overall activity of the protein. There is evidence (Hickman and Levy, 1988; Zottola et al., 1995; Yin et al., 2000) that MFS family members are capable of dimerization. A fused transporter would provide a mechanism by which this dimerization can be controlled, perhaps preventing homodimerization while ensuring heterodimerization. The existence of two transporters that can become more specialized in a particular function may allow denitrification to occur over a wider range of environmental conditions. Indeed, the presence of multiple NarK-like transporters is not unique to P. denitrificans and no conclusive evidence exists for why such a system provides a significant benefit over a single transporter. It has, however, been suggested that in E. coli, NarU provides an advantage over NarK during slow, nutrient-limiting growth (Clegg et al., 2006).
Data presented here, combined with earlier studies (notably Moir and Wood, 2001; Wood et al., 2002), suggest an overall model for nitrate and nitrite transport in P. denitrificans. Nitrate import is initiated by NarK1, which presumably acts as a nitrate/xH+ transporter. The imported nitrate is converted to nitrite by NarGHI. The accumulated intracellular nitrite may then play a role in reducing the activity of NarK1 and also activate nitrate/nitrite antiport by NarK2 that would dominate. It would seem disadvantageous for nitrate transport to proceed solely via a nitrate/proton symporter as it would require the use of inward proton flow, which could otherwise be utilized for synthesis of ATP. Fusion of the two transporters may allow a degree of cross-talk, contributing to their activity and regulation. It is notable that E. coli transporters can both initiate and maintain transport apparently as efficiently as the P. denitrificans NarK. Additionally, NarK2 alone can initiate transport. This suggests either that a nitrate/xH+ transporter is not absolutely required to initiate transport or that each transporter is capable of multiple modes of transport depending on the intracellular and extracellular environments. It is possible that the transport activities of NarK-like proteins are not as rigorously defined as previously believed.
DNA manipulations were performed by standard methods. Oligonucleotides were synthesized by Sigma-Genosys. Amplification by the polymerase chain reaction (PCR) used KOD DNA polymerase (from Thermococcus kodakaraensis) according to the supplier's instructions (Novagen). All constructs generated by PCR were confirmed to be correct by sequencing. All oligonucleotides used in this study are shown in Table S1.
Construction of strains
Initially, an unmarked deletion was generated in napD in a wild-type Pd1222 P. denitrificans strain. This was performed in a two-step process. First, the 5′ and 3′ flanking regions of napD were cloned and the kanR cassette inserted between them. This was cloned into pJQ200ks (gentamicin-resistant) which is incapable of replication in P. denitrificans. Chromosomal integrants in which double crossover events had replaced the napD ORF with the kanamycin-resistance cassette but lost the pJQ200ks backbone were selected as kanamycin-resistant gentamicin-sensitive strains. Correct integration of the cassette was confirmed by PCR screening. Second, the deletion was made unmarked using a construct in which the napD flanks were cloned into the pRVS1 vector (van Spanning et al., 1991), which is also incapable of replication in P. denitrificans. Single crossover events were selected via resistance to rifampicin (P. denitrificans), streptomycin (pRVS1) and kanamycin (napD::kanR ). This strain was then screened for a second crossover event in which the kanR cassette was removed via homologous recombination. This strain was selected essentially as described in van Spanning et al. (1991) and identified by the growth of kanamycin-sensitive white colonies in the presence of 200 μg ml−1 X-gal (5-bromo-4-chloro-3-indolyl-α-D-galactoside). Putative strains were confirmed to be correct by PCR screening (ΔnapD). Unmarked deletions in narK (ΔnarK) and narK2 (ΔnarK2) were created in the ΔnapD strain by an identical strategy using the relevant constructs. Full details of construct generation and strategy can be found in the Supporting information. Constructs and strains are listed in Table S2.
Cloning of P. denitrificans narK domains
The narK ORF was amplified from P. denitrificans genomic DNA using AG3 and AG4, digested with EcoRI and HindIII and ligated into EcoRI/HindIII-digested pEG276. Regions corresponding to the ORFs of narK1 and narK2 were cloned in the same manner using oligonucleotides AG3 and AG27 for narK1 and AG24 and AG4 for narK2. Hexahistidine-tagged versions of the above constructs were created in the same manner but using antisense oligonucleotide AG45 for narK and narK2 and AG46 for narK1.
Construction of narK mutants
Clones corresponding to the ORFs of narK, narK1 and narK2 were also generated in pTZ19R using the above oligonucleotide and restriction site combinations. Inverse PCR was then used to generate a number of mutations using the following oligonucleotide combinations on the appropriate clone: R66L – AG163 and AG164, R269L – AG157 and AG158, R520L – AG159 and AG160 and R736L – AG165 and AG166. The mutagenized ORFs were then subcloned into the EcoRI and HindIII sites of pEG276 for expression in P. denitrificans. Hexahistidine-tagged versions of the mutant ORFs were generated as described for their wild-type counterparts. The mutants generated are detailed in Table S2.
Cloning of E. coli transporters
The narK ORF was amplified from E. coli genomic DNA using oligonucleotides AG131 and AG132. This PCR product was digested with EcoRI and BamHI and cloned into the EcoRI and BamHI sites of pTZ19R. Inverse PCR was used to mutagenize the internal HindIII site in narK using oligonucleotides AG135 and AG136. The narK ORF was then subcloned into the EcoRI and HindIII sites of pEG276 to generate pEG276-coliNarK. The narU ORF was amplified from genomic DNA using oligonucleotides AG133 and AG134 and the PCR product digested with EcoRI and HindIII and cloned into the EcoRI and HindIII sites of pEG276 to generate pEG276-coliNarU.
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are detailed in Table S2. P. denitrificans strains were grown in LB medium or in a defined mineral-salt medium (Robertson and Kuenen, 1983) supplemented with 20 mM succinate as a carbon and energy source. Aerobic growth was achieved in 5 ml of growth medium in 50 ml universals which were incubated in a rotary shaker at 250 r.p.m. and 37°C. Anaerobic growth was conducted in 300 ml of growth medium in 300 ml bottles, incubated stationary at 37°C. For anaerobic growths, cultures were supplemented with sodium nitrate (1–50 mM as appropriate). Anaerobic cultures were inoculated with 1% v/v of freshly grown aerobic overnight culture in LB and cell density determined at OD600. E. coli strains were grown in LB medium in 5 ml of aerobic cultures as described for Paracoccus strains. Antibiotic-resistant strains were supplemented with antibiotics at the following concentrations; rifampicin (50 μg ml−1), kanamycin (50 μg ml−1), spectinomycin (50 μg ml−1), streptomycin (20 μg ml−1), carbenicillin (100 μg ml−1) and gentamicin (20 μg ml−1). Growth on solid media used liquid growth medium supplemented with 1.5% bacteriological agar.
Analysis of extracellular nitrite
Cells were pelleted from anaerobic culture via centrifugation at 14 000 g for 1 min. Nitrite concentration in the medium was estimated colourimetrically by the method of Nicholas and Nason (1957).
Preparation of P. denitrificans extracts
Paracoccus denitrificans strains were grown in 1 l cultures of LB and harvested at 6000 g for 20 min. Cell pellets were re-suspended in 10 ml of 20 mM Tris pH 7.5 to which 1 mg ml−1 lysozyme, 75 μg DNaseI and 1/5 of a protease inhibitor tablet were added. This suspension was incubated on ice for 20 min. Each sample was French-pressed three times at 1000 psi. Cell debris was removed by centrifugation at 12 000 g for 30 min. The supernatant was centrifuged at 50 000 r.p.m. 2 h to collect the membranes, which were re-suspended in 5 ml of 20 mM Tris pH 7.5 and stored at −80°C.
Alternatively, P. denitrificans strains were grown aerobically in 50 ml of LB, or anaerobically in 50 ml of mineral-salt medium supplemented with 20 mM sodium nitrate, to an OD600 of ∼1 before harvesting at 6000 g for 10 min. Pellets were re-suspended in BugBuster (Novagen) at 0.2 g dry pellet ml−1 and incubated at room temperature with rocking for 30 min.
10 μl samples of lysate or 3 μl of membrane extracts containing equal protein concentrations (total 30 μg) were run on an SDS-PAGE gel for analysis. Western blots to detect hexahistidine tags were performed using a peroxidase conjugate of a monoclonal antipentahistidine antibody (Qiagen) according to the manufacturer's instructions. NarGH was detected using a polyclonal antibody raised in sheep which was detected using an anti-sheep alkaline phosphatase-conjugated secondary antibody. In each case the markers used were SeeBlue Plus 2 (Invitrogen).
MALDI-TOF/TOF analysis was conducted on purified NarK-His using an Applied Biosystems 4700 Proteomics Analyser at the University of York.
S.J.F., D.J.R., J.W.B.M. and A.D.G. gratefully acknowledge funding from Biotechnology and Biological Sciences Research Council Grant (BB/D523019/1). D.J.R. is the recipient of a Royal Society Wolfson Foundation Merit Award.