ABSTRACT
Rhodobacter capsulatus fixes atmospheric dinitrogen via two nitrogenases, Mo- and Fe-nitrogenase, which operate under different conditions. Here, we describe the functions in nitrogen fixation and regulation of the rcc00574 (cooA) and rcc00575 (cowN) genes, which are located upstream of the structural genes of Mo-nitrogenase, nifHDK. Disruption of cooA or cowN specifically impaired Mo-nitrogenase-dependent growth at carbon monoxide (CO) concentrations still tolerated by the wild type. The cooA gene was shown to belong to the Mo-nitrogenase regulon, which is exclusively expressed when ammonium is limiting. Its expression was activated by NifA1 and NifA2, the transcriptional activators of nifHDK. AnfA, the transcriptional activator of Fe-nitrogenase genes, repressed cooA, thereby counteracting NifA activation. CooA activated cowN expression in response to increasing CO concentrations. Base substitutions in the presumed CooA binding site located upstream of the cowN transcription start site abolished cowN expression, indicating that cowN regulation by CooA is direct. In conclusion, a transcription factor-based network controls cowN expression to protect Mo-nitrogenase (but not Fe-nitrogenase) under appropriate conditions.
INTRODUCTION
Biological nitrogen fixation, the reduction of dinitrogen (N2) to ammonia (NH3), is an ancient process that developed when fixed nitrogen became limiting on the early earth (1). N2 reduction is catalyzed by nitrogenases exclusively found in diazotrophic prokaryotes and absent in eukaryotes, some of which form symbiotic associations with diazotrophs (2). All diazotrophs possess molybdenum-dependent nitrogenases containing a unique iron-molybdenum cofactor (3). In addition to Mo-nitrogenase, some species harbor alternative Mo-independent nitrogenases containing either an iron-vanadium or an iron-only cofactor (4). V- and Fe-nitrogenases exhibit lower N2-reducing activities than Mo-nitrogenase but are especially important for diazotrophic growth under Mo-limiting conditions. Since nitrogenase-mediated ammonia production is a highly energy-requiring process, ammonium strictly inhibits expression of nitrogen fixation (nif) genes in most diazotrophs.
N2 reduction by Mo- and V-nitrogenases is strongly inhibited by carbon monoxide (CO) (5–10). It is assumed that CO blocks electron flow in Mo- and V-nitrogenases at different stages, because hydrogen production by Mo-nitrogenase is largely unaffected by CO, whereas V-nitrogenase-catalyzed H2 evolution decreases gradually with increasing CO concentrations (6). The effect of CO on Fe-nitrogenase activity has not yet been examined.
CO is naturally generated by volcanic activity and forest fires, as well as by cellular processes based on enzymes like heme oxygenase and FeFe-hydrogenase maturase (11, 12). Even Mo-nitrogenase produces CO by reduction of CO2, albeit at low rates (13, 14). On the other hand, CO is a substrate for Mo-nitrogenase, but its CO-reducing activity is very low (15). In contrast, CO is a much better substrate for V-nitrogenase (15, 16). To date, it is unknown to what extent, if at all, Fe-nitrogenase is able to reduce CO.
CowN (Rru_A3516) promotes diazotrophic growth of Rhodospirillum rubrum in the presence of CO (17). Expression of cowN is activated by RcoM (Rru_A3515), a heme-binding CO-responsive regulator of the LytTR family, which is encoded by a gene located immediately upstream of cowN (18). In addition to RcoM, R. rubrum possesses a second heme-binding CO-responsive regulator, CooA (Rru_A1431), which belongs to the Crp/Fnr family. CooA is essential for CO dehydrogenase gene activation but dispensable for cowN expression (19, 20).
Genes similar to R. rubrum cowN are widespread in bacteria (17). Many cowN genes are genetically linked to rcoM or cooA genes, suggesting that they are controlled by the respective CO-responsive regulator RcoM or CooA. Notably, cowN genes are restricted to bacteria that also contain the structural genes of Mo-nitrogenase, nifHDK, suggesting that CowN is specific for diazotrophs. In line with this assumption, cowN is linked to nif genes in some species, including Rhodobacter capsulatus (see below). Despite its frequent occurrence in diazotrophs, a functional correlation between CowN and N2 fixation has been shown only in R. rubrum (17).
Our objective was to analyze the functions and regulation of cowN (rcc00575) and cooA (rcc00574) in R. capsulatus, a photosynthetic alphaproteobacterium that is able to synthesize two nitrogenases, Mo-nitrogenase and Fe-nitrogenase (21). We provide genetic evidence that CowN specifically protects Mo-nitrogenase (but not Fe-nitrogenase) against CO inhibition. In addition, we unravel the regulatory cascade controlling cooA and cowN expression in response to the cellular nitrogen status and CO.
MATERIALS AND METHODS
Strains, plasmids, and growth conditions.The bacterial strains and plasmids used in this study are listed in Table 1. Methods for conjugational plasmid transfer from Escherichia coli S17-1 to R. capsulatus and for mutant selection were previously described (22–24). R. capsulatus strains were grown in modified Rhodobacter capsulatus V (RCV) minimal medium as described previously (25). For growth under a mixed N2-CO atmosphere, 3-ml cultures were placed in screw-cap 17-ml Hungate tubes. After exchange of air against N2, appropriate amounts of CO were added with a syringe. Subsequently, the cultures were incubated in the light, and growth was followed by measuring the optical density at 660 nm.
Bacterial strains and plasmids
Construction of R. capsulatus cooA and cowN mutant strains.R. capsulatus mutants were generated as previously described (24–27). Briefly, genes of interest were cloned into mobilizable suicide vectors before gentamicin (Gm) cassettes were inserted to disrupt the genes. The cooA gene (rcc00574) was disrupted by a Gm cassette cloned into a unique SmaI site, resulting in hybrid plasmid pMF7 (cooA::Gm). The coding sequence of cowN (rcc00575) was mutagenized using the primers 5′-GGATCCTCGCGCTGGGTCGCCTATTTCG-3′and 5′-GGATCCTGCCCGCCATCCGCGCCGCC-3′ to create a BamHI site (underlined). This BamHI site was used to insert a Gm cassette, resulting in hybrid plasmid pYP206 (cowN::Gm). Plasmids pMF7 and pYP206 were conjugationally transferred into R. capsulatus. Selection for Gm resistance and loss of the vector-encoded tetracycline (Tc) resistance identified potential R. capsulatus cooA (MF7) and cowN (YP206) mutants, which were verified by PCR (data not shown).
Construction of cooA-lacZ and cowN-lacZ reporter strains and β-galactosidase assays.R. capsulatus reporter strains containing a promoterless lacZ gene transcriptionally fused to promoters of interest were constructed as previously described (25, 26). Briefly, appropriate DNA fragments carrying the respective promoter and part of the coding region were cloned into vector pUC18. Subsequently, a cassette carrying a promoterless lacZ gene, a tetracycline resistance (Tcr) gene, and a transfer origin (oriT) from plasmid pYP35 was cloned into the SmaI site within cooA or the artificially created BamHI site within cowN (see above), resulting in reporter plasmids pYP208 (cooA-lacZ) and pYP209 (cowN-lacZ). In addition, a 278-bp SalI-BamHI fragment carrying the cowN promoter (including the CooA binding site) was inserted into the broad-host-range vector pBBR1MCS-4. Subsequently, the lacZ-Tc-oriT cassette from plasmid pYP35 was cloned into the BamHI site, resulting in reporter plasmid pYP188 (cowN-lacZ) (Fig. 1A). Plasmid pYP188 served as the template for site-directed mutagenesis to generate CooA binding site variants, resulting in pYP188-E, pYP188-P, and pYP188-EP. Reporter plasmids were conjugationally transferred into R. capsulatus. Selection for Tc resistance yielded strains carrying plasmids pYP208 and pYP209 chromosomally integrated by single recombination, whereas pYP188 and its derivatives were episomally maintained. The reporter strains were grown in the light until late exponential phase, and LacZ (β-galactosidase) activity was determined as described previously (28).
Mutational and expression analysis of cooA and cowN genes. (A) R. capsulatus strains AM164, MF7, and YP206 contain gentamicin (Gm) cassettes inserted into the PstI site within fdxD (rcc00573), the SmaI site within cooA (rcc00574), and an artificial BamHI site (BamHI*) within cowN (rcc00575) (see Materials and Methods). The arrowheads indicate the orientation of the Gm resistance gene. The solid squares indicate putative CooA binding sites. The nifH and fdxD genes are preceded by highly conserved RpoN binding sites (boxed N). Reporter plasmids pYP208 (cooA-lacZ) and pYP209 (cowN-lacZ), which are based on narrow-host-range vectors, were chromosomally integrated by single recombination events. Plasmid pYP188 (cowN-lacZ), which is based on a broad-host-range vector, is able to replicate in R. capsulatus. (B) The cooA promoter contains three overlapping sequences (GG-N10-GC) resembling minimal RpoN binding sites. (C) The cowN TSS was determined by 5′-RACE (see Materials and Methods). Putative −35 and −10 regions are underlined. The opposed arrows indicate the putative CooA binding site. To disrupt the CooA recognition sequence, site-directed base substitutions creating EcoRI and PstI sites were introduced. The Gm cassettes and lacZ genes are not drawn to scale.
Determination of the cowN transcription start site (TSS) by 5′-RACE.RNA isolation from R. capsulatus cultures grown under nitrogen-fixing conditions, and 5′ rapid amplification of cDNA ends (RACE) experiments were done as previously described (29, 30). Briefly, the adapter oligonucleotide 5′-GTCAGCAATCCCTAACgag (with uppercase and lowercase letters indicating desoxyribonucleotides and ribonucleotides, respectively) was ligated to tobacco acid pyrophosphatase (TAP)-treated RNA, and cDNA was synthesized by reverse transcription using the cowN-specific oligonucleotide 5′-CCGCGCCGCCAGCATCTCGC. Subsequently, cDNA served as a template for PCR amplification using the adapter and the cowN-specific oligonucleotides as primers. Finally, the DNA sequences of the amplification products were determined.
RESULTS
CowN promotes Mo-nitrogenase-dependent growth in the presence of CO.The R. capsulatus cooA (rcc00574) and cowN (rcc00575) genes are located upstream of the structural genes of Mo-nitrogenase, nifHDK, and the fdxD gene (Fig. 1A), raising the possibility that their products are functionally linked. FdxD promotes Mo-nitrogenase-dependent growth in the presence of oxygen (25). To determine the roles of CooA and CowN in nitrogen fixation in the presence of carbon monoxide (CO), ΔcooA and ΔcowN mutant strains were constructed (see Materials and Methods) (Fig. 1A).
First, we investigated the roles of CooA and CowN in Mo-nitrogenase-dependent growth. For this purpose, wild-type and mutant strains were grown in minimal medium lacking a fixed nitrogen source under different N2-CO gas mixtures (see Materials and Methods) (Fig. 2A). The medium used in these assays (RCV plus Mo) contained 10 μM molybdate, which is sufficient to repress synthesis of the Fe-only nitrogenase (21). Thus, any growth observed under these conditions could be ascribed to Mo-nitrogenase activity. In the absence of CO, the ΔcooA and ΔcowN strains grew like the wild type, demonstrating that CooA and CowN per se are dispensable for Mo-nitrogenase activity (Fig. 2A). In the presence of CO, however, diazotrophic growth of the ΔcooA and ΔcowN strains was abolished at CO concentrations above 0.5% and 0.25%, respectively, while the wild type grew at CO concentrations up to 4%. These observations demonstrate the importance of CooA and CowN for Mo-nitrogenase-dependent growth of R. capsulatus in the presence of CO. Diazotrophic growth of a strain lacking FdxD was slightly inhibited by CO compared to the wild type, but growth inhibition was less pronounced than that by the ΔcooA and ΔcowN strains (Fig. 2A).
Diazotrophic growth of R. capsulatus wild-type and mutant strains in the presence of CO. R. capsulatus strains were grown in RCV minimal medium under a mixed N2-CO atmosphere with the indicated CO concentrations in the headspace (see Materials and Methods). The strains used were the wild-type (B10S), ΔfdxD (AM164), ΔcooA (MF7), ΔcowN (YP206), ΔnifDK (BS85), ΔnifDK ΔcooA (BS85-MF7), and ΔnifDK ΔcowN (BS85-YP206) strains. (A and B) +Mo indicates addition of 10 μM molybdate to support the Mo requirement of Mo-nitrogenase and to repress Fe-nitrogenase. (C and D) −Mo indicates that molybdate was omitted from the medium to derepress Fe-nitrogenase. (B and D) To achieve growth with a fixed nitrogen source, 10 mM serine was added.
To determine the roles of CooA and CowN in Fe-nitrogenase-dependent growth, a strain lacking Mo-nitrogenase (ΔnifDK), a ΔnifDK ΔcooA mutant, and a ΔnifDK ΔcowN strain were analyzed essentially as described above (Fig. 2C). The only difference was that molybdate was omitted from the growth medium (RCV minus Mo) to induce expression of Fe-nitrogenase. The ΔnifDK parental strain grew diazotrophically at CO concentrations up to 1%, demonstrating that molybdate concentrations were low enough to relieve repression of Fe-nitrogenase genes. Notably, the wild type grew diazotrophically at CO concentrations up to 4%, indicating that the RCV-minus-Mo medium contained trace amounts of molybdate that were still sufficient for Mo-nitrogenase activity. The ΔnifDK ΔcooA and ΔnifDK ΔcowN double-mutant strains with N2 as the sole nitrogen source grew like the parental strain, ΔnifDK, suggesting that CooA and CowN do not influence Fe-nitrogenase-dependent growth in R. capsulatus.
To test the effect of CO under conditions independent of nitrogen fixation, wild-type and mutant strains were grown with a fixed nitrogen source, serine, at different CO concentrations (Fig. 2B and D). All strains grew comparably well at different CO concentrations, ruling out the possibility that strains lacking CooA, CowN, or FdxD were generally sensitive to CO.
Expression of cowN is induced by CO and inhibited by ammonium.In the presence of CO, CowN supported growth dependent on Mo-nitrogenase, whose expression is inhibited by ammonium (25, 31–34), leading us to ask whether cowN expression was regulated by CO and/or ammonium. To examine cowN regulation, we constructed an R. capsulatus reporter strain carrying a transcriptional cowN-lacZ fusion integrated into the chromosome (see Materials and Methods) (Fig. 1A).
The cowN-lacZ reporter strain was grown in minimal medium with serine or ammonium at different CO concentrations prior to determination of LacZ (β-galactosidase) activity (Fig. 3). R. capsulatus grows efficiently with serine as a nitrogen source, but in contrast to ammonium, serine does not inhibit nitrogen fixation. Expression of cowN was clearly induced by CO in both serine and ammonium cultures, but ammonium cultures showed much lower cowN expression than serine cultures at any given CO concentration. These findings are consistent with the observation that CowN promotes growth under nitrogen-fixing (ammonium-starved) conditions in the presence of CO (Fig. 2A).
Expression of cowN in response to CO and different nitrogen sources. Cultures of R. capsulatus wild type (B10S) carrying a chromosomally integrated transcriptional cowN-lacZ fusion were grown in RCV minimal medium under a mixed N2-CO atmosphere with the indicated CO concentrations in the headspace. The cultures were grown with 10 mM serine or 10 mM ammonium as a fixed N source. LacZ (β-galactosidase) activity is given in Miller units (28), with the results representing the means and standard deviations of at least three measurements.
CooA induces cowN expression in response to CO.To identify regulators controlling cowN expression, the cowN-lacZ fusion described above was introduced into R. capsulatus strains lacking the putative CO-responsive regulator CooA or the nitrogen-responsive regulators NtrC, RpoN, NifA1, NifA2, and AnfA. In the absence of ammonium, the primary nitrogen regulator, NtrC, activates transcription of nifA1, nifA2, and anfA (35–38). In turn, NifA and AnfA activate transcription of Mo- and Fe-nitrogenase genes, respectively, in concert with the nitrogen fixation-specific sigma factor RpoN (31, 39, 40).
Reporter strains were grown in the presence or absence of 0.5% CO, and cowN expression was determined (Fig. 4A). In the ΔcooA strain, cowN expression was strongly reduced compared to the wild type and no longer responded to CO, suggesting that CooA acts as the CO-responsive activator of cowN transcription. However, residual cowN expression, even in the absence of CooA, suggests that an additional, CO-independent promoter contributes to cowN expression. In the ΔntrC and ΔrpoN strains, cowN expression was lower than in the wild type but remained CO responsive. In contrast, transcription of Mo- and Fe-nitrogenase genes is completely abolished in strains lacking NtrC or RpoN (31, 38). Disruption of either nifA1 or nifA2 did not affect cowN expression, but expression was clearly reduced in the ΔnifA1 ΔnifA2 double mutant. Hence, NifA1 and NifA2 substitute for each other in cowN regulation. Notably, cowN expression was enhanced in the ΔanfA strain, suggesting that AnfA, the transcriptional activator of Fe-nitrogenase genes, acts as a repressor of cowN expression. Similarly, Azotobacter vinelandii AnfA has activator and repressor functions, depending on the cellular nitrogen status (41, 42). It is worth emphasizing that NifA-dependent cowN activation and AnfA-mediated cowN repression required CooA and thus appear to be indirect (see below).
Expression of cowN and cooA in R. capsulatus wild-type and mutant strains. R. capsulatus strains carrying chromosomally integrated transcriptional cowN-lacZ (A) or cooA-lacZ (B) fusions were grown in RCV minimal medium with or without 0.5% CO in the headspace. The parental strains used were the wild-type (B10S), ΔcooA (MF7), ΔntrC (PBK2), ΔrpoN (YP201), ΔnifA1 (YP203), ΔnifA2 (YP202), ΔnifA1 ΔnifA2 (YP203-YP202), ΔanfA (KS94A), and ΔfdxD (AM164) strains. LacZ (β-galactosidase) activity is given in Miller units (28), with the results representing the means and standard deviations of at least three measurements.
NifA activates cooA expression under nitrogen-fixing conditions.Maximal cowN expression required ammonium-starved conditions and the presence of CO (Fig. 3), and CO-dependent induction was mediated by CooA (Fig. 4A). This led us to ask whether cooA expression also responded to ammonium and/or CO. To answer this question, expression of a chromosomally integrated transcriptional cooA-lacZ fusion was examined in R. capsulatus wild-type and mutant strains lacking CooA, NtrC, RpoN, NifA1, NifA2, or AnfA (see Materials and Methods) (Fig. 1A).
Reporter strains were grown in the presence or absence of CO prior to measuring cooA expression (Fig. 4B). In contrast to cowN expression (Fig. 4A), however, cooA expression in general was much lower and did not respond to CO irrespective of the genetic background. In the wild type, cooA expression was lower than in the ΔcooA background, indicating that CooA represses its own synthesis. Maximal cooA expression required NtrC, RpoN, and either NifA1 or NifA2, while expression was repressed by AnfA, essentially as described above for cowN (Fig. 4A). Hence, cooA was coordinately expressed with Mo-nitrogenase genes, a finding that correlates well with the importance of CooA in Mo-nitrogenase-dependent growth (Fig. 2A).
Expression of cooA was not affected in a ΔfdxD strain (Fig. 4B), whereas cowN expression was clearly impaired in the mutant (Fig. 4A). Together, these findings indicate that FdxD affects cowN expression by a mechanism independent of CooA and, correspondingly, of NifA, which activates cooA expression. In conclusion, the CO inhibition of diazotrophic growth observed in the ΔfdxD strain (Fig. 2A) is likely to be due to reduced cowN expression.
The cooA promoter lacks canonical binding sites for NifA (TGT-N10-ACA) and RpoN (CTGG-N8-TTGC) present in most R. capsulatus nif promoters described previously (25, 34, 43). NifA may, however, bind less conserved sites or, alternatively, may bind at a great distance from the transcription start site (44, 45). It is worth noting that the cooA promoter contains three overlapping sequences resembling the minimal consensus sequence of RpoN binding sites (GG-N10-GC) (Fig. 1B), suggesting that the cooA promoter may very well be a direct target of RpoN and, correspondingly, of NifA. Irrespective of whether NifA directly or indirectly activates cooA expression, the involvement of a CO-responsive regulator, CooA, adds a further level of complexity to nif gene regulation in R. capsulatus (see below).
Expression of cowN depends on an inverted-repeat sequence (IRS) probably serving as a CooA binding site.The cowN TSS was mapped 22 bp upstream of the ATG start codon by 5′-RACE experiments (see Materials and Methods) (Fig. 1C). Two DNA sequences upstream of the TSS, CTGACA and TATGAAA, exhibit clear similarity to the R. capsulatus −35 and −10 regions, respectively (46), suggesting that cowN transcription depends on the housekeeping sigma factor RpoD.
The −35 region overlaps an IRS, TGTCA-N6-TGACA (Fig. 1C), that is similar to the proposed CooA binding site consensus sequence, tGTCg-N6-tGACa (where lowercase letters indicate less conserved nucleotides) (17). To examine the role of this IRS in cowN regulation, the broad-host-range plasmid pYP188 containing a transcriptional cowN-lacZ fusion was constructed (see Materials and Methods) (Fig. 1A). Subsequently, pYP188 derivatives with base substitutions in the distal IRS half (pYP188-E), the proximal IRS half (pYP188-P), or both halves (pYP188-EP) were generated.
R. capsulatus strains carrying pYP188 and its derivatives were grown in the presence or absence of CO prior to estimation of cowN expression (Fig. 5). Like the chromosomal cowN-lacZ fusion (Fig. 4A), cowN was strongly expressed in a CO-dependent manner from plasmid pYP188. Disruption of either half-site of the IRS was sufficient to abolish cowN expression, demonstrating that the IRS acts as a cis-regulatory element important for cowN expression. Base substitutions in the proximal IRS half may affect binding of both CooA and RNA polymerase, while mutations in the distal IRS half are likely to affect CooA binding only but leave RNA polymerase binding unaffected. We conclude that the IRS functions as a CooA binding site, supporting the view that CooA directly mediates CO-responsive activation of cowN in concert with RpoD.
Effects of CooA binding site mutations on cowN expression. R. capsulatus wild-type (B10S) strains carrying plasmids pYP188, pYP188-E, pYP188-P, and pYP188-EP were grown in RCV minimal medium with or without addition of 0.5% CO to the headspace. Plasmid pYP188 and its derivatives carry transcriptional lacZ fusions to the wild-type cowN promoter or promoter variants with mutated CooA binding sites (see Materials and Methods) (Fig. 1). LacZ (β-galactosidase) activity is given in Miller units (28), with the results representing the means and standard deviations of five measurements.
DISCUSSION
CO is a serious challenge for biological nitrogen fixation, as it inhibits nitrogenase-catalyzed N2 reduction (5, 7–10). In this study, CowN (Rcc00575) and CooA (Rcc00574) were identified as important players in N2 fixation by R. capsulatus in the presence of CO (Fig. 2). This is the first report comparing the effects of CO on Mo-nitrogenase- and Fe-nitrogenase-dependent growth, and we showed that CowN specifically protects Mo-nitrogenase against CO inhibition, while it does not support Fe-nitrogenase activity.
CowN-protected Mo-nitrogenase tolerates higher CO concentrations than Fe-nitrogenase, whereas unprotected Mo-nitrogenase (in a strain lacking CowN) is more sensitive than Fe-nitrogenase. CO tolerance of Fe-nitrogenase against up to 1% CO may result from CO reduction by the alternative nitrogenase. It is worth noting that V-nitrogenase exhibits much higher CO-reducing activity than Mo-nitrogenase (15, 16). Thus, CO reduction activity might be a general feature of alternative nitrogenases.
The mechanism by which CowN protects Mo-nitrogenase against CO inhibition is not understood. CowN-like proteins are small (∼100 amino acid residues) and do not encompass any known functional domain. However, we can exclude the possibility that CowN removes CO by a general mechanism, because CowN supported Mo-nitrogenase activity but did not protect Fe-nitrogenase (Fig. 2A and C). Possibly, Mo-nitrogenase protection against CO requires physical interaction with CowN. Mo-nitrogenase protection against another diatomic gas, O2, involves complex formation between Mo-nitrogenase and FeSII-type ferredoxins, also called Shetna proteins (47, 48). The fesII homolog of R. capsulatus, fdxD, forms part of the nifHDK-fdxD-cooA-cowN gene region, and its product, FdxD, promotes Mo-nitrogenase-dependent growth in the presence of oxygen (25). Thus, genes mediating protection of Mo-nitrogenase against the diatomic gases O2 and CO are clustered in R. capsulatus (Fig. 6A).
Models of CowN function (A) and cowN regulation (B). (A) Mo-nitrogenase protection gene cluster in R. capsulatus. CowN and FdxD protect Mo-nitrogenase (but not Fe-nitrogenase) against CO inhibition and O2 damage, respectively (reference 25 and this study). (B) Model of cooA and cowN regulation in response to nitrogen and CO. Green and red indicate activation and repression of nitrogen fixation genes, respectively. At low fixed-nitrogen concentrations (−N), NtrC is activated by phosphorylation. Phosphorylated NtrC (NtrC-P) activates RpoD-dependent promoters (boxed D). NifA1, NifA2, and AnfA activate RpoN-dependent promoters (boxed N). Expression of cooA is activated by NifA1 and NifA2 and repressed by AnfA and CooA. Upon coordination of CO, CooA activates cowN expression by binding an inverted-repeat sequence in the cowN promoter (solid square). For further details, see the text.
Four lines of evidence support the view that CooA functions as a CO-binding regulator directly activating cowN expression in R. capsulatus. First, regulatory genes in bacteria often cluster with their target genes, as is the case for cooA and cowN (Fig. 1A and 6A). Second, CooA exhibits clear similarity to the heme-containing CO-responsive regulator CooA, which activates CO dehydrogenase gene expression in R. rubrum but is dispensable for cowN expression in the bacterium (17, 19, 20). In particular, amino acid residues involved in heme coordination (corresponding to Pro2 and His77 in R. rubrum CooA) are conserved in R. capsulatus CooA. Third, mutations in the putative CooA binding site upstream of R. capsulatus cowN abolished activation of cowN expression in response to CO (Fig. 5). Fourth, disruption of the cooA gene strongly reduced cowN expression and CO-responsive induction (Fig. 4). Assuming that CooA functions as the direct activator of cowN expression, the nitrogen fixation phenotype observed for the ΔcooA strain can be attributed to lack of CowN. Notably, Mo-nitrogenase-dependent growth was slightly more tolerant of CO in the ΔcooA strain than in the ΔcowN mutant (Fig. 2A), suggesting that some cowN expression occurs independently of CooA, which is consistent with residual expression in the ΔcooA strain (Fig. 4A).
Figure 6B summarizes our current model of cooA and cowN regulation in response to nitrogen and CO. When ammonium is limiting, the primary nitrogen sensor, NtrC, activates transcription of nifA1, nifA2, and anfA (35, 36, 38). Either NifA1 or NifA2 is sufficient to activate transcription of nifHDK and cooA, while AnfA activates anfHDGK transcription (references 37 and 40 and this study). The presence of three putative RpoN binding sites in the cooA promoter (Fig. 1B) suggests that RpoN (in concert with NifA) directly binds the cooA promoter. Irrespective of whether cooA is directly or indirectly activated by NifA, cooA is coordinately induced with the Mo-nitrogenase genes. The cooA gene is the first NifA-activated gene coding for a transcriptional activator, thus adding a further level of control to the nitrogen-regulatory cascade in R. capsulatus. Upon coordination of CO, CooA activates cowN expression, indicating that NifA-mediated nitrogen regulation of cowN is indirect. In other words, nitrogen regulation of cowN occurs at the level of cooA transcription, while CO regulation of cowN is mediated by the CooA protein.
Disruption of anfA increased cooA expression, suggesting that AnfA represses cooA transcription and, subsequently, cowN expression (Fig. 4). As a consequence, AnfA limits CowN-mediated protection of Mo-nitrogenase against CO inhibition. At present, we do not understand the biological relevance of this control. We assume that repression of cooA by AnfA (and CooA itself) allows fine tuning of Mo-nitrogenase activity in response to CO.
In conclusion, the widespread occurrence of cowN-like genes in diazotrophs suggests that CowN-mediated protection of Mo-nitrogenase against CO inhibition is a common mechanism. Future studies are needed to identify the mechanism underlying Mo-nitrogenase protection by CowN. In addition to its role in nitrogen fixation, CowN may have yet unrecognized functions, as implied by significant cowN expression in the presence of ammonium (Fig. 3), a condition that strictly inhibits nif gene expression.
ACKNOWLEDGMENTS
We thank Johanna Roßmanith for help with 5′-RACE experiments to determine the cowN transcription start site.
This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG) (Ma 1814/4-1) to B.M.
FOOTNOTES
- Received 10 April 2014.
- Accepted 18 July 2014.
- Accepted manuscript posted online 28 July 2014.
- Address correspondence to Bernd Masepohl, bernd.masepohl{at}rub.de.
↵* Present address: Maria Fehringer, Max-Planck-Institut für Molekulare Biomedizin, Münster, Germany.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
