Role of the nifB1 and nifB2 Promoters in Cell-Type-Specific Expression of Two Mo Nitrogenases in the Cyanobacterium Anabaena variabilis ATCC 29413

ABSTRACT

Anabaena variabilis ATCC 29413 has one Mo nitrogenase that is made under oxic growth conditions in specialized cells called heterocysts and a second Mo nitrogenase that is made only under anoxic conditions in vegetative cells. The two large nif gene clusters responsible for these two nitrogenases are under the control of the promoter of the first gene in the operon, nifB1 or nifB2. Despite differences in the expression patterns of nifB1 and nifB2, related to oxygen and cell type, the regions upstream of their transcription start sites (tss) show striking homology, including three highly conserved sequences (CS). CS1, CS2, and the region just upstream from the tss were required for optimal expression from the nifB1 promoter, but CS3 and the 5′ untranslated region (UTR) were not. Hybrid fusions of the nifB1 and nifB2 upstream regions revealed that the region including CS1, CS2, and CS3 of nifB2 could substitute for the similar region of nifB1; however, the converse was not true. Expression from the nifB2 promoter region required the CS1, CS2, and CS3 regions of nifB2 and also required the nifB2 5′ UTR. A hybrid promoter that was mostly nifB2 but that had the region from about position −40 to the tss of nifB1 was expressed in heterocysts and in anoxic vegetative cells. Thus, addition of the nifB1 promoter region (from about position −40 to the tss of nifB1) in the nifB hybrid promoter supported expression in heterocysts but did not prevent the mostly nifB2 promoter from also functioning in anoxic vegetative cells.

IMPORTANCE In the filamentous cyanobacterium Anabaena variabilis, two Mo nitrogenase gene clusters, nif1 and nif2, function under different environmental conditions in different cell types. Little is known about the regulation of transcription from the promoter upstream of the first gene of the cluster, which drives transcription of each of these two large operons. The similarity in the sequences upstream of the primary promoters for the two nif gene clusters belies the differences in their expression patterns. Analysis of these nif promoters in strains with mutations in the conserved sequences and in strains with hybrid promoters, comprising parts from nif1 and nif2, provides strong evidence that each promoter has key elements required for cell-type-specific expression of the nif1 and nif2 gene clusters.

INTRODUCTION

Filamentous heterocystous cyanobacteria fix atmospheric nitrogen under oxic growth conditions in specialized cells called heterocysts that differentiate in a semiregular pattern in a filament. They comprise about 5 to 10% of the cells in filaments in cultures grown in an environment that is deficient in fixed nitrogen (1–5). Heterocysts are microoxic; they have a glycolipid layer that restricts oxygen diffusion into the cell, lack components of oxygen-evolving photosystem II, and have high levels of respiration, which together protect the oxygen-labile nitrogenase from oxygen (6–10). Anabaena variabilis ATCC 29413 (here referred to as A. variabilis) is unusual among heterocyst-forming cyanobacteria because it has three nitrogenases, each of which is expressed under different environmental conditions (reviewed in references 11 and 12). The heterocyst-specific Mo nitrogenase, encoded by the nif1 genes, is the most commonly found nitrogenase in cyanobacteria. This nitrogenase is made only in heterocysts in cultures grown in an oxic environment that is low in fixed nitrogen but with molybdate (13, 14). This Mo nitrogenase is expressed late in differentiation, after the heterocyst becomes microoxic (11). The second Mo nitrogenase, encoded by the nif2 genes, is expressed only in anoxic vegetative cells in an environment with molybdate that is low in fixed nitrogen (15). In an oxic environment that is low in fixed nitrogen and low in molybdate but with vanadate (V), A. variabilis synthesizes a heterocyst-specific V nitrogenase encoded by the vnf genes (13, 16). The organization of the nif1 and nif2 nitrogenase gene clusters in A. variabilis is very similar (12, 13, 17–19). The nif1 cluster of A. variabilis is similar to the nif clusters in other heterocystous cyanobacteria but is most similar to the sole nif cluster in Anabaena sp. strain PCC 7120. The nif1 genes are expressed only in heterocysts, even under anoxic growth conditions (17, 20).

We have shown that the heterocyst-specific nif1 genes in A. variabilis are primarily under the control of the promoter for nifB1, the first gene in the cluster, which drives transcription of a 15-kb region (21, 22). Additional heterocyst-specific but weak promoters in nifU1, nifE1, and a strong promoter upstream of hesA1 also contribute to transcription (22). RNA processing of some transcripts in the nif1, nif2, and vnf gene clusters leads to a greater abundance of transcripts for genes located immediately downstream of the RNA processing site. For nifH1, which encodes dinitrogenase reductase and is the most abundant nif1 transcript, this increase results from the stability of the transcript, which depends on a conserved stem-loop structure at the RNA processing site (22). NifB, which is essential, is involved in the synthesis of the FeMo cofactor of dinitrogenase (23, 24). Although nifB1 is expressed only in heterocysts and nifB2 is expressed only in anoxic vegetative cells (17), the regions upstream of the transcription start site (tss) of nifB1 and nifB2 are remarkably similar to each other and to conserved sequences (CS) upstream of nif genes in other cyanobacteria that have been shown to be important for nitrogenase expression (25).

The sequencing of a large number of genomes of nitrogen-fixing cyanobacteria has revealed the strong conservation not only of nitrogenase gene clusters but also of a gene called cnfR (formerly called patB) in both heterocystous and nonheterocystous cyanobacteria (25–29). CnfR is a regulatory protein with a C-terminal helix-turn-helix (HTH) DNA-binding domain and two N-terminal 4Fe-4S binding sites, similar to bacterial ferredoxins that may be involved in oxygen sensing (25–27). In Anabaena sp. PCC 7120, cnfR mutants have delayed heterocyst formation and greatly reduced nitrogen fixation (26, 27). A cnfR mutant of a nonheterocystous cyanobacterium, Leptolyngbya boryana, which fixes nitrogen under microoxic growth conditions, showed no nif transcription (28). We have recently shown in A. variabilis that CnfR1 and CnfR2 activate expression of nif1 and nif2, respectively, and are likely key mediators of the developmental control of nif1 and nif2 expression via the nifB1 and nifB2 promoters (29). In this work, we examine the role of the conserved sequences upstream of the transcription start sites of nifB1 and nifB2 in cell-type-specific expression of nifB1 and nifB2 and discuss the role of CnfR1 and CnfR2 in nifB1 and nifB2 cell-type-specific regulation.

RESULTS

Role of conserved sequences in expression of nifB1.

We aligned the regions upstream of nifB among many nitrogen-fixing cyanobacteria from about position −350 from the transcription start site (tss) to just downstream of the tss for nifB1 (30, 31) and for nifB2 (see Fig. S1 in the supplemental material). The alignment revealed a high degree of similarity among the sequences (Fig. S2); however, the upstream regions of the two copies of nifB in A. variabilis are not closely related to each other. The nifB1 upstream region is most closely related to similar regions of other heterocystous cyanobacteria, while the nifB2 upstream region is most closely related to the similar region of nifB in Chroococcidiopsis, a primitive coccoidal cyanobacterium (32) (Fig. S3). The nifB upstream regions in both heterocystous and nonheterocystous cyanobacteria include three conserved motifs (labeled CS1, CS2, and CS3, shown in green in Fig. 1) upstream of the tss and another duplicated conserved region between CS2 and CS3 (turquoise in Fig. 1). In order to determine the role of these conserved regions in expression of nifB1, fusions of various fragments upstream of nifB1, with or without these conserved regions, were cloned as transcriptional fusions to a promoterless lacZ reporter (Fig. 2A). The plasmids containing these fusions were integrated into the chromosome of A. variabilis using an internal fragment of the fructose transport operon (frtABC) such that single-crossover recombination between the plasmid and the chromosome produced an frtABC mutant unable to grow heterotrophically with fructose (21, 33). In contrast, a single crossover in the region upstream of nifB1 yielded a strain that was frtABC⁺ (strain BP682), providing a positive control, since the latter recombination event resulted in lacZ expression from an extensive chromosomal region upstream of nifB1, rather than the shorter cloned region (Fig. 2A). The positive control, strain BP682, and the strain with all the conserved regions, JJ33, gave the highest β-galactosidase activity under nitrogen starvation conditions, while the one with the shortest fragment (strain JJ31), lacking the conserved sequence sites CS1 and CS2, gave a very low level of activity of about 30 units (Fig. 2B), very similar to that for the negative-control strain, JJ53, with a promoterless lacZ, which also gave values of about 30 units (typical range, 25 to 35 units). A strain with a fragment lacking only CS1, JJ12, had a much lower level of β-galactosidase activity than the strain with CS1, CS2, and CS3 (JJ33) (Fig. 2B). Although the loss of CS1 decreased the levels of β-galactosidase, expression of lacZ was still heterocyst specific in strain JJ12 (Fig. 2C). A larger fragment that included more of the region upstream of the conserved sequence sites resulted in strain JJ15, which had a somewhat lower level of activity than the positive-control strain BP682 but had greatly decreased activity compared to strain JJ33 (Fig. 2B), likely because the region upstream of the conserved sequences has a negative effect on transcription of nifB1. To determine whether the long 5′ untranslated region (UTR) between the tss and the start of nifB1 is important for transcription, a strain with a promoter region fragment containing the upstream elements but lacking 300 bp between the tss and the start of nifB1, strain JJ91, was made. Loss of the sequence encoding the 5′ UTR of nifB1 in strain JJ91 led to a slight decrease in expression in medium without N but increased lacZ expression in medium with N (Fig. 2B). In cells grown in medium without N, lacZ expression was still heterocyst specific (Fig. 2C). This indicates that the 5′ UTR is not essential for expression of nifB1 but plays a role in repression of nifB1 in cells grown with fixed nitrogen. In summary, the tss and the three similar conserved motifs, CS1, CS2, and CS3, are sufficient for heterocyst-specific nifB1 expression.

FIG 1.

Alignment of the nifB1 and nifB2 promoter regions. Green or turquoise, conserved sequence sites; pink, conserved extended −10 promoter region; red, transcription start sites for nifB1 (30, 31) and nifB2 (Fig. S1). Vertical lines with numbers indicate fusions sites (FS) for the constructs shown in Fig. 2 to 7.

FIG 2.

Expression of nifB1 promoter fragments. (A) Map of nifB1 promoter fragments fused to lacZ. WT, wild type. (B) The β-galactosidase activity in the strains was measured 28 h after nitrogen step-down. Green bars, growth without fixed nitrogen; purple bars, growth with fixed nitrogen. The promoterless lacZ control strain, JJ53, had β-galactosidase activity of about 30 ± 3 units. (C) Fluorescence in cells of strains JJ12 and JJ91, grown under oxic conditions in the absence of fixed N for 28 h. Green, cleavage of fluorescein-β-d-galactopyranoside; red, autofluorescence, diminished in heterocysts; arrows, representative heterocysts.

The need for CS1 and CS2 was confirmed by creating mutations in CS1, CS2, or CS3, and each of the mutant sequences was fused to lacZ (Fig. 3A). The loss of either CS1 (strain JJ101) or CS2 (strain JJ102) reduced β-galactosidase activity substantially, with a further reduction in expression being seen for a strain with mutations in both CS1 and CS2 (strain SM78) (Fig. 3B). Loss of CS3 alone had no effect on expression (strain SM74), and the effects of combinations of mutations in CS1 and CS3 (strain SM76) and CS2 and CS3 (strain SM75) were similar to those of mutations in CS1 or CS2 alone (Fig. 3B). Mutations in all three conserved sequence sites (strain SM79) gave results similar to those obtained with mutations in both CS1 and CS2 (strain SM78) (Fig. 3B). Although the levels of activity in the strain with the loss of CS1 alone (strain JJ101) (Fig. 3B) were slightly lower than the levels in strain JJ12 (Fig. 2B), which lacks both CS1 and the upstream region, this may be due to the presence of the upstream nifB1 region in strain JJ101, which had reduced activity compared to that of a strain lacking the upstream region (compare strain JJ15 versus strain JJ33; Fig. 2B). In summary, both CS1 and CS2 were important for expression of nifB1, while CS3 was not, and loss of both CS1 and CS2 abolished lacZ expression. In contrast, the 5′ UTR was not required for lacZ expression; however, loss of this region resulted in increased expression in cells grown with fixed nitrogen compared to that for the control strain.

FIG 3.

Expression of nifB1 CS1, CS2, and CS3 mutation fragments. (A) Map of nifB1 promoter fragments fused to lacZ. (B) The β-galactosidase activity in the strains was measured 28 h after nitrogen step-down. In cultures grown with N, the number of β-galactosidase activity units was between 32 and 45 (data not shown).

Reciprocal fusions of nifB1 and nifB2 upstream regions.

To identify the region of the nifB1 upstream sequence that is necessary for heterocyst-specific expression of nifB1, we created reciprocal fusions of the nifB1 and nifB2 upstream regions at fusion sites (FS) 6 and 9 (see these junctions in Fig. 1) and an additional nifB1::nifB2 fusion at fusion site 10 (Fig. 4A). Fusion of the upstream region of nifB2 with its CS1, CS2, and CS3 sequences to the downstream region of nifB1 at junction 6, about 40 bp upstream of the nifB1 tss, in strain SM34 gave wild-type levels of heterocyst-specific expression of lacZ, while the fusion at junction 9, just downstream from the tss, in strain SM35 gave only background levels of activity (Fig. 4B). The fusion of the upstream region of nifB1 to nifB2 at junction 6, about 40 bp upstream of the nifB2 tss, in strain SM38, gave weak β-galactosidase activity that was not induced in the absence of fixed nitrogen (Fig. 4B) and was not visible in filaments by confocal microscopy (data not shown). In contrast, the fusion of the upstream region of nifB1 to nifB2 at junction 9, just downstream from the tss, in strain SM42 gave wild-type levels of heterocyst-specific expression of lacZ (Fig. 4B). Moving the junction of the upstream region of the nifB1::nifB2 hybrid to junction 10, downstream of the tss in strain SM33 (Fig. 1), resulted in higher levels of expression, but the level of expression in medium with N was also higher (Fig. 4B). These results indicated that the region from about position −40 to just past the tss of the nifB1 promoter was necessary for heterocyst-specific expression. In summary, the nifB2 region from CS1 to CS3 can replace the similar upstream region of nifB1 to drive heterocyst-specific expression of β-galactosidase. Replacement of the nifB1 UTR by the nifB2 region downstream of the tss, which is not similar to the same region of nifB1, did not alter the heterocyst-specific expression of β-galactosidase; however, this is perhaps not surprising because the 5′ UTR of nifB1 is not required for heterocyst-specific expression. The region from position −40 to the tss of nifB1 is required for oxic expression in heterocysts.

FIG 4.

Oxic expression of nifB1 and nifB2 promoter fragment fusions. (A) Map of nifB1 and nifB2 promoter hybrids fused to lacZ. (B) The β-galactosidase activity in the strains was measured 28 h after nitrogen step-down. Green bars, growth without fixed nitrogen; purple bars, growth with fixed nitrogen. (C) Fluorescence in strains SM43 and SM34, grown under oxic conditions in the absence of fixed N for 28 h. Green, cleavage of fluorescein-β-d-galactopyranoside; red, autofluorescence, diminished in heterocysts; arrows, representative heterocysts.

We also constructed fusions to determine whether the nifB1 region with the conserved sequences can function in place of the homologous region of nifB2 to drive expression of lacZ in vegetative cells 6 h after nitrogen step-down under anoxic conditions (Fig. 5A). The nifB1 region upstream of CS1 at junction 1 fused to the downstream region of nifB2 in strain SM46 gave good expression of lacZ at 6 h after nitrogen step-down under anoxic conditions (Fig. 5B). In contrast, loss of the nifB2 CS1 by fusion of nifB1 to nifB2 at junction 2 in strain SM56 reduced expression by about 80%, as did the loss of the nifB2 CS1 and CS2 in the hybrid of nifB1 to nifB2 at junction 4 in strain SM49 (Fig. 5B). A hybrid of nifB1 to nifB2 at junction 6, between CS3 and the extended −10 region in strain SM38, completely abolished lacZ expression at 6 h after nitrogen step-down under anoxic conditions (Fig. 5B). Anoxic expression of lacZ in vegetative cells of strain SM56 with the nifB1 CS1 region was visible but much reduced compared to that in strain SM46 with the nifB2 CS1 region (Fig. 5C). While the conserved region upstream of the nifB2 tss is important for expression under anoxic conditions 6 h after nitrogen step-down, other regions are also required. A strain with a hybrid of nifB2 to nifB1 at junction 10, strain SM36, showed very low levels of β-galactosidase activity compared to strain BP770, which has the wild-type nifB2 promoter region (Fig. 5B), indicating that a sequence downstream of junction 10 is required for nifB2 expression. In summary, strong expression from the nifB2 promoter under anoxic conditions in vegetative cells requires the region from CS1 to CS3 from nifB2, and it requires the 5′ UTR of nifB2.

FIG 5.

Anoxic expression of nifB1 and nifB2 promoter fragment hybrids. (A) Map of nifB1 and nifB2 promoter fragment hybrids fused to lacZ. (B) The β-galactosidase activity in the strains was measured 6 h after nitrogen step-down under anoxic conditions. Green bars, growth without fixed nitrogen; purple bars, growth with fixed nitrogen. (C) Fluorescence in strains SM46 and SM56, grown 6 h after nitrogen step-down under anoxic conditions. Green, cleavage of fluorescein-β-d-galactopyranoside; red, autofluorescence. The numbers in the lower right indicate relative exposure levels (exp.) for the fluorescein fluorescence.

Expression from a hybrid nifB2-nifB1-nifB2 promoter.

Since the conserved sequences of the nifB2 upstream region function in place of the homologous region of nifB1 (strain SM34) (Fig. 4B and C) and the sequence encoding the 5′ UTR of the nifB1 transcript is not essential for heterocyst-specific expression (strain JJ91) (Fig. 2B and C), we created a fusion strain, SM54, with a hybrid promoter that comprised the nifB2 region upstream of fusion site 6 and downstream of fusion site 9 with the nifB1 region from junction 6 to junction 9 (Fig. 6A). Strain SM54 had high levels of β-galactosidase activity in heterocysts of filaments grown under oxic conditions without N for 28 h (Fig. 6B and C). β-Galactosidase activity was also detected in strain SM54 grown under oxic conditions with N (Fig. 6B), but since heterocysts are generally absent in the presence of fixed N under oxic and anoxic conditions, the activity was in vegetative cells and therefore represents some loss of nitrogen control in these cells. Strain SM54 also had high levels of β-galactosidase activity in vegetative cells grown under anoxic conditions for 6 h in the absence of fixed nitrogen (Fig. 6B and C), and these cultures showed no expression under anoxic conditions in the presence of fixed nitrogen (Fig. 6B). Strain SM55, which contained a hybrid of the promoter region of nifB2 to nifB1 at junction 7, also showed high levels of β-galactosidase activity in cells grown under oxic conditions for 28 h in the absence of fixed nitrogen (Fig. 6B) and showed repression of lacZ expression in the presence of fixed nitrogen. Under anoxic conditions there was only background β-galactosidase activity for strain SM55, which lacks the sequence for the 5′ UTR of nifB2 that is required for anoxic expression from the nifB2 promoter (strain SM36) (Fig. 5B). Replacement of the 3′ region of nifB1 at junction 9 in strain SM55 with the nifB2 region to create strain, SM71 resulted in high levels of β-galactosidase activity in cells grown for 28 h in the absence of fixed nitrogen (Fig. 6B) but also showed weak repression of lacZ expression in the presence of fixed nitrogen (Fig. 6B), indicating that the nifB2 region from junction 9 into the nifB2 gene was responsible for the loss of nitrogen control in strain SM71 compared to that in strain SM55, which had more stringent nitrogen control. Strain SM71 had high levels of β-galactosidase activity, similar to strain SM54, in cells grown under anoxic conditions for 6 h in the absence of fixed nitrogen (Fig. 6B). We then determined that replacement of the region just downstream from the tss (junction 8 to junction 9) of nifB1 with the region from nifB2 (strain SM72) had no effect on expression under either oxic growth conditions for 28 h or anoxic growth conditions for 6 h. Thus, it appears that the region from position −40 to the nifB1 tss is necessary and sufficient for expression in heterocysts but that this region from position −40 also functions in anoxic vegetative cells, as long as the nifB2 promoter regions upstream of junction 6 and downstream of junction 9 are also present.

FIG 6.

Expression of promoter fragment hybrids with the nifB1 −10 region. (A) Map of nifB1 and nifB2 promoter fragment hybrids fused to lacZ. (B) β-Galactosidase activity was measured in strains grown under oxic conditions for 28 h after nitrogen step-down. β-Galactosidase activity was measured in strains grown under anoxic conditions for 6 h after nitrogen step-down. Green bars, growth without fixed nitrogen; purple bars, growth with fixed nitrogen. (C) Fluorescence in strain SM54, grown under oxic conditions for 28 h after nitrogen step-down (left two panels) or 6 h after nitrogen step-down under anoxic conditions (right two panels). Green, cleavage of fluorescein-β-d-galactopyranoside; red, autofluorescence, diminished in heterocysts; arrows, representative heterocysts.

Requirement for sequences in the coding region of nifB2.

While the region downstream of the tss of nifB1 is not required for oxic gene expression (strain JJ91) (Fig. 2B), the same was not true for nifB2. Strain SM36, with contained a hybrid of nifB2 to nifB1 at junction 10 (Fig. 5B), had low levels of β-galactosidase activity, indicating that a region downstream of the tss of nifB2 is important for expression. Therefore, we analyzed expression of lacZ in strains in which the 3′ end of the region fused to lacZ extended into the nifB1 or nifB2 coding regions (Fig. 7A). Oxic expression of lacZ under the control of the promoter region from position −40 to the tss of nifB1 was not greatly affected by changing the 3′ end of the lacZ fusion from nifB2 (in strains SM54 and SM67) to nifB1 (in strain SM69) (Fig. 7B). However, the nifB2 promoter region failed to drive oxic expression of lacZ in strain SM65 (Fig. 7B). Under anoxic nitrogen-fixing conditions, only strains BP770 and strain SM54, in which lacZ was fused to a region extending 126 bp into the nifB2 coding region, supported lacZ expression. Strain JL2, in which lacZ was fused to a region extending only 45 bp into the nifB2 coding region, showed very low levels of expression of lacZ under anoxic conditions. Strain SM65, lacking the +126 region of nifB2, and strain SM67, lacking the entire nifB2 upstream region, had no anoxic lacZ activity. Thus, it appears that there are sequences in the coding region of nifB2 but not that of nifB1 that are important for gene expression under anoxic conditions in vegetative cells.

FIG 7.

Expression of promoter fragment hybrids with altered 3′ ends. (A) Map of nifB1 and nifB2 promoter fragment hybrids fused to lacZ. (B) β-Galactosidase activity was measured in strains grown under oxic conditions for 28 h after nitrogen step-down. β-Galactosidase activity was measured in strains grown under anoxic conditions for 6 h after nitrogen step-down. Green bars, growth without fixed nitrogen; purple bars, growth with fixed nitrogen.

DISCUSSION

Transcription of nifB, the first gene in the large cluster of genes that is responsible for the synthesis of nitrogenase, requires an activator protein, CnfR (originally called PatB) (26–29), and also depends on highly conserved sequences upstream of the nifB transcription start site (25). An interaction between CnfR and these conserved nifB sequences has been implicated in the nonheterocystous cyanobacterium Leptolyngbya boryana (25), and direct binding of CnfR to the conserved sequences upstream of nifB was shown for CnfR from the unicellular nitrogen-fixing strain Cyanothece sp. strain ATCC 51142 (34). Thus, it seems clear that the conserved regions upstream of the tss of nifB are important for their activation by CnfR. Anabaena variabilis ATCC 29413 is unique among well-studied nitrogen-fixing cyanobacteria in having two distinct nif systems that are cell type specific (12, 17). We have examined the role of the upstream regions of nifB1 and nifB2 in the cell-type-specific activation of the nif1 genes in heterocysts by CnfR1 and of the nif2 genes in anoxic vegetative cells by CnfR2 (29). Despite the similarity of the conserved regions upstream of the tss of nifB1 and nifB2 (Fig. 1), the expression of lacZ from promoter regions containing the nifB1 versus the nifB2 conserved sequences was different.

High levels of expression of lacZ from the nifB1 upstream region required the first two conserved sequence sites, CS1 and CS2, but not CS3. Tsujimoto et al. (25) drew very different conclusions from their data on the importance of the homologs of CS1, CS2, and CS3 of L. boryana for transcriptional activation of nifB by CnfR. Using Synechocystis sp. strain PCC 6803 for heterologous expression of CnfR from L. boryana, they showed that neither CS1 nor CS2 from L. boryana was required for activation of that nifB promoter by CnfR. In fact, in the absence of CS1 and CS2, nifB expression was increased, suggesting that CS1 and CS2 repress nifB expression. Further, while the loss of CS3 had no effect on nifB1 expression in A. variabilis, CS3 was essential for expression of nifB from L. boryana in Synechocystis sp. PCC 6803 (25). To further complicate the interpretation of the role of CS1, CS2, and CS3 in nifB expression, Balassy and Zhang have provisionally shown, also using Synechocystis sp. PCC 6803 for heterologous expression of cnfR and nifB from Cyanothece sp. strain 51142, that the CS1, CS2, and CS3 sites are all important for CnfR-dependent expression of nifB (34). An important difference in the contradictory results cited above is that both groups used a non-nitrogen-fixing organism, Synechocystis sp. strain PCC 6803, to study the role of the conserved sequences, while our studies were done in the native host. It may be that other transcription factors positively or negatively affect the ability of CnfR to activate the nifB promoter from these conserved sequence sites.

The region containing CS1, CS2, and CS3 upstream of nifB2 functioned well in place of the CS1, CS2, and CS3 region upstream of nifB1 to drive heterocyst-specific expression of β-galactosidase. In contrast, strong expression from the nifB2 promoter under anoxic conditions in vegetative cells specifically required the CS1 to CS3 region from nifB2. Expression from the nifB2 promoter not only required the nifB2-specific conserved sequences but also required the 5′ UTR and sequences between positions +45 and +126 in the coding region of nifB2. In contrast, expression from the nifB1 upstream region required neither the 5′ UTR of nifB1 nor any sequences in the nifB1 coding region. Substitution of the 5′ UTR of nifB2 for the 5′ UTR of nifB1 did not affect nifB1 expression under growth conditions without N but resulted in some loss of nitrogen control. The importance of the 5′ UTR in nitrogen control can be clearly seen in the results from the hybrid promoter experiments (Fig. 6). Under aerobic conditions, the hybrid promoter is activated in heterocysts by CnfR1 and does not function in aerobic vegetative cells. CnfR1 acting on this hybrid promoter, like the nifB1 promoter, requires the nifB1 5′ UTR for good nitrogen control. In the absence of the 5′ UTR of nifB1 (in the hybrid promoters), there is a loss of nitrogen control (compare the results for strains SM55 and SM72). In contrast, under anaerobic conditions in vegetative cells, expression of the hybrid promoter is driven by CnfR2. The 5′ UTR of nifB2 is required for activation by CnfR2, and it provides nitrogen control (compare the results for strains SM55 and SM72).

Both the nifB1 and nifB2 promoters have strikingly conserved sequences upstream of the transcription start site that form a canonical extended −10 promoter (TGNTATAAAT) (35), but they lack a recognizable −35 region. They represent a type 2 promoter, which typically requires a transcription factor and may use an alternative sigma factor (36). While the region that includes the nifB2 conserved sequence sites CS1, CS2, and CS3 functioned in place of the similar region of nifB1 for heterocyst-specific expression, the region from position −40 to the tss of nifB1 could not be replaced by the similar region from nifB2. The hybrid promoter in strain SM54, which is mostly the nifB2 promoter with only the region from position −40 to the tss of nifB1 in place of the similar region of nifB2, functioned well in heterocysts; thus, it appears that the region from about position −40 to the tss of nifB1 is necessary and sufficient for expression in heterocysts. However, this region from position −40 to the tss of nifB1 also functioned in anoxic vegetative cells, as long as the nifB2 upstream CS1 to CS3 regions and the 5′ UTR downstream of the nifB2 tss were also present. The requirement for the region from position −40 to the tss of nifB1 implies that an alternative sigma factor could bind specifically to this region; however, if this is the case, then that sigma factor must also be present in anoxic vegetative cells since the hybrid promoter works in both cell types.

The specificity in the regulation of nifB2 compared to nifB1 extends beyond the stringency of the nifB2 upstream regulatory sequences to include the specificity of the regulatory proteins. A model that summarizes our current knowledge and that has a hypothetical transcription factor that may act in the region from position −40 to the tss of nifB2 is shown in Fig. 8. We have shown previously that CnfR2 cannot activate expression of the nifB1 promoter even when CnfR2 is made in heterocysts under the control of the cnfR1 promoter (29). Therefore, it seems likely that CnfR2 cannot bind to the conserved sequences of nifB1. The fact that CnfR2 cannot activate nifB1 in anoxic vegetative cells supports the hypothesis that CnfR2 does not bind the nifB1 upstream region. In addition, there may be a transcription factor (transcription factor 2 [TF2] in Fig. 8A) that is present in anoxic vegetative cells but not in heterocysts that works in conjunction with CnfR2 to activate nifB2 expression. This model is supported by the fact that when CnfR2 is made in heterocysts under the control of the cnfR1 promoter, CnfR2 cannot activate expression of nifB2 (29). The hybrid promoter in strain SM54, which is mostly the nifB2 promoter with only the nifB1 region from position −40 to the tss, was activated in anoxic vegetative cells by CnfR2. This indicates that the nifB1 region from position −40 to the tss in the hybrid promoter had no effect on the ability of CnfR2 to bind and activate from the mostly nifB2 promoter. In contrast, CnfR1 is not specific for the nifB1 promoter. If CnfR1 is made from the cnfR2 promoter in anoxic vegetative cells, it can activate expression of the nifB2 promoter (29), suggesting that CnfR1 binds both to the nifB2 conserved sequence sites and to the nifB1 conserved sequence sites. In anoxic vegetative cells, CnfR1 made from the cnfR2 promoter activates expression of both nifB1 and nifB2 (unpublished data). If CnfR1 can activate nifB2, why does it not do so in heterocysts? One possible explanation is that heterocysts lack a transcription factor (TF2 in Fig. 8A) that is required for nifB2 activation by either CnfR1 or CnfR2. Since the strain SM54 hybrid promoter, which is mostly the nifB2 promoter with only the nifB1 region from position −40 to the tss, was activated in heterocysts by CnfR1, the region from position −40 to the tss is implicated as the site that normally restricts nifB2 expression to anoxic vegetative cells. Replacement of only this region of the nifB2 promoter with the similar regions from the nifB1 region resulted in a promoter that was activated by CnfR1 in heterocysts, while it maintained its ability to be activated by CnfR2 in anoxic vegetative cells (Fig. 8). This implies that the putative additional transcription factor, found only in anoxic vegetative cells, binds to the region from position −40 to the tss of nifB2 to drive nifB2 transcription. The ability of CnfR2 to activate the hybrid promoter in strain SM54, which has the region from position −40 to the tss of nifB1, implies either that no additional transcription factor is required for promoters with this nifB1 region (i.e., in the wild-type nifB1 promoter and in the hybrid promoter in strain SM54) or that this region binds a different transcription factor that is present in both heterocysts and vegetative cells and that this transcription factor interacts with both CnfR1 and CnfR2. For simplicity, the model in Fig. 8 shows no transcription factor specific for the region from position −40 to the tss of nifB1, although if there is such a factor for nifB2, there may also be one for nifB1. The location of the binding site indicates that such a transcription factor may be an alternative sigma factor.

FIG 8.

Model for regulation of nifB1 and nifB2 by CnfR1 and CnfR2. (A) Anoxic vegetative cell. (Promoter construct 1) CnfR2, made from the cnfR2 promoter only under anoxic conditions, probably with another transcription factor (TF2), activates nifB2 in anoxic vegetative cells. (Promoter construct 2) CnfR2 activates the hybrid strain SM54 promoter that is primarily nifB2 with only the 40-bp region upstream of the tss of nifB1 (shown as a blue region). (Promoter construct 3) The cnfR1 gene is not expressed in vegetative cells, so CnfR1 does not activate expression of nifB1. (B) Heterocyst. (Promoter construct 1) CnfR1, made from the heterocyst-specific cnfR1 promoter, activates nifB1 in heterocysts. (Promoter construct 2) CnfR1 activates the hybrid strain SM54 promoter that is primarily nifB2 with only the 40-bp region upstream of the tss of nifB1 (shown as a blue region), confirming that CnfR1 activates and likely binds to the nifB2 upstream region. (Promoter construct 3) The cnfR2 gene is not expressed in heterocysts and does not activate expression of nifB2. CnfR1 can activate the nifB2 promoter but requires TF2, which is absent in heterocysts. β-gal, β-galactosidase.

In A. variabilis, as well as in L. boryana (25) and in Cyanothece (34), there are interactions between highly conserved sequences upstream of nifB1 and nifB2 with their cognate activator proteins, CnfR1 and CnfR2, respectively. These interactions induce expression of each nifB gene. Since the nifB1 and nifB2 promoters are the primary promoters for the large nif1 and nif2 gene clusters (21), respectively, these protein/DNA interactions largely drive nif gene expression. The cell-type-specific expression of CnfR1 in heterocysts and CnfR2 in anoxic vegetative cells is important in determining the sites of expression of nif1 and nif2. However, the conserved sequences upstream of the tss of nifB1 and nifB2 also play a role in cell-type-specific expression, and other transcription factors, possibly sigma factors acting in the region from position −40 to the tss, may also have a function.

MATERIALS AND METHODS

Strains and maintenance conditions.

Strains of A. variabilis FD, a derivative of A. variabilis ATCC 29413 that can grow at 40°C (37), were maintained on agar-solidified Allen and Arnon (AA) medium (38) supplemented, when appropriate, with 5 mM NH₄Cl, 10 mM N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid (TES), pH 7.2, 40 to 50 μg ml⁻¹ neomycin sulfate (Nm), or 3 μg ml⁻¹ each of spectinomycin (Sp) and streptomycin (Sm). Strains were grown photoautotrophically in liquid cultures in an 8-fold dilution of AA medium (AA/8) containing 5 mM NO₃⁻ or in AA/8 supplemented with 5 mM NH₄Cl and 10 mM TES, pH 7.2, at 30°C with illumination at 100 to 120 μE m⁻² s⁻¹. Antibiotics, when used in liquid, included Nm (5 μg ml⁻¹) and Sp (0.5 μg ml⁻¹).

Construction of strains.

The details of the construction of all plasmids are described in Table S1 in the supplemental material, and the primers used in the construction of plasmids are shown in Table S2. Promoter fragments of various lengths upstream of nifB1 were made by PCR amplification using the primers listed for each plasmid. These nifB1 promoter fragments were inserted between the BglII/SmaI sites in the vector pBP609 to create various P_nifB1::lacZ fusions that were conjugated into wild-type strain FD as previously described (39). Single recombinants harboring the P_nifB1::lacZ fusions were selected using an Nm^r Sm^r Sp^r strain, followed by screening for exconjugants that recombined in the frtBC region, resulting in the loss of heterotrophic growth of the strain on fructose in the dark (21). To create strains containing mutations in CS1 (strain JJ101) and CS2 (strain JJ102) of the nifB1 promoter, PCR fragments were amplified using primer pair pnifBL-6/pnifB25-R or pnifB26-R for the upstream fragment and primer pair pnifB25-L or pnifB26-L/pnifB-R for the downstream fragment. The products were digested with BglII/XhoI (upstream fragment) or XhoI/SmaI (downstream fragment) and triply ligated into the BglII/SmaI sites of pBP609 to create P_nifB1::lacZ fusions with mutations in CS1 or CS2, and these plasmids were conjugated into wild-type strain FD followed by screening for exconjugants that recombined in the frtBC region, as described above. Mutants containing deletions in the CS3 region and in multiple CS regions (strains SM74 to SM79) were created in a manner similar to that described above for strains JJ101 and JJ102, except that the upstream and downstream PCR fragments for each strain were fused by PCR, sequenced, and then cloned into the BglII/SmaI site of pBP609.

The P_nifB2::lacZ fusions were constructed by PCR using primers pnifB2-L1/p2nifB-R(BP) (BP770) or primers pnifB2-L1/pnifB2-R1 (JL2), and these nifB2 promoter fragments were fused to lacZ by insertion between the BglII/SmaI sites of pBP744. The nifB2::lacZ fusions were sequenced to verify that they contained no mutations, plasmids were transferred to strain FD as described above, and PCR was used to verify that the fusions in the resulting strains, BP770 and JL2, had recombined by single crossover into the frtBC region.

The nifB1::nifB2 and nifB2::nifB1 hybrid promoter fusions were created by PCR amplification of nifB1 and nifB2 fragments (containing complementary sequences for fusion) using the primers listed for each plasmid (Table S1), followed by fusion PCR to make the full-length fragment. These fragments were ligated into the BglII/SmaI sites of pBP744, sequenced, and inserted into strain FD by conjugation as described above to create various nifB1::nifB2::lacZ and nifB2::nifB1::lacZ fusions that were verified by PCR. The nifB2::nifB1::nifB2 fusions (strains SM54, SM71, and SM72) were created in a similar manner. PCR amplification of initial fragments required the use of existing promoter fusions as the templates; both the primers and the templates used for each plasmid are listed in Table S1. These full-length fragments were ligated into the BglII/SmaI sites of pBP744, sequenced, and inserted into the frtBC region of FD by conjugation to create the nifB2::nifB1::nifB2::lacZ fusions.

Acetylene reduction assays.

For acetylene reduction assays of cultures grown under anoxic conditions, 1.0 ml of acetylene gas was added to Hungate tubes 30 min prior to the end of the 6-h anoxic growth period. Samples (250 μl) of headspace gas were removed via a gas-tight needle/syringe and injected into a Shimadzu gas chromatograph equipped with a 6-ft Poropak N column. The column temperature was 75°C.

β-Galactosidase assays.

Oxic cultures were grown for about 10 generations in AA/8 with 5 mM NaNO₃ or with 5 mM NH₄Cl and 10 mM TES, pH 7.2, and then diluted 1:100 in the same medium and grown to an optical density at 720 nm (OD₇₂₀) of 0.1 to 0.2. These young cultures were washed and then diluted with AA/8 to an OD₇₂₀ of 0.05 to 0.1 to achieve nitrogen (N) step-down or to an OD₇₂₀ of 0.035 to 0.045 to continue growth with nitrogen (5 mM NH₄Cl, 10 mM TES, pH 7.2). Three, 2-ml biological replicates of each culture were grown for 28 h at 30°C with shaking and illumination at 100 to 120 μE m⁻² s⁻¹ in 12-well microtiter plates prior to the β-galactosidase assays. For anoxic nitrogen step-down experiments, cells were grown under oxic conditions in the light with shaking in AA/8 with 5.0 mM fructose, 5.0 mM NH₄Cl–10 mM TES, pH 7.2, to an OD₇₂₀ of about 0.5 to 0.6. Cells were washed with AA/8 and resuspended in AA/8 with 10 mM fructose and 10 μM dichlorophenyldimethylurea (DCMU; to inhibit oxygen evolution from photosystem II) to an OD₇₂₀ of about 0.4. Three, 8-ml biological replicates were aliquoted into 16-ml Hungate tubes and flushed with dinitrogen for 10 min. The cultures were then incubated for 6 h at 30°C with illumination at 100 to 120 μE m⁻² s⁻¹, and induction of nitrogenase activity was verified by acetylene reduction assays. β-Galactosidase assays were performed in 96-well flat bioassay microtiter plates using 250 μl of sample for at least 3 biological replicates, with quadruple technical replicates being used for each biological replicate. OD₇₂₀ measurements were determined for each well. After incubation of the plate at 30°C for 10 min, 25 μl of a 10× permeabilization buffer (0.5 M NaPO₄, pH 7.4, 1.0% Sarkosyl, 1.6 mg/ml ortho-nitrophenyl-β-galactoside [ONPG]) was added to each well. Upon development of a yellow color, 50 μl of 1 M NaCO₂ was added to stop the reaction. OD₄₂₀ and OD₆₆₅ measurements were taken to determine the amount of ortho-nitrophenyl (ONP; OD₄₂₀) and to correct for chlorophyll and light scattering from permeabilized cells (OD₆₆₅). Calculations were performed using the following equation, developed empirically: nitrogenase activity = 1,000 × {[OD₄₂₀ – (1.58 × OD₆₆₅)]/[OD₇₂₀ × time of assay (in minutes)]}. The values from the quadruple technical replicates were averaged for each of the three biological replicates. These three averages for the biological replicates were used to calculate the average, and the standard deviation values are shown in the graphs as error bars.

In situ localization of β-galactosidase.

Cells, induced under either oxic or anoxic growth conditions as described above for the β-galactosidase assays, were washed twice with water and fixed for 15 min at 25°C with 0.04% glutaraldehyde. Cells were washed twice with water to remove the glutaraldehyde and incubated in the dark at 37°C for 30 min with 100 μM 5-dodecanoyl-aminofluorescein di-β-d-galactopyranoside in 25% dimethyl sulfoxide. Excess substrate was removed by washing twice with water, and cells were resuspended in antifade Vectashield mounting medium (catalog number H-1000; Vector Laboratories). The cells were visualized on a Zeiss LSM700 confocal microscope using a plan Apochromat 63× (numerical aperture, 1.4) oil differential inference contrast M27 objective. Expression of lacZ in cells was visualized using excitation and emission wavelengths (from an argon ion laser) specific for the detection of fluorescein fluorescence, while cyanobacterial autofluorescence from phycobiliproteins was visualized using excitation and emission wavelengths specific for the detection of rhodamine fluorescence.

RNA isolation and 5′ RACE.

RNA was isolated using the Tri Reagent (Sigma) as previously described (21) and subjected to DNase digestion (Turbo DNA-free kit; Ambion). 5′ rapid amplification of cDNA ends (RACE) was performed as described previously (40) with the following modifications. A total of 5 to 10 μg of RNA was treated with DNase, followed by ethanol precipitation, and suspended in 44 μl RNase-free water. The sample was treated with 40 U of RNA 5′-polyphosphatase (Epicentre, Madison, WI). Next, 150 pmol of the RNA adapter RNAoligo09 (29) was added, and the sample was extracted with phenol-chloroform-isoamyl alcohol and then with chloroform, followed by ethanol precipitation. The pellet was resuspended in 14 μl of water, heated to 90°C for 5 min, and ligated to the adapter overnight at 17°C using T4 single-stranded RNA ligase (New England BioLabs). The ligated RNA was extracted with organic solvents (as described above) and ethanol precipitated with 2 pmol of cDNA primers, resuspended in 20 μl of water, and reverse transcribed using SuperScript III reverse transcriptase (Invitrogen) according to the recommended protocol using the following primer for nifB2: nifB2-TAP2. PCR was performed using left primer P1 oligo B and primer nifB2-TAP4. The PCR band was excised and sequenced to determine the sequence of the 5′ end.