Regulation of Nitrogenase Gene Expression by Transcript Stability in the Cyanobacterium Anabaena variabilis

Abstract

The nitrogenase gene cluster in cyanobacteria has been thought to comprise multiple operons; however, in Anabaena variabilis, the promoter for the first gene in the cluster, nifB1, appeared to be the primary promoter for the entire nif cluster. The structural genes nifHDK1 were the most abundant transcripts; however, their abundance was not controlled by an independent nifH1 promoter, but rather, by RNA processing, which produced a very stable nifH1 transcript and a moderately stable nifD1 transcript. There was also no separate promoter for nifEN1. In addition to the nifB1 promoter, there were weak promoters inside the nifU1 gene and inside the nifE1 gene, and both promoters were heterocyst specific. In an xisA mutant, which effectively separated promoters upstream of an 11-kb excision element in nifD1 from the downstream genes, the internal nifE1 promoter was functional. Transcription of the nif1 genes downstream of the 11-kb element, including the most distant genes, hesAB1 and fdxH1, was reduced in the xisA mutant, indicating that the nifB1 promoter contributed to their expression. However, with the exception of nifK1 and nifE1, which had no expression, the downstream genes showed low to moderate levels of transcription in the xisA mutant. The hesA1 gene also had a promoter, but the fdxH gene had a processing site just upstream of the gene. The processing of transcripts at sites upstream of nifH1 and fdxH1 correlated with increased stability of these transcripts, resulting in greater amounts than transcripts that were not close to processing sites.

INTRODUCTION

Nitrogen fixation is characteristic of many cyanobacteria, including both filamentous and unicellular strains. In many filamentous cyanobacteria, specialized cells called heterocysts, which develop in response to starvation for fixed nitrogen, are responsible for nitrogen fixation under aerobic growth conditions. Heterocysts have a thick cell wall layer, lack oxygen-producing photosynthesis, and have high rates of respiration, which together provide a micro-oxic environment that supports the activity of the oxygen-labile enzyme nitrogenase (1–3). The nitrogen that is fixed in heterocysts moves to vegetative cells, which in turn provide fixed carbon to heterocysts to support nitrogen fixation. Heterocyst differentiation is a complex developmental process that occurs initially in about 10 to 15% of cells in a filament. The cells that differentiate do not occur randomly in the filament, and differentiation of adjacent cells into heterocysts is relatively uncommon. The differentiation of a heterocyst depends on a regulatory cascade that not only directs the development of a vegetative cell into a heterocyst, but is also important for the establishment and maintenance of heterocysts in a semiregular pattern in the filaments (3–5). The spacing of heterocysts ensures that all cells in the filament maintain a constant supply of both fixed nitrogen and fixed carbon. Intercellular communication between heterocysts and vegetative cells, which requires specific junction proteins and also appears to depend on intercellular channels, is critical for the maintenance of the filament (6–8).

The filamentous heterocystous cyanobacterium Anabaena variabilis is unique among the well-studied cyanobacteria in that it has three nitrogenases that function under different environmental conditions (reviewed in reference 9). The primary nitrogenase is the heterocyst-specific Mo nitrogenase encoded by the nif1 genes (10, 11); however, there is also an alternative heterocyst-specific V nitrogenase, encoded by the vnf genes, that is expressed only when Mo is limiting (10, 12). The second Mo nitrogenase, encoded by the nif2 genes (13–15), functions under strictly anoxic conditions in vegetative cells and heterocysts. The genes for all three nitrogenases are tightly repressed in cells grown with a source of fixed nitrogen, such as ammonium or nitrate.

The expression of the nitrogenase genes is a late step in the differentiation of heterocysts, since nitrogenase activity requires that the heterocyst be micro-oxic (9). The synthesis and assembly of nitrogenase is a complex process requiring the activities of at least a dozen proteins. Most of the proteins are encoded in diazotrophic bacteria in large nif operons that may include multiple genes under the control of a single promoter. In proteobacteria, the nif gene operons include nifHDK, comprising the structural genes for nitrogenase; nifBQ, whose products are involved in Fe-Mo cofactor assembly; nifUVSM, involved in Fe-S cluster formation; and nifENX, whose products serve as scaffolding proteins for the assembly of the nitrogenase complex (16, 17). The promoter for each of these operons is thought to control the timing of expression and the amount of protein that is optimal for a functional nitrogenase. The best-understood nif systems are in the proteobacteria, where nitrogenase genes are under the control of the Ntr global nitrogen regulatory system, which induces transcription of the nif regulators NifA and NifL (18). In conjunction with the activator protein NifA, the sigma-54 RNA polymerase, in the absence of oxygen and fixed nitrogen, transcribes the nif genes (18). In cyanobacteria, there is no homologue of nifA or nifL, so the regulation is different from that in the proteobacteria. Although the global nitrogen-regulatory protein NtcA is required for nitrogen fixation in heterocyst-forming cyanobacteria, its role may be indirect, since it is also required for heterocyst formation (19). Cyanobacteria also lack a sigma factor that is specifically associated with global nitrogen regulation; however, SigE, appears to be important, but not absolutely required, for expression of the nif genes in Anabaena sp. strain PCC 7120 (20).

An active nitrogenase requires the products of many genes, several of which are required for synthesis and insertion of the FeMo cofactor that is found in most nitrogenases (reviewed in reference 21). In cyanobacteria, these genes include, in a single cluster, nifB, fdxN, nifS, nifU, nifH, nifD, nifK, nifE, nifN, nifX, and nifW, as well as hesAB and fdxH and four other open reading frame (ORFs) within the cluster (Fig. 1A). The nifHDK operon is interrupted by an 11-kb excision element that is removed from the chromosome of heterocysts by XisA late in development to allow transcription of nifHDK (22–24). Primarily on the basis of Northern blot analysis, the nif cluster in Anabaena spp. has been reported to minimally comprise several operons: nifB-fdxN-nifS-nifU (25, 26), nifHDK (27–29), hesAB (30), and fdxH (31). In addition, nifVZT (32) comprises a second nif gene cluster in Anabaena spp. Apparent transcription start sites have been identified upstream of nifB (25, 26, 33–35), nifH (26, 36), hesA (30, 35), and fdxH (31). Although the putative nifH transcription start site has been identified several times, we were unable to drive transcription of lacZ using a DNA fragment that included the entire niU1-nifH1 intergenic region of A. variabilis, and we showed that the 5′ end of the nifH1 transcript in A. variabilis is a processed transcript rather than a transcription initiation site; thus, there is no promoter upstream of nifH1 (33). In contrast, nifB1 has a promoter that initiates transcription, and there is a promoter within the coding region of nifU1 (33), which has been verified by transcriptome sequencing (RNA-seq) (35).

FIG 1.

Map of the nif1 cluster and lacZ fusions. (A) Map of the nif1 cluster of A. variabilis without the 11-kb excision element, whose location is shown by an arrow. The solid rightward arrows indicate transcription start sites (tss) upstream of nifB1, in nifU1, in nifE1, and upstream of hesA1, while the dashed arrows indicate transcription processing sites upstream of nifH1, in ava3935, and upstream of fdxH, as determined previously (33) and in this study. (B) β-Galactosidase activities for strains containing fusions of four different regions upstream of nifE1 to a promoterless lacZ. In JU479 and JU480, only the 500-bp or 900-bp region, respectively, was fused to lacZ, while in JU483 and JU490, the entire chromosomal region upstream of nifE1 with (JU483) or without (JU490) the nifB1 promoter was fused to lacZ. Cells were either grown with fixed nitrogen (+N) or starved for 24 h without fixed nitrogen (−N) prior to the β-galactosidase assay. WT, wild type. (C) Diagram showing the two recombination events for the lacZ fusion strains, resulting from either a crossover in the putative nifE1 promoter region, leading to the entire chromosomal region upstream of nifE1 driving expression of lacZ, or a crossover in the frtBC internal fragment of the frtABC operon, leading to a truncated 500-bp or 900-bp region upstream of nifE1 driving expression of lacZ.

We show here that the nif1 cluster in A. variabilis from nifB1 through nifW1 depends primarily on the nifB1 promoter but that weak promoters in nifU1 and in nifE1 contribute to transcription. Further, all three of these promoters provide heterocyst-specific gene expression. While hesA1 has its own promoter, its transcription also depends to some extent on the nifB1 promoter, as does the fdxH1 transcript, which is a processed transcript rather than a primary transcript.

MATERIALS AND METHODS

Strains and growth conditions.

A. variabilis strain FD, a derivative of A. variabilis ATCC 29413, can grow at 40°C. FD and its derivatives were maintained on agar-solidified Allen and Arnon (AA) medium (37) supplemented, when appropriate, with 5 mM NH₄Cl; 10 mM TES (N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid), pH 7.2; 25 to 40 μg ml⁻¹ neomycin sulfate; 3 μg ml⁻¹ each of spectinomycin and streptomycin; or 5 μg ml⁻¹ erythromycin. The strains were grown photoautotrophically in liquid cultures in an 8-fold dilution of AA medium (AA/8) 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 microeinsteins m⁻² s⁻¹. Antibiotics, when used, included neomycin (5 μg ml⁻¹), erythromycin (1 μg ml⁻¹), and spectinomycin (0.5 μg ml⁻¹).

For growth assays, the optical density at 720 nm (OD₇₂₀) was measured during exponential growth over three cell divisions, which required that healthy, growing cells be diluted to an OD₇₂₀ of about 0.02 in appropriate medium (2.0 ml of diluted culture in each well of a 12-well plate) at the start of the experiment. Growth was observed to be exponential up to an OD₇₂₀ of about 0.2 and then slowed as the cells became light limited by self-shading. Growth was measured every 3 h, and the exponential-growth data were used to calculate the growth rate (μ) [ln(N₂) − ln(N₁)/(t₂ − t₁,), where N is the OD₇₂₀ and t is the time in days]. Growth experiments were done using at least triplicate cultures.

Construction of strains.

The xisA mutant was constructed as follows. Plasmid pBP645 was constructed by inserting a 3.5-kb MscI/KpnI fragment from pAAWY19720, containing the xisA region, between the KpnI/HincII sites in pUC18Nm. A 757-bp EcoRV-ScaI fragment was removed from pBP645 by self-ligation, resulting in the deletion of a part of the coding region of the xisA gene. A 6-kb SacI fragment of pRL2948a was cloned into the SacI site in pBP668, creating plasmid pBP669. The xisA deletion strain, BP669, was created by introducing the mobilizable plasmid pBP669 into strain FD by conjugation. Single recombinants were selected by erythromycin resistance (Em^r), followed by isolation of double-recombinant mutants with the deleted copy of xisA, using sacB with sucrose selection (38) as described previously (39).

Fusions of possible promoter fragments to lacZ were made by cloning various PCR fragments, amplified from FD DNA, into the BglII/SmaI sites in pJU411, pBP609, or pBP639 as shown in Table 1. Plasmids pJU411 and pBP609 have an internal fragment of the fructose transport genes, frtABC (40), that allows screening for single recombinants at the frtABC locus, which result in a mutant that cannot use fructose. Plasmid pBP639 lacks the frt region and is used to force a cross in the cloned promoter fragment. The promoter fragments were sequenced to verify that they contained no mutations. Recombination of these plasmids with promoter fragments fused to lacZ in the frt region of A. variabilis by single crossover following conjugation (41) resulted in strains with the same names as the plasmids. Strains resulting from a single crossover in the frt region were identified by screening for colonies that were unable to grow in the dark with fructose. Strains resulting from a single crossover in the nif1 region were screened to determine whether they had a Nif⁻ phenotype by their inability to grow on AA agar plates lacking a source of fixed nitrogen. Figure 1C shows an example of the creation of two strains by different single-crossover recombination events with the same plasmid.

TABLE 1.

Strains and plasmids

Strain or plasmid	Relevant characteristicsa	Source or reference
Strains
FD	A. variabilis ATCC 29413 wild-type parent strain	55
BP640	pBP640 integrated into the nifU region of strain FD via single recombination	This study
BP641	pBP641 integrated into the nifUH region of strain FD via single recombination	This study
BP642	pBP642 integrated into the nifK region of strain FD via single recombination	This study
BP643	pBP643 integrated into the nifE region of strain FD via single recombination	This study
BP644	pBP644 integrated into the nifEN region of strain FD via single recombination	This study
BP669	pBP669 integrated into the chromosome of strain FD via double recombination, creating a deletion in the coding region of xisA	This study
BP682	pBP682 integrated into the nifB promoter region of strain FD via single recombination	This study
BP683	pBP683 integrated into the nifK region of strain FD via single recombination	This study
BP692	pBP692 integrated into the nifNX region of strain FD via single recombination	This study
BP754	pBP754 integrated into the frtBC region of strain FD via single recombination	This study
BP755	pBP755 integrated into the frtBC region of strain FD via single recombination	This study
BP756	pBP756 integrated into the frtBC region of strain FD via single recombination	This study
JU466	nifB1 deletion mutant of FD	33
JU479	pJU479 integrated into the frtBC region of strain FD via single recombination	This study
JU480	pJU480 integrated into the frtBC region of strain FD via single recombination	This study
JU483	pJU480 integrated into the nifE region of strain FD via single recombination	This study
JU490	pJU480 integrated into the nifE region of strain JU466 via single recombination	This study
Plasmids
pAAWY19720	6.2-kb library clone of A. variabilis containing the xisA region; Cmr	JGI
pBP285	Kmr Nmr cassette in a polylinker (C.K.3 [56]) with a transcriptional terminator	40
pBP609	Tc cassette PCR amplified using pBR322-L/pBR322-R and inserted between the BglII/SmaI sites of pJU411	This study
pBP639	Tc cassette PCR amplified using pBR322-L/pBR322-R and inserted between the BglII/SmaI sites of pJU410	This study
pBP640	1,422-bp nifU region; made with primers nifUH-L(BP)/nifUH-R(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP641	669-bp nifUH intergenic region; made with primers nifH-L(BP)/nifH-R(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP642	1,496-bp nifK region; made with primers nifK-L(BP)/nifK-R(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP643	1,492-bp nifE region; made with primers nifE-L1(BP)/nifE-R1(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP644	1,509-bp nifEN region; made with primers nifE-L2(BP)/nifE-R2(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP645	3.5-kb MscI/KpnI fragment of pAAWY19720 containing the xisA region; inserted between the KpnI/HincII sites in pUC18Nm; Kmr	This study
pBP668	757-bp EcoRV-ScaI fragment removed by self-ligation of pBP645, deleting a part of the coding region of the xisA gene; Kmr	This study
pBP669	6-kb SacI fragment of pRL2948a cloned into the SacI site on pBP668; Kmr Cmr Emr	This study
pBP682	2,316-bp nifB promoter region; made with primers pnifB-L(BP)/pnifB-R(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP683	1,244-bp nifK region; made with primers pnifK2-L(BP)/pnifK2-R(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP692	1,240-bp nifNX region; made with primers pnifN2-L(BP)/pnifN2-R(BP); inserted into the BglII/SmaI sites of pBP639	This study
pBP754	462-bp nifE internal-fragment region; made with primers nifE-L3 (BP)/nifE-R1 (BP); inserted into the BglII/SmaI sites of pBP609	This study
pBP755	487-bp nifE internal-fragment region; made with primers nifE-L4/nifE-R1 (BP); inserted into the BglII/SmaI sites of pBP609	This study
pBP756	536-bp nifE internal-fragment region; made with primers nifE-L5/nifE-R1 (BP); inserted into the BglII/SmaI sites of pBP609	This study
pJU410	Vector with a promoterless lacZ for promoter fusions and an internal frtABC fragment for recombination; Kmr	33
pJU411	1.8-kb frtBC PCR fragment used as region of homology for recombination; inserted into the HindIII site of pJU410	33
pJU479	621-bp nifE upstream region; made with primers nifE-L2/nifE-R10; inserted into the BglII/SmaI sites of pJU411	This study
pJU480	867-bp nifE upstream region; made with primers nifE-L3/nifE-R10; inserted into the BglII/SmaI sites of pJU411	This study
pRL2948a	Source of mobilization site, oriT, and sacB gene, which confers sucrose sensitivity; Cmr Emr	C. P. Wolk
pUC18	Cloning vector; Apr	57
pUC18Nm	1-kb HincII fragment of pBP285 containing a Kmr Nmr cassette; inserted between the DraI/SspI sites in pUC18; Kmr	This study

^aCm^r, chloramphenicol resistance; Tc^r, tetracycline resistance; Ap^r, ampicillin resistance; Sm^r, streptomycin resistance; Km^r, kanamycin resistance; Sp^r, spectinomycin resistance; Em^r, erythromycin resistance; Nm^r, neomycin resistance.

RNA isolation and analysis.

FD and BP669, an xisA mutant incapable of removing the 11-kb excision element that interrupts nifD1 (22), was grown in AA/8 supplemented with 5 mM NH₄Cl and 10 mM TES, pH 7.2, for approximately 10 generations. Fifty-milliliter cultures at an OD₇₂₀ of 0.2 to 0.3 were washed free of fixed nitrogen and inoculated into 500 ml AA/8. These cultures were grown for 24 h with aeration to induce heterocysts and nitrogen fixation. Cells were removed at 24 h for the 0-h time point for RNA isolation to measure transcript decay. Then, 100 μg ml⁻¹ rifampin was added to the cultures to inhibit new transcription. At 10, 20, 30, 45, and 60 min after the addition of rifampin, 50 ml of cells was quickly chilled by pouring over ice in 2 50-ml centrifuge tubes, harvested, and frozen. RNA was isolated using Tri-Reagent (Sigma) as previously described (33) and subjected to DNase digestion (Turbo DNA-free kit; Ambion).

RT-qPCR was used to determine the quantities and decay rates of nifB1, nifS1, nifU1, nifH1, nifD1, nifK1, nifE1, nifN1, nifX1, nifW1, hesA1, and fdxH transcripts. cDNA was made from 10 to 40 ng of RNA (from strain FD or BP669, the xisA mutant) using Bio-Rad iScript Reverse Transcription Super Mix (catalog no. 170-8841) in a 10-μl reaction mixture. cDNA was diluted 1:20 to 0.05 to 0.2 ng μl⁻¹. Quantitative PCRs (qPCRs) were done using SsoAdvanced SYBR Green SuperMix (catalog no. 1725261) in a 10-μl reaction volume and contained 0.2 ng to 0.8 ng cDNA, 5 pmol of gene-specific primers (Table 2), and 1× reaction supermix. The following gene-specific primers were used: rnpB, qrnpB-F/qrnpB-R; nifB1, qnifB-L/qnifB-R; nifS1, qnifS1-L/qnifS1-R; nifU1, qnifU1-L/qnifU1-R; nifH1, qnifH1-F/qnifH1-R; nifD1, qnifD-L/qnifD-R; nifK1, qnifK-L/qnifK-R; nifE1 5′, qnifE (Feb)-L/qnifE (Feb)-R; nifE1 3′, qnifE (Sept)-L/qnifE (Sept)-R; nifN1, qnifN-L/qnifN-R; nifX1, qnifX-L/qnifX-R; nifW1, qnifW-L/qnifW-R; hesA1, qhesA-L/qhesA-R; and fdxH, qfdxH-L/qfdxH-R. The qPCR program included denaturation at 95°C for 1 min, followed by 40 cycles at 95°C for 10 s, 55°C for 30 s, and 72°C for 30 s. Data analysis was performed using Bio-Rad CFX Manager 3.0. Quantification cycle (C_q) values for gene-specific transcripts were normalized to rnpB, a structural RNA component of RNase P commonly used for reverse transcription (RT)-qPCR (42, 43). The C_q values for rnpB did not decrease over the 60-min time course of the half-life experiments. The averages and standard deviations for triplicate biological samples are shown (see Fig. 4 and 5). The RNA half-life was calculated using time points after the addition of rifampin that best represented 2-fold decay. The specific time points used depended on the particular transcript, since some transcripts showed an increase up to 30 min after the addition of rifampin. Since levels of induction of the nitrogenase genes can be variable across biological replicates, half-life calculations were made separately for each biological replicate, and the half-lives were averaged. This increase in transcript after the addition of rifampin was also reported for short-half-life transcripts in the slow-growing marine 1yanobacterium Prochlorococcus (44).

TABLE 2.

Oligonucleotides

FIG 4.

Transcript abundances of nif cluster genes in the wild-type strain compared to an xisA mutant. (A) The amount of transcript for genes in the nif cluster (relative to nifB1) was determined by RT-qPCR using RNA isolated from the wild-type strain, FD, and from an xisA mutant, BP669, that cannot remove the 11-kb excision element (Fig. 1A). Strains were grown with ammonium, and then the nif genes were induced by 24 h of starvation for fixed nitrogen. (B) Transcript levels in BP669 are shown relative to the wild-type strain, FD, in order to more clearly visualize the low levels of transcript for nifK1, nifE1, nifN1, nifX1, and nifW1 in the mutant. The error bars indicate standard deviations.

FIG 5.

Stability of nif cluster transcripts. Transcript stability was determined by measuring transcript amounts by RT-qPCR from RNA collected at 10, 20, 30, 45, and 60 min after the addition of rifampin to cells that had been induced to fix nitrogen by the removal of fixed nitrogen from the medium 24 h before the start of the experiment (time zero). The transcripts were normalized to rnpB. Measurement of half-life is described in Materials and Methods. The data are based on triplicate biological samples. The error bars indicate standard deviations.

5′ rapid amplification of cDNA ends (RACE) was performed as described previously (45), with the following modifications. A total of 20 μg of RNA was treated with DNase, followed by ethanol precipitation, and suspended in 88 μl RNase-free water. The sample was divided into two reaction mixtures; half of the RNA was treated with 20 U of tobacco acid pyrophosphatase (TAP) (Epicentre, Madison, WI), whereas the other half of the RNA was not treated. Next, 150 pmol of the RNA adapter, RNAoligo09, was added to each tube before the tubes were 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 (NEB). The ligated RNA was extracted with organic solvents and ethanol precipitated with 2 pmol of cDNA primers, resuspended in 20 μl of water, and reverse transcribed using Superscript III (Invitrogen) according to the manufacturer's protocol using the following primers: hesA, hesA-RPE2; fdxH, fdxH-RPE3; and nifW-Ava3936, hesA-RPE1 and fdxH-RPE4. PCR was performed using the left primer P1 oligo A and the following right primers: hesA, hesA-RPE4 and nifW-Ava3936, and qnifW-R and qAva3936-R. A second RNA linker primer, P1 oligo B, was used with fdxH-RPE1 to amplify fdxH. The PCR bands were excised and sequenced to determine the 5′ ends.

β-Galactosidase assays.

Strains for β-galactosidase assays were grown in AA/8 supplemented with 5 mM NH₄Cl and 10 mM TES, pH 7.2, for approximately 5 generations. Cultures at an OD₇₂₀ of 0.1 to 0.15 were washed free of fixed nitrogen, diluted to an OD₇₂₀ of 0.04 for cultures grown with fixed nitrogen (+N) and 0.09 for cultures without fixed nitrogen (−N), and grown with or without ammonium for 24 h prior to β-galactosidase assays. β-Galactosidase assays were done in 96-well plates with biological quadruplicate samples using a modification of the standard Escherichia coli assay (46). Cell density was measured at OD₇₂₀, and then the cells were permeabilized by adding a 10× buffer to yield a final concentration of 0.05 M NaPO₄, pH 7.4, 0.1% Sarkosyl, 160 μg ml⁻¹ ortho-nitrophenyl-β-galactoside, and the reaction mixtures were allowed to incubate for 15 to 30 min at 30°C until a yellow color was visible and then were stopped with sodium carbonate (150 mM final concentration). Samples were read at OD₄₂₀ and OD₆₆₅. The cells lysed partially in the permeabilization buffer, so it was necessary to include a correction factor for light scattering, determined empirically (by comparing OD₄₂₀ values for samples that were centrifuged to those for samples that were not) to be most reproducible at OD₆₆₅, that measured both the light scattering and the chlorophyll that increased the apparent OD₄₂₀ of the samples. Activity was determined essentially as described by Miller but using the following formula: 1,000 × [(OD₄₂₀ of sample − OD₄₂₀ of blank) − 1.58 × (OD₆₆₅ of sample − OD₆₆₅ of blank)]/(OD₇₂₀ of sample − OD₇₂₀ of blank) × t (min).

In situ localization of β-galactosidase.

Strains for in situ localization of β-galactosidase expression were grown in AA/8 supplemented with 5 mM NH₄Cl and 10 mM TES, pH 7.2, for approximately 5 generations. Cultures at an OD₇₂₀ of 0.1 to 0.15 were washed free of fixed nitrogen. The cells were diluted to 0.05 to 0.1 in AA/8 and grown for 24 h without ammonium prior to in situ localization. The cells were washed twice with water and fixed for 15 min at 25°C with 0.04% glutaraldehyde. The 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 the cells were resuspended in Vectashield (Vector Laboratories), an antibleaching agent. The cells were imaged on a Zeiss epifluorescence microscope using a fluorescein filter set (excitation, 450 to 490 nm; dichroic, 510 nm; barrier, 520 nm) with a 560-nm short-pass filter. Images were taken using a Retiga EXi (QImaging) cooled charge-coupled device camera with IP Labs 4.0 software (BD Biosciences).

RESULTS

A promoter in nifE1.

We have previously shown that in A. variabilis there is a promoter driving nifB1 that, supplemented by a promoter in nifU1, also drives expression of nifH1, the most abundant nif transcript. Within nifD1, there is an 11-kb excision element that separates nifBSUHD1 from nifDKENXW1 (Fig. 1A); however, that element is removed during heterocyst differentiation so that the entire cluster can be driven by the nifB1 promoter (47). In a unicellular nitrogen-fixing strain, Cyanothece sp. strain PCC 8801, transcription start sites were identified upstream of nifE, nifX, and nifW (48), so it seemed likely that there would be a promoter upstream of nifE1 in A. variabilis. To determine whether the nifEN1 genes had their own promoter, we constructed fusions of various-size fragments upstream of nifE1 to a promoterless lacZ (Fig. 1B) and integrated them in the chromosome so that lacZ was driven only by these nifE1 upstream fragments (frt cross) (Fig. 1C). The control was JU483, a strain in which lacZ was driven by the entire chromosomal region upstream of nifE1 (promoter cross) (Fig. 1C). Neither the 500-bp fragment (JU479) nor the 900-bp fragment (JU480) produced β-galactosidase activity higher than that of the noninduced (+N) control cultures (Fig. 1B), and the values were very similar to those of the control strain JU467 (33), which lacks an insert but apparently expresses some activity by read-through from the plasmid. In contrast, the promoter cross in the wild-type strain (JU483), which included the nifB1 promoter, provided strong, inducible expression. We have shown previously that expression of nifH1 in a nifB1 promoter deletion mutant is about 35% that of the wild type, because there is a weak promoter in nifU1 (33). Similarly, lacZ fusions to nifE1 in a nifB promoter deletion strain (JU490) produced expression levels of about 50% that of the wild type (JU483) (Fig. 1B), suggesting that nifE1 was transcribed by another promoter, probably the promoter in nifU1, but that nifE1 lacked its own promoter.

In order to further characterize the control of transcription of nifEN1, we constructed strains in which various short DNA regions within the nif1 operon controlled the expression of the genes downstream of that region, independent of the nifB1 promoter. These strains were constructed by single-crossover recombination in the chromosome, as shown for one example (BP683) in Fig. 2A. The other constructs were made the same way but using the various nif1 fragments shown below the nif1 genes in Fig. 2A, and the relevant characteristics of each strain are described below. These constructs effectively separated transcription of the nif1 operon into two chromosomal regions, one driven by the nifB1 promoter and the other driven by the fragment used for recombination. This allowed independent assessment of the transcription of the two regions, which are separated by the integrated plasmid (Fig. 2A). The assay for function of the nif1 operon was the ability of these strains to grow under nitrogen-fixing conditions. In the control strain BP682, the nif1 operon was driven by a region that extended 2,257 bp upstream of the nifB1 gene, the first gene in the operon. Growth of that strain was somewhat slower than that of the wild-type strain, FD (Fig. 2B). Strain BP640, which included the weak nifU1 promoter but not the strong nifB1 promoter, grew much more slowly than the strains with the nifB1 promoter, consistent with prior results showing that the nifU1 promoter is weak compared to the nifB1 promoter (33). Strain BP641, which lacked both the nifB1 promoter and the nifU1 promoter, did not grow under diazotrophic conditions (Fig. 2B), confirming the importance of these two promoters for transcription of nifHDK1. In strains BP642, BP683, and BP643, transcription of nifHDK1 was driven by the normal nifB1 and nifU1 promoters, but transcription of nifENXW1 was driven by three overlapping regions in the nifK1-nifE1 region (Fig. 2A shows the locations of the start sites for the fragments driving transcription of nifENXW1). In BP642, the region driving transcription of nifENXW1 included nifK1 and all the genes downstream of nifK1. In BP683, the region driving transcription of nifENXW1 included ava3931 and all the genes downstream of ava3931. In BP643, the region driving transcription of nifENXW1 excluded ava3931 but included the intergenic region between ava3931 and nifE1 and all the genes downstream of nifE1 (Fig. 2A). Failure of these strains to grow under diazotrophic conditions (Fig. 2B) supported the lacZ reporter data (Fig. 1B) showing that there was no promoter upstream of nifEN1. However, strain BP644 grew as well as the control strain, BP682, indicating a promoter within nifE1. Single-crossover recombination of the fragment used to produce BP644 resulted in a strain in which the nifB1 and nifU1 promoters drove expression of nifBSUHDKE1 and the 5′ end of the truncated nifN1, while the second copy of the duplicated region, beginning within the nifE1 gene, drove expression of nifN1 and the downstream genes. Growth of BP644 in the absence of fixed nitrogen (Fig. 2B) implied that there was a promoter within the nifE1 gene that drove transcription of nifN1 and the downstream genes. The promoter must be downstream from the nifE-L2(BP) left primer used to make the nifE1-nifN1 fragment in pBP644, since that fragment supported growth. Further experiments to identify this promoter within nifE1 are described below. Strain BP692 also grew as well as the control strain, BP682 (Fig. 2B). Single-crossover recombination of the fragment used to produce BP692 resulted in a strain in which the nifB1 and nifU1 promoters drove expression of nifBSUHDKEN1 (and part of nifX1), while the region beginning inside nifN1 drove expression of nifX1 and the downstream genes. Growth of this strain (Fig. 2B) indicated that there might also be a promoter within or just downstream of nifN1 that drove transcription of nifX1 and the downstream genes. RT-qPCR data measuring transcription of nifX1 and nifW showed low, but detectable, levels of transcript for these genes in BP692 (data not shown), suggesting that there may be a weak promoter driving their expression.

FIG 2.

Growth of nif mutant strains. (A) nif mutants were constructed by single recombination between nif chromosomal fragments cloned in a plasmid and the homologous region of the chromosome, resulting in duplicated copies of the cloned fragments. An example, strain BP683, is shown; the upstream region, including the nifB1 and nifU1 promoters (not shown), drives a truncated nifE1 gene, while the complete nifE1 gene, downstream of the plasmid, functions only if it has a promoter. The cloned fragments used to construct each of the described nif mutants (made in the same manner as shown for BP683) are shown below the nif cluster map. (B) Growth of mutants in a medium lacking fixed nitrogen. The error bars indicate standard deviations.

The growth data shown in Fig. 2B indicated that strain BP644 has a promoter located at least 440 bp inside nifE1. About 590 bp inside nifE1, we identified a region that has sequence similarity to the −10 promoter region of nifB1 (Fig. 3A), and a primary transcription start site was identified nearby by RNA-seq (35); therefore, internal fragments of nifE1 that included or excluded that putative promoter region were fused to a promoterless lacZ and integrated into the chromosome in the frt region. Strain BP754, containing a lacZ fusion that lacked the putative transcription start site, showed no expression of lacZ in response to nitrogen deprivation (Fig. 3B). Similarly, strain BP755, with a lacZ fusion that extended 19 bp upstream from the dotted underlined C in Fig. 3A, showed no expression of lacZ in response to nitrogen deprivation (Fig. 3B). In contrast, strain BP756, with lacZ fused to a fragment that extended 68 bp upstream from the putative transcription start site, showed increased expression of lacZ in response to nitrogen deprivation (Fig. 3B). The low level of activity indicated either that this was a weak promoter or that additional DNA elements not included in the 68-bp region upstream of the putative transcription start site in BP756 were required for activity. We were unable to precisely define the transcription start site within nifE1 using 5′ RACE, which failed to produce sufficient product to sequence; however, based on RNA-seq data (35) and our results, it seems likely that the promoter is close to the putative −10 region shown in Fig. 3A.

FIG 3.

Localization of the nifE1 internal promoter. (A) Comparison of the promoter regions of nifB1 and nifE1. The underlined nucleotide is the transcription start site for nifB1, while the dotted underlined nucleotide is the possible tss for nifE1, based on the similarity to nifB1. (B) Fragments of nifE1, with or without the putative promoter region shown in panel A, were fused to a promoterless lacZ as shown, and β-galactosidase activity was determined for the strains containing the fusions. Cells were either grown with fixed nitrogen (+N) or starved for 24 h without fixed nitrogen (−N) prior to the β-galactosidase assay. Asterisks indicate conserved nucleotides.

Transcript levels in an xisA mutant.

An xisA mutant is unable to excise the 11-kb excision element that interrupts nifD1; hence the strain cannot fix nitrogen. The mutant would also not be able to use the nifB1 and nifU1 promoters to transcribe genes downstream from the 11-kb excision element and would not be able to transcribe nifDKENXW1 unless one of more of these genes had a promoter. As expected, RT-qPCR data showed that transcript levels for the nifB1, nifS1, and nifU1 genes upstream of the 11-kb element were not affected much in the xisA mutant, whereas the genes immediately downstream of the excision element, including nifD1 and nifK1, were dramatically reduced in the xisA mutant compared to the wild-type strain (Fig. 4A). Transcript levels for nifN1, nifX1, and nifW1 were higher than for nifD1 and nifK1 but lower than for the nifB1, nifS1, and nifU1 genes upstream of the 11-kb element (Fig. 4A). Because we had identified a promoter in nifE1, we determined transcript levels for nifE1 using two sets of primers, one set located upstream of the promoter and the other set downstream from the promoter. Although transcript levels for nifE1 were low, there was much more nifE1 transcript downstream from the internal nifE1 promoter in the xisA mutant than upstream of the nifE1 promoter (Fig. 4B), confirming the presence of the promoter in nifE1, which also allowed expression of nifN1 and genes downstream of nifN1. While the low levels of transcript for nifD1 and nifK1 in the xisA mutant were expected, the large reduction in the amount of nifH1 transcript was not, since it is made using the nifB1 promoter and the internal nifU1 promoter (33). However, the nifH1 transcript is processed, so it is possible that the very long transcript that would result from failure to excise the 11-kb element would destabilize the processed nifH1 transcript, resulting in its degradation. Transcript levels for hesA and fdxH, which are reported to be under the control of their own promoters (30, 31), were much lower in the xisA mutant, indicating that their expression may also be influenced by the nifB1 and nifU1 promoters.

Stability of nif1 transcripts.

We examined the stability of the nif gene transcripts in the wild-type strain, FD, by quantitation of mRNA after the addition of rifampin, which inhibits initiation but not elongation by RNA polymerase (49). Cells were deprived of fixed nitrogen for 24 h to induce expression of the nif1 genes prior to the addition of rifampin, and mRNA was measured by RT-qPCR at various times after the addition of rifampin. The patterns of mRNA degradation were strikingly different for nifB1, fdxN1, nifS1, and nifU1, the genes upstream of nifH1, compared to nifH1 and the downstream genes, nifD1, nifK1, nifE1, nifN1, and nifX1 (Fig. 5). The genes upstream of the RNA-processing site, located between nifU1 and nifH1, decayed with a half-life of about 10 to 15 min, and 80% of the mRNA had disappeared within 30 min after the addition of rifampin. The nifH1, nifD1, and nifK1 genes downstream from the RNA-processing site had longer half-lives, with 50 to 90% of the initial mRNA present 30 min after the addition of rifampin (Fig. 5). This was due to two contributing factors. (i) There was an initial increase in the amount of transcript at 10 to 20 min after the addition of rifampin for nifH1 and all the genes downstream of nifH1. (ii) The half-lives of the transcripts for these genes (17 to 34 min) were longer than those for the transcripts for nifB1-nifU1. Transcripts for the genes farther downstream, nifE1, nifN1, and nifX1, had half-lives similar to those of nifB1, nifS1, and nifU1; however, like nifH1-nifK1, they also showed a pattern of increased transcript levels immediately after rifampin addition, with large amounts of transcript still present 30 to 45 min after rifampin addition (Fig. 5). This phenomenon has been observed for downstream genes in the multigene ATPase operon of Prochlorococcus (44) and is likely explained in part by the fact that RNA polymerase molecules that were already transcribing at the time of rifampin addition would complete their transcripts. Genes farther from the promoter would have more RNA polymerase molecules that would continue to make transcripts after rifampin addition than genes close to the promoter. In addition, there are promoters in nifU1 and nifE1 that could contribute, by the same mechanism, to transcripts for the downstream genes. Lastly, it is possible that some low level of transcription continued even after the addition of rifampin. Transcripts with long half-lives that are made after rifampin addition by any of these mechanisms, such as nifH1, would appear to increase more than genes with short half-lives, because the longer-lived transcripts would accumulate more than those with shorter half-lives.

Processed nif1 transcripts.

The hesA, hesB, and fdxH genes are not part of the nif cluster; however, they are conserved in all cyanobacterial nif clusters, and according to RNA-seq data, they are upregulated under conditions of nitrogen depletion (34, 35). Both hesA and fdxH have been shown to have a promoter (30, 31); however, the pattern of RNA decay for hesA and fdxH was similar to that for nifH1, nifD1, nifK1, nifE1, nifN1, and nifX1, with much of the transcript appearing to be resistant to rifampin, suggesting that RNA processing may also play a role in the transcription of these genes (Fig. 5).

We used 5′ RACE, which can distinguish between processed and primary transcripts (45), to map the 5′ ends of the hesA1 and fdxH1 transcripts. In this technique, unless the RNA already has a monophosphate end because it is a cleavage product from a primary transcript, it must be treated with TAP, which hydrolyzes the 5′ triphosphate to a monophosphate so that it can be ligated to the RNA linker. Primary transcripts require the TAP treatment because they have a 5′ triphosphate. Using 5′ RACE with and without TAP treatment, we performed RNA ligase-mediated RT-PCR and then recovered and sequenced the cDNA bands to determine the 5′ ends of hesA1 and fdxH1. No product was seen for hesA1 unless the RNA was treated with TAP (Fig. 6); therefore, it was a primary transcript, and the 5′ end was identical to the transcription start site reported for hesA in Anabaena sp. PCC 7120. In contrast, for fdxH1, a strong band was visible from RNA samples that were not treated with TAP (Fig. 6), indicating that the transcript is processed from a larger transcript. The processing site for the fdxH1 transcript exactly matched the published transcription site for fdxH in Anabaena sp. PCC 7120 (30) and in both strains is located 132 bp upstream from the translation start site. In this region, we also identified the 5′ end of a weak processed transcript that extended from near the end of ava3935 into nifW1 (Fig. 6), which may provide increased stability for the ava3936 and nifW1 transcripts.

FIG 6.

Mapping nif transcript 5′ ends. RNA isolated from cells grown without fixed nitrogen for 24 h was analyzed by 5′ RACE. Lanes: 1, 4, and 9, DNA ladder; 2 and 3, a band for a primary transcript 39 bp upstream of hesA1 with (+) but not without (−) TAP treatment of RNA; 5 and 6, a processed transcript located upstream of hesA1, ava3936, nifW, and 60 bp before the end of ava3935, present with or without TAP treatment of RNA; 7 and 8, a processed transcript located 132 bp upstream of fdxH with or without TAP treatment of RNA.

Heterocyst specificity of the nif1 promoters.

In order to determine whether the secondary promoters in nifU1 and nifE1 provided heterocyst-specific transcription, we localized the expression of β-galactosidase in strains with fusions of a promoterless lacZ to the nifB1 promoter alone, the nifU1 promoter alone, or the nifE1 promoter alone. The nifB1 promoter provided strong heterocyst-specific expression of lacZ in strain BP682 (Fig. 7A and B), while the nifU1 promoter alone (strain JU472) produced much lower levels of lacZ expression, requiring at least 5 times longer exposure times for visible expression (Fig. 7C and D); however, that expression was heterocyst specific. Expression from the nifE1 promoter in BP756 was also heterocyst specific but was weaker than that from the nifU1 promoter (Fig. 7E and F).

FIG 7.

In situ localization of expression of β-galactosidase in strains with fusions of a promoterless lacZ to the nifB1 promoter alone (the 2.3-kb region directly upstream of nifB1) in strain BP682 (A and B), the nifU1 promoter alone (in a 700-bp fragment directly upstream of nifH1) in strain JU472 (C and D), or the nifE1 promoter alone in strain BP756 (shown in Fig. 3) (E and F). (A, C, and E) Bright-field images of filaments. The arrows indicate representative heterocysts. (B, D, and F) Fluorescence images showing the expression of β-galactosidase primarily in heterocysts. Exposure times in seconds are provided.

DISCUSSION

The large (>16-kb) region containing the nitrogenase genes (excluding the excision element[s]) has been thought to comprise multiple operons, usually described as nifB-fdxN-nifS-nifU, nifHDK, nifENXW, hesAB, and fdxH (reviewed in reference 19). Apparent transcription start sites for nifB, nifH, hesA, and fdxH have been reported in strains of Anabaena spp. (26, 30, 31, 33, 35, 36), although we showed that the apparent nifH1 transcription start site in A. variabilis is actually a processed site (33). However, there is no published information on a promoter for nifEN. Transcripts for nifHDK are abundant compared to most of the other nif genes (see Fig. 4A), so their transcript sizes of 1.1 kb (nifH), 2.8 kb (nifHD), and 4.7 kb (nifHDK) were determined over 25 years ago by Northern blotting (29). While fdxH is included in this cluster, the role of its product as the ferredoxin for nitrogen fixation is tangential to nitrogenase biosynthesis and assembly, and it is useful, but not essential, for nitrogen fixation (50). The roles of HesA and HesB are not known, although they are also encoded in all the nif clusters identified in the genomes of nitrogen-fixing cyanobacteria (51).

We have shown in this work and in a previous study that, in addition to the promoter located upstream of nifB1, there are at least two promoters in the nif cluster, one within nifU1 and the other within nifE1. The internal nifU1 transcript was shown to be a primary transcript, and the promoter region in nifU1 was able to drive expression of lacZ (33). Similarly, the nifE1 promoter region, shown in Fig. 3A, was able to drive weak expression of lacZ. These transcription start sites were identified in nifU and nifE in Anabaena sp. PCC 7120 by RNA-seq, using a method that enriched for primary transcripts (35), thus confirming that they exist in both strains of Anabaena spp.

Despite the presence of promoters in nifU1 and in nifE1, it appears likely that the nifB1 promoter serves as the primary promoter for the entire cluster of nif1 genes. Although both internal promoters are weak, the internal nifE1 promoter provides sufficient transcript for nifN1 to support diazotrophic growth at nearly wild-type rates in the mutant strain BP644, in which nifN1 is made only from this promoter (Fig. 2B). This is surprising, since the nifE1 promoter provides less than 20% of the normal amount of nifN1 transcript, as seen in the xisA mutant, in which neither the nifB1 nor the nifU1 promoter can drive expression of nifN1. Mutant BP692 allows expression of the nif1 genes from nifB1 through nifN1 from the nifB1, nifU1, and nifE1 promoters; however, nifX and the downstream genes ava3935, ava3936, and nifW should lack a promoter in this mutant, yet the strain still grew nearly as well as the wild type, suggesting that there may be one or more as yet uncharacterized promoters between nifN1 and the hesA1 promoter.

The characteristics of promoters that lead to expression of genes in heterocysts versus vegetative cells is not well understood, although progress has been made (52). It is clear that SigE plays a role in the expression of the nitrogenase genes, since a sigE mutant showed lower levels of transcripts of nifH and fdxH in Anabaena sp. PCC 7120 (20). We were curious to know whether the weak nifU1 and nifE1 promoters were able to drive heterocyst-specific expression, which they did (Fig. 7). Further, the level of expression of β-galactosidase in heterocysts correlated with the strength of the promoter, with the nifE1 promoter-lacZ fusion requiring at least 7-fold-longer exposure time than the nifB1 promoter-lacZ fusion for visible heterocyst-specific fluorescence. The nifU1 promoter was somewhat stronger than the nifE1 promoter but still very weak compared to the nifB1 promoter (Fig. 7); however, the nifE1 promoter region that was characterized was shorter than the nifU1 promoter region and could lack upstream enhancer elements. Although there is significant sequence similarity in the −10 regions of the nifB1 promoter and the nifE1 internal promoter (Fig. 3), alignment of the two promoters with the nifU1 internal promoter revealed few nucleotides that were conserved among all three promoters, and there was no conservation in the −35 region among the three promoters, thus providing no clue as to sequences that might be important for heterocyst-specific expression.

The hesA gene has been shown to have a promoter (30), which was corroborated by RNA-seq data using a method that enriched for primary transcripts (35), and we have shown here that the 5′ end of hesA1 represents a primary transcript. However, levels of hesA and hesB transcripts were greatly reduced in the xisA mutant compared to the wild-type strain (Fig. 4B), suggesting that the nifB1 and nifU1 promoters normally contribute to transcription of these genes. Similarly, the levels of fdxH transcript were also greatly reduced in the xisA mutant, suggesting that its transcription is also enhanced by the nifB1 and nifU1 promoters. Although the fdxH gene has been shown to have a transcription start site (31), that site was not found in the RNA-seq data using a method that enriched for primary transcripts (35), and we have shown here that the transcript is processed, presumably from the transcript made from the hesA1 promoter but possibly from the nifB1, nifU1, and nifE1 promoters, as well.

In analyzing the hesA1 promoter for these studies, we found a possible error in the annotation of the gene. Although the 100-bp region upstream of the ATG start site in hesA in Anabaena sp. PCC 7120 is identical to that for hesA1 in A. variabilis, the genes are annotated with different start sites. In the IMG database, the deduced start site for hesA1 in A. variabilis is shown to be a GTG that is 45 nucleotides upstream from the ATG that is annotated as the start of hesA in Anabaena sp. PCC 7120 (51) (Fig. 8). If the GTG were the correct start codon, then the transcription start site for hesA, which has been mapped now in both strains as well as by RNA-seq (30, 35), would be 7 nucleotides inside the hesA gene. Therefore, it seems likely that the ATG located 39 bp downstream from the transcription start site in both strains is the correct translation start site. Among diverse cyanobacteria, the ATG site is more highly conserved than the upstream GTG site (Fig. 8).

FIG 8.

Alignment of the hesA region upstream of the start of the gene. The transcription start sites for Anabaena (Nostoc) sp. PCC 7120 and A. variabilis are boxed in red. The annotated translation start site for hesA in A. variabilis (GTG) upstream of the tss is highlighted in yellow. The annotated translation start site for hesA in Anabaena (Nostoc) sp. PCC 7120 (GTG) 45 bp upstream of the ATG start site is highlighted in green. IMG identification information for these genes is provided in the table. The sequence alignment was done using Clustal (58). Asterisks indicate conserved nucleotides.

The nifH1 transcript is very abundant in filaments starved for fixed nitrogen (27, 28), reflecting the need for large amounts of the multifunctional NifH protein that is required not only for catalytic activity of the enzyme but also for FeMo cosynthesis (53). The high levels of nifH1 transcript suggested that it has its own strong promoter; however, in A. variabilis, transcription of nifH1 depends primarily on the nifB1 promoter, with some contribution from the weaker promoter in nifU1 (33). The relative abundances of transcripts for nif1 (Fig. 4A) indicate that the amounts are controlled, despite the lack of discrete operons and the presence of only one strong promoter that is distant from many of the genes, including the key structural genes nifHDK1, whose products are required in large amounts. If the amount of transcript is not controlled by the strength of the promoter, then it must be controlled by the stability of the transcript. We have shown that there are processing sites just upstream of two abundant transcripts, nifH1 and fdxH1, and that these transcripts also have long half-lives. The long half-life could be the result of secondary structure that protects the transcript from rapid degradation, which is suggested by the secondary structure at the site of processing of the nifH1 transcript (33). Similarly, the processing site for fdxH1 is just inside the 5′ end of a possible complex stem-loop structure (ΔG = −23 kcal mol⁻¹) (54) that starts 4 nucleotides upstream from the processing site and extends 85 bp toward fdxH (Fig. 9).

FIG 9.

RNA secondary structure of the fdxH1 region around a transcription processing site (tps) located 132 bp upstream from the translation start site of fdxH. The model was created using mFold (54).

In summary, transcriptional control of the large cluster of nitrogenase genes includes synthesis of a very long primary transcript from the nifB1 promoter, supplemented by at least two weaker promoters in nifU1 and nifE1 and a promoter for hesA1. Processing of this transcript at positions upstream of nifH1 and fdxH1 correlates with increased stability of the transcripts, resulting in increased amounts compared to transcripts that are not close to processing sites. A summary of the locations of transcription start sites and transcription-processing sites is provided in Fig. 1A.