ABSTRACT
In the cyanobacterium Anabaena variabilis ATCC 29413, aerobic nitrogen fixation occurs in micro-oxic cells called heterocysts. Synthesis of nitrogenase in heterocysts requires expression of the large nif1 gene cluster, which is primarily under the control of the promoter for the first gene, nifB1. Strong expression of nifH1 requires the nifB1 promoter but is also controlled by RNA processing, which leads to increased nifH1 transcript stability. The processing of the primary nifH1 transcript occurs at the base of a predicted stem-loop structure that is conserved in many heterocystous cyanobacteria. Mutations that changed the predicted secondary structure or changed the sequence of the stem-loop had detrimental effects on the amount of nifH1 transcript, with mutations that altered or destabilized the structure having the strongest effect. Just upstream from the transcriptional processing site for nifH1 was the promoter for a small antisense RNA, sava4870.1. This RNA was more strongly expressed in cells grown in the presence of fixed nitrogen and was downregulated in cells 24 h after nitrogen step down. A mutant strain lacking the promoter for sava4870.1 showed delayed nitrogen fixation; however, that phenotype might have resulted from an effect of the mutation on the processing of the nifH1 transcript. The nifH1 transcript was the most abundant and most stable nif1 transcript, while nifD1 and nifK1, just downstream of nifH1, were present in much smaller amounts and were less stable. The nifD1 and nifK1 transcripts were also processed at sites just upstream of nifD1 and nifK1.
IMPORTANCE In the filamentous cyanobacterium Anabaena variabilis, the nif1 cluster, encoding the primary Mo nitrogenase, functions under aerobic growth conditions in specialized cells called heterocysts that develop in response to starvation for fixed nitrogen. The large cluster comprising more than a dozen nif1 genes is transcribed primarily from the promoter for the first gene, nifB1; however, this does not explain the large amount of transcript for the structural genes nifH1, nifD1, and nifK1, which are also under the control of the distant nifB1 promoter. Here, we demonstrate the importance of a predicted stem-loop structure upstream of nifH1 that controls the abundance of nifH1 transcript through transcript processing and stabilization and show that nifD1 and nifK1 transcripts are also controlled by transcript processing.
INTRODUCTION
Anabaena variabilis is a filamentous cyanobacterium that fixes nitrogen under oxic growth conditions in cells called heterocysts. Heterocysts, which are specialized micro-oxic cells that develop in a semiregular pattern in the filament in response to nitrogen stress, fix atmospheric dinitrogen to ammonium using the enzyme nitrogenase (1–3). The micro-oxic conditions of the heterocyst result from the glycolipid layer that may restrict oxygen diffusion into the cell, the absence of oxygen-evolving photosystem II, and increased respiration, all of which serve to protect nitrogenase from oxygen (4–6). While all heterocyst-forming cyanobacteria make a Mo nitrogenase that functions in heterocysts, A. variabilis is unusual because it has three nitrogenases, two Mo nitrogenases and one V nitrogenase (reviewed in reference 7). The primary nitrogenase, which is expressed in heterocysts in cultures growing in air in an environment with adequate molybdate, is a Mo nitrogenase encoded by the nif1 genes (8, 9). The Nif1 nitrogenase is expressed late in heterocyst differentiation, after the cell becomes micro-oxic (10). Under similar growth conditions, but in an environment deficient in molybdate but with adequate vanadate, heterocysts make an alternative, V nitrogenase, encoded by the vnf genes (8, 11). The V nitrogenase is completely repressed in the presence of molybdate. The second Mo nitrogenase, encoded by the nif2 genes, is made only in anoxic vegetative cells in an environment with adequate molybdate (12). Synthesis of all three nitrogenases is repressed in cells grown with a source of fixed nitrogen (+N).
Nitrogenase synthesis requires the products of at least a dozen genes, most of which are encoded by conserved genes that are found in large nif clusters in all nitrogen-fixing bacteria. Dimers of NifD (α-subunit) and NifK (β-subunit) make up the heterotetrameric dinitrogenase with two M clusters (comprising MoFe7S9C-homocitrate) (13–16). NifH is dinitrogenase reductase with an [Fe4-S4] cofactor, which transfers electrons to dinitrogenase (17). NifS transfers sulfur from cysteine to NifU (18), which acts as a scaffolding protein for [Fe-S] cluster assembly (15, 19). The [Fe-S] clusters are transferred to NifB to make NifB-co, an [Fe6-S9] cluster that serves as the precursor to the M cluster (20, 21). NifE and NifN, which are homologs of NifD and NifK, form a similar heterotetrameric complex and serve as a scaffold for M cluster assembly prior to its transfer to aponitrogenase (15, 22). NifW is involved in the processing of homocitrate (23), while NifX may function as a reservoir of FeMo cofactor (24). NifV makes homocitrate, a component of FeMo cofactor (15), while NifZ is involved in P cluster assembly (25, 26). Missing in cyanobacteria are the genes for NifQ, the Mo donor to FeMo cofactor (27); NifM, which stabilizes NifH (28, 29); and NafY, which stabilizes the open conformation of apo-MoFe protein prior to the insertion of the M cluster (30, 31). In A. variabilis, the organizations of the nif clusters for nif1 and nif2 are similar, but the nif2 cluster lacks the 11-kb excision element that interrupts nifD1 in the nif1 cluster and the nifEN genes are fused into a single open reading frame (ORF) in the nif2 cluster (7). The Mo-repressed vnf genes in A. variabilis are not tightly clustered but are found within a region of about 40 kb, which includes many uncharacterized genes (7). In addition to the structural genes, vnfH, vnfDG, and vnfK, there are the genes for scaffolding proteins, vnfEN, and the vanadate transport genes, vupABC (32, 33).
In the Proteobacteria, nif genes are under the control of the NtrBC nitrogen-regulatory system, which controls synthesis of the regulatory proteins NifA and NifL (34). Activation of nif genes in the absence of oxygen and fixed nitrogen (−N) requires NifA or VnfA, as well as the alternative δ54 RNA polymerase (34). In cyanobacteria, there are no homologs of NtrBC, NifA, NifL, or VnfA. While no sigma factor specifically associated with nitrogen regulation, like the δ54 factor in Proteobacteria, has been identified in cyanobacteria, the sigma factor encoded by sigE is important, but not essential, for expression of the nif genes in Anabaena sp. strain PCC 7120 (35).
Regulation of nitrogenase expression in cyanobacteria is not well understood. The global regulator NtcA affects the expression of genes that respond to nitrogen availability in all cyanobacteria; however, in heterocyst-forming cyanobacteria, it plays a key role in sensing nitrogen starvation and initiating the complex process of heterocyst differentiation (1, 2). NtcA has been reported to bind weakly to a region upstream of nifH in Anabaena sp. PCC 7120, and a putative noncanonical NtcA binding site was identified (36, 37). However, a chromatin immunoprecipitation sequencing (ChIP-seq) analysis of RNA from Anabaena sp. PCC 7120 showed that NtcA binds to a site inside nifH in Anabaena sp. PCC 7120 (38). We were unable to detect binding of NtcA to the intergenic nifU1-nifH1 region of A. variabilis, and mutations in the putative NtcA binding site in the strain did not affect nitrogenase expression (39). This is consistent with the fact that there is no promoter in the nifUH1 intergenic region.
Within the large nif cluster of Anabaena sp. PCC 7120, putative transcription start sites (tss) have been identified for nifB, nifH, hesA, and fdxH (36, 39–45). Transcription of the nif1 gene cluster of A. variabilis depends primarily on the promoter for the first gene in the cluster, nifB1 (39). The apparent transcription start sites for nifB1 and nifH1 in A. variabilis are identical to those mapped in Anabaena sp. PCC 7120, and there are weak, heterocyst-specific promoters within the coding regions of nifU1 and nifE1 (7, 39, 46). However, we found no evidence for a promoter in the 300-bp nifU1-nifH1 intergenic region driving expression of nifH1 (39). In contrast, we were able to verify primary transcription start sites upstream of nifB1 and hesA1 that matched the published sequence (7, 39, 46). Further, we demonstrated that the 5′ ends of nifH1 and fdxH1 in the nif1 cluster, as well as the V-nitrogenase genes, vnfH and vnfDG, have processed RNA ends, not primary transcript ends (39, 46, 47). In a mutant strain that lacks xisA, the gene that makes the excisase that removes the 11-kb element from nifD1, the nifB1 promoter can no longer drive expression of the genes downstream from the excision element, resulting in little or no expression of those genes (46). Even hesA1, which has its own promoter, is not expressed as well in the xisA mutant as in the wild-type (WT) strain, which suggests that the nifB1 promoter controls the expression of a gene that is 14 kb away.
The nifH gene is highly expressed because it is required in large amounts both for assembly of nitrogenase and for the function of the enzyme (16). In A. variabilis, the nifH1 transcript is much more abundant than the other nif1 transcripts, even though nifH1 lacks its own promoter, because the transcript is very stable (39, 46). The 5′ end of the processed transcript is at the base of a predicted stem-loop structure that may be responsible for the stability of the transcript (39, 46). We have examined the region between nifU1 and nifH1, including the predicted stem-loop structure located 123 bp upstream of nifH1 and an antisense RNA whose transcript begins near the predicted stem-loop structure, to determine the role of the region in the transcript abundance of nifH1.
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 derivatives of the strain were maintained on agar-solidified Allen and Arnon (AA) medium (48) supplemented, when appropriate, with 5 mM NH4Cl, 10 mM N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid (TES), pH 7.2, 40 μg ml−1 neomycin sulfate, 3 μg ml−1 each of spectinomycin and streptomycin, or 5 μg ml−1 erythromycin. 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). Growth, as the growth constant μ, was measured in triplicate cultures as the optical density at 720 nm (OD720) in very dilute cultures (to prevent self-shading) that grew exponentially over three cell divisions, as described previously (46).
Construction of strains.The construction of all plasmids is briefly described in Table 1, and all the primers used in the construction of strains are shown in Table 2. Strains BP703, BP706, and BP712 were constructed by making two PCR products that included the mutations of interest flanked by wild-type regions for recombination, allowing us to replace the wild-type region in the chromosome with the mutations in the PCR products. Mutations that destroyed the predicted stem-loop structure in the nifUH1 intergenic region (BP706) were constructed using primer sets nifHLoop1-L/nifHLoop1-R and nifHMutLoop2-L/nifHLoop2-R. Mutations that restored a stem-loop with different nucleotides (BP703) were constructed using primer sets nifHLoop1-L/nifHLoop1-R and nifHResLoop2-L/nifHLoop2-R. Mutation in the promoter for the antisense gene sava4870.1 was constructed using primer sets antiloopmut-1/antiloopmut-2 and antiloopmut-3/antiloopmut-4. The pairs of PCR products were triply ligated into the PstI/KpnI sites of pUC18 to create pBP701, pBP700, and pBP711, respectively. A 6-kb SacI fragment of pRL2948a, containing the mob site for mobilization, sacB for sucrose selection of double recombinants, and an erythromycin resistance (Emr) gene (49), was inserted into the SacI site in pBP700, pBP701, and pBP711 to create pBP703, pBP706, and pBP712, respectively. Mutant strains BP703, BP706, and BP712 were created by introducing the mobilizable plasmids pBP703, pBP706, and pBP712, respectively, into strain FD by conjugation, as previously described (50). Single recombinants were selected by Emr, followed by isolation of double-recombinant mutants using sacB selection (51) on AA plates supplemented with 5 mM NH4Cl, 10 mM TES, 10% sucrose. PCR was used to verify that there were no wild-type copies of the region of interest. Further, the nifUH1 region was sequenced in BP703, BP706, and BP712 to verify the mutations.
Plasmids and strains
Primers
Strains JU420, JU422, and JU423 were constructed as follows. A 4.1-kb BamHI region from 33D12 containing nifSUHD1 was cloned in pUC118, to create pEL1. A SmaI fragment containing the neomycin resistance (Nmr) cassette from pBP285 was inserted into the nifU1 gene of pEL1 at the EcoRV site to create pJU375. pJU376 was constructed by inserting the 5-kb EcoRV fragment of pRL2948 containing sacB and the chloramphenicol resistance (Cmr) cassette into the SmaI site of pJU375. pJU376 was partially digested with AgeI to remove a 377-bp fragment containing the 5′ end of nifH1 and religated to create pJU408. This plasmid was introduced into FD by conjugation to create a deletion of the 5′ end of nifH1 via double recombination, strain JU408, using sacB selection as described above. To generate the mutations in JU420, JU422, and JU423, PCR was performed on pEL1 using the following primer pairs: JU420-nifH69L/JU420-nifH69R, JU422 nifH95-L/JU422 nifH95-R, and JU423-120L/JU423-120R. These initial PCR fragments were ligated to circularize into a plasmid. The plasmids were named pJU378, pJU380, and pJU381, respectively. pJU397, pJU399, and pJU400 were made by inserting sacB and the Cmr cassette from pRL2948 cut with SacI into the SacI site in pJU378, pJU380, and pJU381, respectively. Plasmids pJU397, pJU399, and pJU400 were conjugated into JU408, and double recombinants were recovered using sucrose selection, and verified by screening for Nm sensitivity to generate JU420, JU422, and JU423, respectively. The mutations in the nifUH1 region were verified by sequencing.
Fusions of nifU1::lacZ and nifUH1::lacZ (pBP750 to pBP753 and pJU472) were made by cloning various PCR fragments, amplified from FD DNA, into the BglII/SmaI sites in pJU411 or pBP609, as described in Table 1. Plasmids pJU411 and pBP609 have an internal fragment of the fructose transport genes, frtABC (52), that allows screening for single recombinants at the frtABC locus, which results in a mutant that cannot use fructose. Promoter fragments were sequenced to verify that they contained no mutations. Recombinations of these plasmids with promoter fragments fused to lacZ into the frt region of A. variabilis by single crossover following conjugation (53) were identified by screening for colonies that were unable to grow in the dark with fructose (BP750 to BP753 and JU472).
RNA isolation and analysis.RNA decay was measured as previously described (46). Strains were grown in AA/8 supplemented with 5 mM NH4Cl and 10 mM TES, pH 7.2, and 50-ml cultures at an OD720 of 0.2 to 0.3 were washed free of fixed nitrogen and grown for 24 h with aeration to induce heterocysts and nitrogen fixation. Cells were removed at 24 h for the 0-h time, and 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 chilled quickly by pouring over ice in 2 50-ml centrifuge tubes, harvested, and frozen. RNA was isolated using Tri-Reagent (Sigma), as previously described (39), and subjected to DNase digestion (Turbo DNA-free kit; Ambion).
Reverse transcription-quantitative PCR (RT-qPCR) was used to determine the quantities and decay rates of transcripts. cDNA was made from 40 ng of RNA using Bio-Rad iScript Reverse Transcription Super Mix (catalog no. 170-8841) in a 10-μl reaction mixture. The cDNA was diluted 1:20 to 0.2 ng μl−1 for qPCR. qPCRs were done using Bio-Rad SsoAdvanced SYBR Green SuperMix (catalog no. 1725261) in a 10-μl reaction volume and contained 0.8 ng cDNA, 5 pmol of gene-specific primers, and 1× reaction supermix. The following gene-specific primer pairs were used: rnpB, qrnpB-F/qrnpB-R; nifB1, qnifB1-L/qnifB1-R; nifS1, qnifS1-L/qnifS1-R; nifH1, qnifH1-L/qnifH1-R; nifD1, qnifD-L/qnifD-R; nifK1, qnifK-L/qnifK-R; nifE1, qnifE (Sept)-L/qnifE (Sept)-R; and nifX1, qnifX-L/qnifX-R. Additionally, the following primer pairs were used to determine the stability around the predicted stem-loop structure: loop 5′, antinifH111/antinifH-5′; loop-span, qnifHloop-L/qnifHloop-R; and loop 3′, nifH-RTL/nifH-RTR. In some experiments, strand-specific cDNA was made from 50 ng of RNA using SuperScript III reverse transcriptase from Invitrogen. The manufacturer's recommended protocol was followed, using gene- and strand-specific primers rnpB-R and antinifH111 for looking at the expression of antisense RNA in BP712 and the WT. Additionally, we made strand-specific cDNA to look at the 3′ end of the antisense sava4870.1 transcript, using rnpB-R and antinifH111, antinifH164, antinifH216, antinifH245, or antinifH352. The reaction temperature for strand-specific cDNA was 52°C for 1 h, followed by an inactivation step at 70°C for 15 min; the cDNA was diluted 1:10 for qPCR. The qPCRs were done using Bio-Rad SsoAdvanced SYBR Green, as described above, using 4 μl of diluted strand-specific cDNA. The following primer pairs were used to amplify antisense RNA products in FD and BP712: rnpB, qrnpB-F/qrnpB-R; “antinif,” antinifH-5′/antinifH111; R20, nifUH-R20/antinifH111; R21, nifUH-R21/antinifH111; and nifH, nifH-RTR/nifH-RTL. Additionally, the following primer pairs were used to identify the 3′ end of the sava4870.1 antisense transcript: 111 bp, antinifH-5′/antinifH111; 164 bp, antinifH-5′/antinifH164; 216 bp, antinifH-5′/antinifH216; 245 bp, antinifH-5′/antinifH245; and 352 bp, antinifH-5′/antinifH352. The qPCR program for all reactions was 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. The averages and standard deviations of triplicate biological samples are shown. The RNA half-life was calculated using time points after the addition of rifampin that best represented 2-fold decay, as described previously (46).
5′ rapid amplification of cDNA ends (RACE) was performed as described previously (46). Briefly, 20 μg of DNase-free RNA was divided into two reaction mixtures; half was untreated, while half was treated with 20 U of tobacco acid pyrophosphatase (TAP) (Epicentre, Madison, WI). The RNA adapter RNAoligo09 (150 pmol) was added to each tube and was extracted with organic solvents and ethanol precipitated. The pellet, resuspended in 14 μl of water, was 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: nifD, nifD1-RPE1; nifK, qnifK-R; and Ava_3931, nifE-PCR3. PCR was performed using the left primer P1 oligo B and the following right primers: nifD, nifD-RPE3; nifK, nifK1-R; and Ava_3931, qAva3931-R. PCR bands were excised and sequenced to determine the 5′ ends.
β-Galactosidase assays.β-Galactosidase assays were done in 96-well plates with biological quadruplicate samples and a modification of the standard Escherichia coli assay (54) for cyanobacteria, as described previously (46), using cells permeabilized in 0.05 M NaPO4, pH 7.4, 0.1% Sarkosyl, 160 μg ml−1 orthonitrophenyl-β-galactoside. Samples were read at OD420 and OD665. Activity was determined essentially as described by Miller (54) 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 (in minutes). The averages and standard deviations of triplicate biological samples are shown.
Acetylene reduction assays.Nitrogenase activity was measured in vivo by acetylene reduction assays. Five-milliliter aliquots of cultures at an OD720 of about 0.15 were added to 16-ml Hungate tubes. The tubes were sealed with gas-tight stoppers, injected with 1.0 ml acetylene gas, and shaken at 30°C in the light for 60 min. Samples (250 μl) of headspace gas were injected into a Shimazu gas chromatograph equipped with a 6-ft Poropak N column. The average and standard deviation of triplicate biological samples are shown.
Naming the antisense RNA.We followed the recommended convention for naming of small RNAs, which is to begin with the letter “s,” then the three-letter genome identifier (ID) used in the KEGG database, “ava,” and then the number that indicates its genomic location to the nearest kilobase plus 1 (55). As recommended by the Bacterial Small Regulatory RNA Database (BSRD), we added a period and the number “1” at the end to indicate the number of small RNAs (sRNAs) identified in this location (which is currently one). Thus, the name is sava4870.1 based on a transcription start site at position 4869891 (kb 4869 + 1 = 4870).
RESULTS
The tps is required for nifH1 expression.We have shown previously that the large nif1 cluster encoding the heterocyst-specific Mo nitrogenase is controlled primarily by the first promoter of the cluster, located 328 bp upstream of nifB1, with additional weak promoters inside nifU1, inside nifE1, and upstream of hesA1 (Fig. 1) (7, 39, 46). A smaller nifH1 transcript is made by processing of the longer transcript at a transcriptional processing site (tps) located 123 bp upstream of nifH1 at the base of a predicted stem-loop structure. Additional processing sites were identified between nifH1 and nifD1, between nifD1 and nifK1, and between nifK1 and Ava_3931 (see below) and between hesB1 and fdxH1 (46).
Map of the nif1 gene cluster showing tss and tps. The 263-bp intergenic region between nifU1 and nifH1 is shown, with the predicted stem-loop structures (bottom) just downstream of the tps for nifH1 at −123 from the start of nifH1 and the tss for the antisense RNA that begins at −148 from nifH1. The processing site upstream of nifD1 is at −44 from the nifD1 start site. The processing site upstream nifK1 is at −131 from the nifK1 start site. The processing site upstream of Ava_3931 is near the start of the gene.
We have shown previously that a promoter in the nifU1 gene provides weak, heterocyst-specific expression of nifH1 independent of the primary nifB1 promoter (39, 46); however, the importance of sequences in the nifUH1 intergenic region for nifH1 expression is not known. We constructed fusions of the nifU1 internal promoter and various-length regions of the nifUH1 intergenic region to a promoterless lacZ in strains BP750, BP751, BP752, and BP753 (Fig. 2A). There was only basal-level expression of lacZ from the nifU1 promoter in fragments that lacked the processing site located 123 bp upstream of nifH1 (strains BP750, BP751, and BP752); however, strain BP753, with lacZ fused to the nifU1 internal promoter at a site 43 bp downstream from the processing site, had levels of β-galactosidase similar to those of the positive control, JU472, in which lacZ was fused at the start of nifH1 (39). This result is consistent with expression of lacZ from the weak nifU1 promoter (Fig. 2B), suggesting that the processing site is important for expression of nifH1.
Expression from the nifU1 internal promoter. (A) Construction of the nifU1::lacZ fusion strains. (B) β-Galactosidase activity, expressed in Miller units, of the strains grown with fixed nitrogen (+N) or 24 h after removal of fixed nitrogen (−N). In BP753, lacZ is fused 43 bp downstream from the tps, and in JU472, it is fused 146 bp downstream from the tps (39). The error bars indicate standard deviations.
The predicted stem-loop structure is important for nifH1 transcript processing and stability.The sequence near the 5′ end of the processed nifH1 transcript is conserved among many heterocyst-forming cyanobacteria, and that sequence forms the 5′ region of the predicted stem in the stem-loop. The stem-loop structure is also conserved, although the structure is variable in size and shape among the cyanobacterial strains (see Fig. S1B in the supplemental material; see also sequence data in reference 69). We have shown previously that the nifH1 transcript is much more stable than the nifU1 transcript (46); therefore, we were interested in the role of the stem-loop structure in the processing and stability of the nifH1 transcript.
In order to determine the importance of the predicted stem-loop region in the stability of the nifH1 transcript, we measured the stability of transcripts in the nifU1-nifH1 intergenic region by RT-qPCR from RNA extracted at various times after the addition of rifampin to block initiation of transcription. Primers were chosen to yield a product from transcripts directly upstream of the predicted stem-loop structure (Loop 5′), spanning the stem-loop structure (Loop-span), or directly downstream of the stem-loop structure (Loop 3′). Transcripts in the Loop 5′ region and the Loop-span region had a half-life of about 12 min, with decay evident immediately after the addition of rifampin (Fig. 3). In contrast, the transcript in the Loop 3′ region was much more stable, lasting longer than 30 min (Fig. 3), and it showed the transient increase in transcript after rifampin addition that was shown previously to be characteristic of the transcripts downstream from RNA-processing sites in the nif1 cluster (46). These data provide strong evidence that the region that comprises the stem-loop structure confers stability of the nifH1 transcript.
Stability of transcripts near the tps of nifH1. (A) 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 (zero time). Primers used for RT-qPCR amplified a region upstream from the tps (Loop 5′), a region spanning the tps (Loop-span), or a region downstream from the tps (Loop 3′). The data are based on triplicate biological samples. The error bars indicate standard deviations. (B) Map showing the locations, relative to the stem-loop structures, of Loop 5′, Loop-span, and Loop 3′ products.
To determine the importance of the sequence of the predicted stem-loop versus the secondary structure, two mutations were created that changed the sequence in this region. One mutation abolished the predicted stem-loop structure (strain BP706), while the second mutation changed the sequence in the region compared to the wild-type strain but maintained a stem-loop structure very similar to that of the wild-type strain with a similar ΔG (strain BP703) (Fig. 4A). The effects of these mutations on processing of the transcript were determined by 5′ RACE. No processed transcript was detected in RNA isolated from strain BP706, which lacked the predicted stem-loop structure. A weak product was detected in RNA isolated from strain BP703, which had a predicted stem-loop structure with altered nucleotides, and the sequence of this product showed that the 5′ end corresponded to the mutation created in BP703 (Fig. 4B). Thus, the stem-loop structure is important in RNA processing, and processing is impaired if the sequence in the predicted stem-loop region is incorrect. However, the fact that the mutant BP703 lacks wild-type levels of transcript suggests that the sequence, as well as the stem-loop structure, is important in normal transcript processing.
Mapping nifH1 transcript 5′ ends. (A) Predicted secondary structures in the region just downstream of the transcriptional processing site for nifH1 for the wild-type strain and the two mutants; one mutant abolished the first stem-loop (BP706), while the second mutant restored the stem-loop with a different sequence (BP703). The ΔG values represent only the first stem-loop structure, not both loops. (B) RNA isolated from cells grown without fixed nitrogen for 24 h was analyzed by 5′ RACE in strains FD, BP706, and BP703. Because the 5′ end of nifH1 is a processed site, the 5′ RACE was done without TAP treatment of the RNA. The bands indicated by the arrows for FD and BP703 were recovered from the gel and sequenced to determine the tps of nifH1 in the strains. The sequence for BP703 confirmed the stem-loop structure shown in panel A.
The effects of these mutations on the stability of genes in the nif1 cluster were determined by RT-qPCR with RNA extracted at various times after the addition of rifampin to block initiation of transcription. The major effect of the mutation in BP706 that destroyed the predicted stem-loop structure was on the nifH1 and nifD1 transcripts, which had much shorter half-lives than the wild-type strain, with little effect on the other nif1 genes tested (Table 3). The restoration of a stem-loop structure with altered nucleotides in strain BP703 resulted in increased half-lives for nifH1 and nifD1 compared to BP706, although they were not as long as those for the wild-type strain (Table 3). These results suggest that while the predicted stem-loop structure plays a role in stability, other factors, including the sequence, are probably also important.
Half-lives of nif1 transcripts in wild-type (FD) and mutant (BP706 and BP703) strains
The effects of the predicted stem-loop mutations on transcript abundance, nitrogenase activity, and growth rate were determined for FD, the wild-type strain; for BP703, lacking a predicted stem-loop structure; and for BP706, with a stem-loop structure with altered bases. Loss of the predicted stem-loop structure in mutant BP706 resulted in loss of about 90% of the nifH1 transcript and about two-thirds of the nifK1 transcript (Fig. 5A), with less effect on transcripts for nifE1 and nifX1 farther downstream. Restoration of the predicted stem-loop in BP703 resulted in a 2- to 3-fold increase in transcript levels for both nifH1 and nifK1 compared to BP706 and essentially normal levels of transcripts for nifE1 and nifX1 (Fig. 5A). Mutant BP706 showed only about 25% of the wild-type levels of nitrogenase activity, while restoration of a predicted stem-loop in BP703 partially restored nitrogenase (Fig. 5B), consistent with the increase in nifH1 transcript in this strain (Fig. 5A). Growth data confirmed that the complete loss of the predicted stem-loop in BP706 was accompanied by slower growth and that restoration of a predicted stem-loop increased the growth rate to about 75% of the wild-type rate (Fig. 5C). Thus, the loss of the stem-loop structure resulted in impaired growth that correlated with reduction in nitrogenase activity and loss of transcript, but restoration of a stem-loop structure with an altered sequence only partially restored growth, nitrogenase activity, and transcript abundance. Interestingly, the reduction of nifH1 transcript to only 10% of wild-type levels in the stem-loop mutant, BP706, reduced the nitrogenase activity only about 75% and the growth rate only about 50% compared to wild-type cells.
Analysis of mutants in the nifU1-nifH1 intergenic stem-loop structure. (A) The amounts of nifB1, nifS1, nifH1, nifK1, nifE1, and nifX1 transcripts in strains FD, BP706, and BP703 were measured by RT-qPCR in RNA harvested from cells 24 h after the removal of fixed nitrogen. The amounts of nifS1, nifH1, nifK1, nifE1, and nifX1 transcripts were normalized to those of nifB1 and compared to the wild-type strain, FD. (B) Nitrogenase activities of the wild-type strain, FD, and mutants BP706 and BP703 were measured by acetylene reduction 24 h after the removal of fixed nitrogen from the medium. Ethylene values were normalized to chlorophyll a (chla). (C) Growth of the wild-type and mutant strains was measured in cells grown for several generations in the absence of fixed nitrogen prior to the assay. The error bars indicate standard deviations.
Mutations in the predicted stem-loop structure that are highly conserved among heterocystous cyanobacteria were also tested to determine their effects on the transcript level, nitrogenase activity, and growth. A 5-bp alteration in mutant strain JU420 that affected only the shorter downstream 3′ stem-loop structure (Fig. 6A) had no effect on nitrogenase activity (Fig. 6B) or on growth (not shown) but resulted in a decrease in the nifH1 transcript level to about 50% of the wild-type level (Fig. 6C). An 8-bp alteration in strain JU422 that changed the sizes of both stem-loop structures and the spacing between them and increased the ΔG of the first predicted stem-loop structure (Fig. 6A) decreased nitrogenase activity, reflecting the greater reduction in the nifH1 transcript level than in JU420 (Fig. 6B). A 6-bp alteration in mutant strain JU423, which altered the sequence very near the processing site and greatly increased the ΔG of the first predicted stem-loop structure (Fig. 6A), had the greatest effect, with a 5-fold reduction in nifH1 transcript levels and a 40% reduction in nitrogenase activity compared to the wild-type strain (Fig. 6B and C). All three mutants gave 5′ RACE products that appeared to be the same size (Fig. 6D); however, the sequences of the 5′ ends of the transcripts revealed minor changes in the processed 5′ end for JU420 and multiple ends for JU422, while JU423 had a processing site that was 8 nucleotides upstream from the site in the wild-type strain (Fig. 6E). While these mutations in the conserved areas in the predicted stem-loop region all had effects on nifH1 transcript levels, like the mutant BP703 with an altered sequence in the 5′ stem-loop (Fig. 3A), the effects on nitrogenase activity and on growth were not severe. In strain JU423, which has a relatively unstable 5′ stem-loop structure and an altered sequence near the processing site (Fig. 6A), the effects were more deleterious, suggesting that the sequence near the processing site, as well as a stable stem-loop structure, is important for proper processing of the transcript, resulting in higher levels of the nifH1 transcript.
Effects on nifH1 expression of mutations in the conserved regions of stem-loop structures. (A) Locations of mutations in JU420, JU422, and JU423 and their predicted effects on RNA secondary structure. ΔG values are shown for only the first predicted stem-loop structure. The mutations are shown in the boxed regions. (B) The nitrogenase activities of the mutants were measured by acetylene reduction 24 h after the removal of fixed nitrogen from the medium. (C) The amounts of nifH1 and nifB1 transcripts were measured by RT-qPCR from RNA harvested from cells 24 h after the removal of fixed nitrogen. The amount of nifH1 transcript is shown relative to that of nifB1. (D) 5′ RACE products for the wild-type strain and each of the mutants. Because the 5′ end of nifH1 is a processed site, the 5′ RACE was done without TAP treatment of the RNA. (E) Sequences of the 5′ RACE products from panel D. The 5′ ends are underlined, and the mutations in JU422 and JU423 are shown in red (the JU420 mutation is further downstream). The error bars indicate standard deviations.
Antisense RNA in the nifU1-nifH1 intergenic region.The RNA-seq data for Anabaena sp. PCC 7120 show small amounts of an antisense RNA with an apparent 5′ end located 148 bp upstream of nifH (42), and a primary transcription start site was mapped to that site by RNA-seq using a method that enriched for primary transcripts (43). Therefore, we looked for antisense RNA in this region in A. variabilis. Preliminary experiments with various primer sets in the nifU1-nifH1 intergenic region using strand-specific cDNA identified an antisense transcript with a 5′ end at about 148 bp upstream from the start of nifH1 (data not shown). Using the nomenclature recommendation for small RNAs (55), we have named this small RNA sava4870.1 (see Materials and Methods). Primers within the nifU1 gene (Fig. 7B) were used to determine by RT-qPCR that sava4870.1 is between 164 and 215 bp in length (Fig. 7A); however, there also appeared to be low-level transcription of antisense RNA that extended further into the nifU1 gene (Fig. 7A). We compared the amounts of antisense transcript using primers that amplified strand-specific cDNA in cells grown with or without fixed nitrogen. Four sets of primers that amplified the strand-specific cDNA were used (Fig. 8D): (i) sava4870.1 antisense RNA (labeled “antinif”), (ii) antisense RNA beginning 29 bp upstream of the putative tss for the antisense RNA (R20), (iii) antisense RNA beginning 70 bp upstream of the putative tss for the antisense RNA (R21), and (iv) antisense RNA beginning 64 bp upstream of the start site of nifH1 to 156 bp into the nifH1 gene (nifH). The last three served as controls for sava4870.1 antisense RNA, since they were not within the sava4870.1 gene. RNA from cells grown without fixed nitrogen showed a 7-fold increase in expression of the antisense transcript for nifH1 (labeled nifH) compared to the control antisense RNA (R20 and R21) (Fig. 8A); however, there was little or no increase in expression of the sava4870.1 antisense RNA (antinif) compared to the control antisense RNA (R20 and R21) (Fig. 8A). In contrast, RNA from cells grown with fixed nitrogen showed increased expression of the sava4870.1 antisense RNA (antinif) compared to the control antisense RNA (R20 and R21) and nifH (Fig. 8B). Expression of the sava4870.1 RNA (antinif) in cells grown with fixed nitrogen compared to those grown without fixed nitrogen was at least 20-fold higher than the antisense transcripts (R20, R21, and nifH) (Fig. 8C). Thus, the sava4870.1 transcript is downregulated under conditions of nitrogen fixation, suggesting that it may play a role in inhibiting nif gene expression in cells growing with fixed nitrogen.
Size of the sava4870.1 antisense RNA. (A) Antisense RNA was measured by RT-qPCR using strand-specific cDNA. (B) Primer sets used to determine the 3′ end of sava4870.1. The tss for sava4870.1 antisense RNA is located 148 bp upstream of the start of nifH1 and 25 bp upstream from the tps for nifH1. The error bars indicate standard deviations. The colors of the products in panel A match the colors of the lines corresponding to the transcripts in panel B.
(A and B) Expression of sava4870.1 antisense RNA in cells grown without (−N) (A) or with (+N) (B) a source of fixed nitrogen. Antisense RNA was measured by RT-qPCR using strand-specific cDNA with primer sets that amplified the regions labeled antinif, R20, R21, or nifH, as shown in panel D. (C) Ratio of antisense RNA expression, +N versus −N, for each region shown in panel D. (D) Map of the nifU1-nifH1 intergenic region showing the locations of the RT-pPCR products. The error bars indicate standard deviations.
To determine the effect of loss of the sava4870.1 transcript, we created a sava4870.1 promoter mutant, strain BP712, in which a 22-bp deletion in the region immediately upstream of the sava4870.1 tss was replaced with an XbaI site. This mutation drastically reduced expression of the sava4870.1 RNA, as measured by expression of sava4870.1 in wild-type versus mutant cells grown with fixed nitrogen, but had little effect on expression of the sava4870.1 RNA in cells grown without fixed nitrogen, consistent with the fact that sava4870.1 RNA is expressed primarily in cells grown with fixed nitrogen (Fig. 9A). Since the mutation in BP712 also destroyed the region just upstream of the nifH1 processing site (the promoter deletion extends to −1 relative to the transcriptional processing site), we also determined its effect on expression of the nifH1 sense transcript. This mutation resulted in a 50% reduction in nifH1 expression (Fig. 9B), similar to the effect seen for mutations in the predicted stem-loop region, such as JU420 and JU422 (Fig. 6C). The only phenotype associated with the BP712 mutant was a delay of about 24 h before initiation of nitrogen fixation and diazotrophic growth; however, once growth began, the cells appeared to grow normally (data not shown). This phenotype of the BP712 mutant more likely reflects the decrease in expression of nifH1 sense transcript rather than the loss of the sava4870.1 transcript, since the antisense RNA is hardly made in the absence of fixed nitrogen. This conclusion is supported by the fact that in BP712 we were not able to detect a 5′ RACE product in RNA from cells grown without fixed nitrogen, suggesting that the nifH1 transcript is not processed normally in BP712 (data not shown). Further, the half-life of the sense transcript downstream from the transcriptional processing site in BP712, about 10 min, as determined by measuring the Loop 3′ product (Fig. 3A), was very similar to the half-life of the sense transcript upstream of the transcriptional processing site, as determined by measuring the Loop 5′ product (Fig. 3A and data not shown). The decrease in nifH1 transcript in the BP712 mutant (Fig. 9B), combined with the lack of a processed nifH1 transcript and the short half-life of the Loop 3′ product in the mutant compared to the long half-life in the same region of the wild-type strain, indicates that the mutation that abolished the promoter for sava4870.1 also had a major effect on the nifH1 transcript, probably at the level of transcript processing.
Expression of sava4870.1 antisense RNA in FD and in mutant BP712. (A) Expression of sava4870.1 antisense RNA was determined by comparing the amount of antisense RNA (antinif) to the amount of antisense RNA upstream of the antisense region (R21) in the wild-type (FD) and mutant (BP712) strains. (B) Expression of nifH1 sense RNA was determined by comparing the amount of nifH1 transcript to the amount of nifB1 transcript in the wild-type (FD) and mutant (BP712) strains. The error bars indicate standard deviations. (C) Map of the nifH1 region showing the primers used for the experiment in panel A. The sense primer for nifH1 (used in the experiment shown in panel B) is at the 3′ end of the gene, outside the region shown.
Processing of the nifD1, nifK1, and Ava_3931 transcripts.We used 5′ RACE, which can distinguish between processed and primary transcripts (58), to map the 5′ ends of the nifD1, nifK1, and Ava_3931 (the conserved gene downstream from nifK1) transcripts. In this technique, primary transcripts are treated with TAP, which hydrolyzes the 5′ triphosphate to a monophosphate so that it can be ligated to the RNA linker. However, processed transcripts do not require TAP treatment because they already have a 5′ monophosphate. 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 the nifD1, nifK1, and Ava_3931 transcripts. For all three transcripts, a band was visible from RNA samples that were not treated with TAP (Fig. 10A), indicating that the transcripts are processed from a larger transcript. Despite the diffuse nature of the nifD1 product, bands for both nifD1 and nifK1 were recovered from the gel and were successfully sequenced to identify the 5′ end. While sequence was obtained for the Ava_3931 product, the sequence at the 5′ end was not of sufficient quality to map the 5′ end, but we estimate that it is close to the start of Ava_3931. The locations of these three processing sites predict transcript sizes for nifH1, nifD1, and nifK1 (Fig. 10B) that are consistent with the sizes that were first reported in 1986 for the nifH, nifD, and nifK transcripts in Anabaena sp. PCC 7120 by Northern blotting (59).
Mapping nifD1, nifK1, and Ava_3931 transcript 5′ ends. (A) RNA isolated from cells grown without fixed nitrogen for 24 h was analyzed by 5′ RACE. A band for a processed transcript 44 bp upstream from the start of nifD1 and a band 131 bp upstream from the start of nifK1 were present with (+) or without (−) TAP treatment of RNA. A band for a processed transcript near the start of Ava_3931 was present with or without TAP treatment of RNA; however, we were unable to determine the 5′ end exactly. (B) Map of transcripts for nifH1, nifD1, nifK1, and Ava_3931 showing the tps and sizes of the deduced transcripts.
DISCUSSION
NifH plays a crucial role in nitrogenase assembly and function (15, 16, 30, 60). First, it is the enzyme that reduces dinitrogenase. As part of the mature nitrogenase complex, NifH is present in a 2:1 ratio to the NifD and NifK subunits. However, NifH also has a major role in the assembly of nitrogenase, specifically in the maturation of the P cluster and in the maturation and insertion of the M cluster (MoFe7S9C-homocitrate) into nitrogenase (16). Thus, it is not surprising that NifH is the most abundant nitrogenase protein. In other diazotrophic bacteria, nifH is the first gene on a larger transcript that includes nifD and nifK, so it was assumed that nifHDK in cyanobacteria also comprised an operon under the control of a strong promoter upstream of nifH (61, 62). However, as we have shown previously, nifH1 in A. variabilis is driven primarily by the distant nifB1 promoter, supplemented by a weak promoter in nifU1 (39). The abundance of the nifH1 transcript compared to nifB1, nifS1, and nifU1 cannot be attributed to the strength of the promoter but does correlate well with the stability of the nifH1 transcript, which is made by RNA processing at a site 123 bp upstream from the nifH1 start site. Although the sequences of the intergenic nifU-nifH regions of many heterocystous cyanobacteria are not well conserved and do not align well, manual alignment revealed conserved regions that produce a predicted stem-loop structure at the 5′ end of the nifH1 transcript that is very well conserved (see Fig. S1 in the supplemental material). Stem-loops stabilize transcripts when they are at the extreme 5′ end of the transcript (63). The stabilization of the 5′ end of the transcript when it is double stranded results from the mRNA not being recognized by 5′ exonucleases. According to Emory et al. (63), up to 2 unpaired nucleotides at the 5′ end of a transcript can be tolerated, while 5 or more unpaired nucleotides at the 5′ end of the transcript prevent a stem-loop from having any stabilizing effect. We analyzed mutations in the predicted stem-loop at the 5′ end of nifH1 to determine the relative importance of the sequence in this region versus the stem-loop structure for nifH1 transcript abundance. Based on several mutants, it appears that both the sequence and the structure contribute to the abundance of the nifH1 transcript, and loss of both reduces the amount of transcript by about 90%. Mutant BP706, which completely lacks the first stem-loop structure, had very little nifH1 mRNA and had slow growth under diazotrophic conditions. The JU423 mutation also had a very deleterious effect on mRNA abundance, probably because the first predicted stem-loop is not very stable (ΔG = −9 versus −21 for FD) and it has several mispaired bases, while the predicted stem-loop in the wild-type strain has only a single mispaired base. The decreased stability of the mutant stem-loop in JU423 increases the probability of a single-stranded target that is recognized for mRNA degradation. The altered stability of the stem-loop in JU423 not only led to a large decrease in transcript levels for nifH1 but also changed the processing site, which was shifted 8 nucleotides upstream compared to the wild-type strain. This suggests that the altered stem-loop structure in the mutant affected the cleavage site that yields the normal 5′-processed nifH1 transcript.
In the BP706 mutant, which completely lacks the first stem-loop structure, there was much less nifH1 transcript than in the wild-type strain, because the nifH1 transcript was more unstable, and the mutation had a polar effect, resulting in loss of the downstream nifK1 transcript (Fig. 5A), while the half-life of nifK1 in the mutant remained nearly the same as in the wild-type strain. While these results may seem contradictory, the quantity of a transcript may decrease while the half-life of the remaining transcript remains the same. Northern blot analysis in Anabaena sp. PCC 7120 identified nifH, nifHD, and nifHDK transcripts but not a nifBSUHDK transcript (59). This suggests that the processing event between the nifU1 and nifH1 transcripts in A. variabilis occurs more rapidly than the processing events within the nifHDK1 transcript. The nifK1 transcript has its own processing site (Fig. 10) that likely affects the half-life of the nifK1 transcript; however, the quantity of nifK1 transcript is sensitive to the stability of the upstream nifH1 transcript, because a destabilized nifHDK1 transcript reduces the abundance of all three individual transcripts. The half-life of the nifK1 transcript after it is processed from the nifHDK1 transcript should not be affected by the mutation in the nifH1 stem-loop structure, which we observed (Table 3). Hence the abundance of the nifK1 transcript is affected more than the half-life in the stem-loop mutants.
While the abundance of the nifH1 transcript was expected, it was surprising that, despite large reductions in the amount of nifH1 transcript in some mutants, there was much less effect on nitrogenase activity. The normal pattern of nitrogenase expression, with a strong peak at about 24 h after nitrogen step down followed by a rapid decline in activity, which remains relatively low (64), suggests that the cells initially make more transcript/enzyme than is required, modulating the amount as the cells become nitrogen replete and store fixed nitrogen in cyanophycin (65). Hence, they may make more enzyme than is required initially to replenish nitrogen reserves and then repress expression, while the mutants may continue to express nitrogenase much longer, which may explain the lack of a major defect in growth of some of the mutants under diazotrophic growth conditions.
Analysis of transcripts in the nifU1-nifH1 intergenic region revealed a small antisense transcript that began just upstream from the transcriptional processing site for nifH1 (but is transcribed in the opposite direction). This transcript is evident in the RNA-seq data for Anabaena sp. PCC 7120 (42), and its 5′ end was mapped in a global search for primary transcription start sites in the strain (43); however, neither its size nor its function has been described. It appears to be about 200 bp, extending well into the nifU1 gene, and is upregulated +N compared to transcripts upstream of the sava4870.1 promoter, though still present in very small amounts. While the amount of antisense transcript increases under −N conditions, that seems to be the result of increased expression of antisense RNA throughout the region, since there is also increased antisense RNA upstream of the sava4870.1 promoter. Compared to the upstream region, sava4870.1 RNA is more highly expressed +N than −N, and a mutation in the promoter for sava4870.1 affected expression only +N, not −N. A mutant lacking the promoter region for sava4870.1 showed greatly reduced nifH1 expression, suggesting that the antisense RNA is required for high-level expression of nifH1. Although it is possible that the antisense sava4870.1 RNA that is already present in heterocysts helps to stabilize the nifH1 transcript, leading to a decrease in nifH1 transcript in the mutant, this seems unlikely, because the mutant and wild-type strains show nearly identical levels of sava4870.1 antisense RNA under nitrogen-deficient growth conditions (Fig. 9A). More likely, the effect of the mutation may not be directly related to loss of the sava4870.1 antisense RNA. The sava4870.1 promoter mutation is very close to the processing site for the nifH1 transcript and did abolish processing of the nifH1 transcript. Further, a mutant with a 38-bp deletion that began 4-bp downstream from the end of nifU1, and thus was within the sava4870.1 gene, had no phenotype (data not shown), supporting the hypothesis that sava4870.1 is not directly involved in nifH1 expression. We also observed that there was a 50- to 100-fold increase in expression of nifH1 antisense transcripts in cells grown in the absence of fixed nitrogen compared to cells grown with fixed nitrogen. While the amount of antisense transcript compared to sense for nifH1 is very low (about 2.5% as much), it is difficult to imagine the role of this antisense transcript in the expression of nitrogenase genes.
In Anabaena sp. PCC 7120, the transcripts for nifHDK, as determined by Northern blots, suggested that they all derived from a single transcript initiated upstream of nifH. The nifH probe hybridized to transcripts of about 4.7 kb, 2.8 kb, and 1.1 kb; nifD hybridized to transcripts of 4.7 kb and 2.8 kb, and nifK hybridized to the 4.7-kb transcript (59). These sizes correspond to transcript junctions between nifH and nifD and between nifD and nifK. If the nifB promoter is the primary promoter for transcription of the entire nif cluster, then it seems unlikely that termination of transcription at these junctions is the only mechanism of transcript regulation, although it could play a role, since the amount of transcript for genes downstream of nifK1 is much less than for nifK1 itself (46). However, we determined that the transcript made from the nifB1 and internal nifU1 promoters is processed just upstream of nifD1, nifK1, and Ava_3931 (Fig. 10), resulting in predicted transcript sizes that are consistent with the reported sizes for the nifH, nifD, and nifK1 transcripts in Anabaena sp. PCC 7120 (59). However, the presence of a strong 2.8-kb band that hybridizes to nifH and nifD in Anabaena sp. PCC 7120 (59) suggests that processing between nifH and nifD is not efficient. The region around the 5′ end of nifD1 has no strong predicted secondary structure; however, the 5′ end of nifK1 is within a strong predicted single stem-loop structure (ΔG = −31.7). The sizes and abundances of the nifD1 and nifK1 transcripts in A. variabilis may reflect a combination of termination of transcription, transcriptional processing, and transcript stability; however, the relative amounts of nifH1, nifD1, and nifK1 transcripts normalized to nifB1 are roughly proportional to their half-lives (Table 3), suggesting that stability is very important in maintaining the stoichiometry of the three transcripts. The 7-fold decrease in the amount of nifE1 transcript compared to nifK1 (46), however, suggests that this large drop may result from both transcriptional termination downstream of nifK1 and the much shorter half-life of nifE1 than of nifK1 (Table 3). Despite multiple attempts, we have not been successful in finding the 5′ end of nifE1 by 5′ RACE.
The data presented here support the conclusion that expression of the nifH1, nifD1, and nifK1 genes, encoding the nitrogenase proteins that are required in the greatest abundance for nitrogen fixation, depends on transcript processing. Further, the stability of the nifH1, nifD1, and nifK1 transcripts correlates with the amount of transcript (46), and at least for nifH1, sequences near the transcriptional processing site are important in processing and transcript stability. The predicted secondary structures in the intergenic regions of nifU1-nifH1 and nifD1-nifK1 may play a role in the regulation of the transcript abundances of nifH1 and nifK1; however, at least in the case of nifH1, the sequences around the processing site appear to be equally important for transcript stability.
ACKNOWLEDGMENT
Support for this research was provided by National Science Foundation grant MCB-1052241.
FOOTNOTES
- Received 29 December 2014.
- Accepted 30 January 2015.
- Accepted manuscript posted online 9 February 2015.
- Address correspondence to Teresa Thiel, thiel{at}umsl.edu.
Citation Pratte BS, Ungerer J, Thiel T. 2015. Role of RNA secondary structure and processing in stability of the nifH1 transcript in the cyanobacterium Anabaena variabilis. J Bacteriol 197:1408–1422. doi:10.1128/JB.02609-14.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.02609-14.
REFERENCES
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.