|Home | About | Journals | Submit | Contact Us | Français|
The filamentous, heterocystous, nitrogen-fixing cyanobacterium Nostoc sp. strain PCC 7120 may contain, depending on growth conditions, up to two hydrogenases directly involved in hydrogen metabolism. HypC is one out of at least seven auxiliary gene products required for synthesis of a functional hydrogenase, specifically involved in the maturation of the large subunit. In this study we present a protein, CalA (Alr0946 in the genome), belonging to the transcription regulator family AbrB, which in protein-DNA assays was found to interact with the upstream region of hypC. Transcriptional investigations showed that calA is cotranscribed with the downstream gene alr0947, which encodes a putative protease from the abortive infection superfamily, Abi. CalA was shown to interact specifically not only with the upstream region of hypC but also with its own upstream region, acting as a repressor on hypC. The bidirectional hydrogenase activity was significantly downregulated when CalA was overexpressed, demonstrating a correlation with the transcription factor, either direct or indirect. In silico studies showed that homologues to both CalA and Alr0947 are highly conserved proteins within cyanobacteria with very similar physical organizations of the corresponding structural genes. Possible functions of the cotranscribed downstream protein Alr0947 are presented. In addition, we present a three-dimensional (3D) model of the DNA binding domain of CalA and putative DNA binding mechanisms are discussed.
Cyanobacteria and green algae are the only organisms known which are equipped with the combination of oxygenic photosynthesis and hydrogenases (19, 21, 39, 44, 53). This combination makes it possible for cyanobacteria to potentially produce hydrogen gas (H2) from solar energy and water, making them an ideal candidate for production of a sustainable energy carrier needed for the future (48). In nature, cyanobacteria recycle the energy-rich H2 and therefore no net production can be detected (25, 48). To create a redundancy of H2 from a cyanobacterial system, the pathways and regulation of the hydrogen metabolism have to be further explored so that the obtained knowledge can be used in metabolic engineering. The field of key regulators of hydrogenases in cyanobacteria is just starting to be explored, and only a few regulators of gene expression involved in hydrogen metabolism are known. Once more light is shed on functions and mechanisms in transcriptional regulation, it will be possible to monitor and modify the bacterial system for higher yield of a desired product, in our case H2.
The model organism used in this study is the filamentous and heterocystous cyanobacterium Nostoc sp. strain PCC 7120 (23). All cyanobacterial hydrogenases are classified as [NiFe] hydrogenases, and when grown under N2-fixing conditions Nostoc PCC 7120 expresses two types of hydrogenases, a bidirectional enzyme existing in vegetative cells and heterocysts and an uptake hydrogenase located in the heterocysts only (20). The uptake hydrogenase, encoded by the hupSL operon, recycles the H2 produced as a by-product from the nitrogenase under N2 fixation. The bidirectional hydrogenase, encoded by hoxEFHUY, consists of two individually regulated clusters, with hoxEF in one and hoxUYH in the other (45). HoxYH forms the small and large subunits of the hydrogenase, and HoxEFU forms the diaphorase part (41, 42). The bidirectional hydrogenase is not connected to N2 fixation and can either split or form H2, depending on the redox potential (48, 50). The biological function of the bidirectional hydrogenase is not fully understood, but there are three main possibilities: it may function as a valve for low-potential electrons generated during the light reaction of photosynthesis (3), be responsible for H2 oxidation in the periplasm and electron delivery to the respiratory chain (42), or remove the excess reductants under fermentative, anoxic conditions (48, 50, 52).
Hydrogenases consist of at least two subunits. The active site is located in the large subunit whereas the [FeS] clusters, transferring the electrons in and out of the active site, are positioned in the small subunit (7, 56). To produce a functional and mature hydrogenase, several maturation proteins are required, those encoded by hypFCDEAB (hydrogenase pleiotropic) and a hydrogenase-specific protease encoded by either hupW (for the uptake hydrogenase) or hoxW (for the bidirectional hydrogenase) (7, 13, 48, 56). HypC functions as a chaperone and keeps the precursor of the large subunit in a conformation accessible for metal incorporation (7). How the maturation is achieved for the small subunit is not known. In Nostoc PCC 7120 the extended hyp operon, including the six hyp genes, hypFCDEAB; the open reading frame (ORF) asr0697 located between hypD and hypE; two downstream ORFs; and five upstream ORFs have all been shown to be part of the same transcriptional unit covering a distance of 14 kb (1). Several transcriptional start points are present within the extended hyp operon, e.g., one 475 bp upstream of hypC (1). Moreover, the five genes upstream of hypFCDEAB were suggested to be involved in the assembly of the [FeS] clusters of the small subunit (1).
In the present study we address the question of transcription factors directly involved in the regulation of the maturation proteins of hydrogenases in Nostoc PCC 7120. The DNA binding protein CalA (cyanobacterial AbrB-like, Alr0946 in the genome) was found to bind specifically to the upstream region of hypC. Among cyanobacteria with sequence genomes, most strains harbor two copies of homologues of cal, which are classified into two clades, A and B (22). Little is known about CalA, and we examine in this work the regulation of calA as well as characterization of the protein, including a putative three-dimensional (3D) model of the DNA binding domain and the similarities to and differences from the AbrB protein in Bacillus subtilis. Putative functions of the cotranscribed downstream protein Alr0947 are also discussed.
The filamentous heterocystous cyanobacterium Nostoc sp. strain PCC 7120, also known as Anabaena sp. strain PCC 7120 (Nostoc PCC 7120), was cultured in BG110 supplemented with 5 mM NH4Cl and 5 mM MOPS (morpholinepropanesulfonic acid; pH 7.8) or in BG110 (38), sparged with air, and grown at 25°C, with a continuous irradiance of 40 μmol of photons m−2 s−1 (46). In the Northern blot experiments some of the cultures of Nostoc PCC 7120 were grown in darkness and sparged with argon 12 h before harvesting. For the real-time quantitative PCR (RT-qPCR) experiments, two overexpression strains of Nostoc PCC 7120 harboring the self-replicating pnir vector (see below) with alr0946 (N7120 OV) or without it (N7120 EV) and the wild-type strain were grown in BG110 supplemented with 5 mM NH4Cl and 5 mM MOPS for 6 days. The cells were harvested by centrifugation at 5,000 × g for 5 min at 4°C, washed three times with BG110, and transferred to BG110 medium supplemented with additional NaNO3 to a final concentration of 6 mM, which induced the overexpression of CalA under the control of the nirA promoter, and were left to grow for another 4 days. Part of the cultures was harvested at day −2 (2 days before changing the nitrogen source to NaNO3), day 0, day 2, and day 4. Escherichia coli strains used in the cloning work and overexpression of His-tagged CalA were grown at 37°C in liquid or solid LB medium with the respective antibiotics added.
Soluble proteins were extracted from Nostoc PCC 7120 cultures grown in BG110 supplemented with 5 mM NH4Cl and 5 mM MOPS (pH 7.8) for 5 days. The cells were harvested by centrifugation at 5,000 × g for 10 min at 4°C and resuspended with the ratio 1:1 (vol/vol) in protein buffer (100 mM Tris-HCl, pH 7.5, 1 mM EDTA, 2 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF], 10% glycerol, 0.5% Triton X-100). The cells were mixed with glass beads (0.3 g per 1 ml cells) and lysed with a Precyllis24 lyser/homogenizer (Berlin Technologier) six times for 30 s, resting for 1 min on ice between the intervals. A second centrifugation step was performed at 14,000 × g for 45 min at 4°C. The protein extract was quickly frozen in liquid nitrogen and stored at −80°C.
Biotin-tagged DNA fragments from the upstream region of hypC (Fig. (Fig.1A)1A) were amplified by PCR with the specific primers listed in Table Table1.1. Streptavidin-covered magnetic beads (Dynabeads M-280; Dynal Biotech) were used according to the manufacturer's instructions. The incubation of the DNA fragments with the magnetic beads and the isolation of the protein-DNA complexes were carried out as described previously (31). The proteins were denatured at 95°C for 5 min in sodium dodecyl sulfate (SDS) sample buffer and separated on a 10 to 15% gradient SDS-polyacrylamide gel (SDS-PAGE). The gel was stained with Coomassie blue according to the manufacturer's instructions. The proteins were identified by mass spectrometry as described previously (32).
PCR amplifications were performed as described previously (1). Obtained DNA fragments were isolated from the agarose gels using the GFX PCR DNA and gel band purification kit (GE Healthcare, Uppsala, Sweden) following the manufacturer's instructions. All primers, except the tag primer (see below), were designed with the Primer3 program (http://frodo.wi.mit.edu/primer3/) and checked for their specificity by BLAST search against the Nostoc PCC 7120 genome. All oligonucleotides used are listed in Table Table1.1. The secondary structure was analyzed with a Primer design utility program (http://www.cybergene.se/EazyPrimer.htm). The tag primer used in cDNA synthesis was designed with the program Tagenerator (26). Sequencing reactions were performed at Macrogen Inc. Computer-assisted sequence analyses were performed using the BioEdit sequence alignment editor version 184.108.40.206.
In Northern blot analyses formaldehyde/denaturing gels were used to separate the total RNA, 7 μg per lane, using the protocol provided with the Hybond-N+ nylon membrane (GE Healthcare). Hybridizations were performed at 65°C, using a probe of genomic DNA from Nostoc PCC 7120 obtained by PCR and gene-specific oligonucleotides (Table (Table1).1). The Rediprime II random prime labeling system (GE Healthcare) with [α-32P]dCTP was used according to manufacturers' instructions. To ensure an even loading of total RNA, the visual appearance of rRNA on the agarose gel was checked, as well as using a probe of genomic DNA from Nostoc PCC 7120 obtained by PCR with gene-specific primers for the constitutive RNA component of the ribozyme RNase P (57). Reverse transcription-PCRs (RT-PCRs) were performed as described previously (1). The reverse primer in the RT-PCR includes a tag marked with an open diamond in Fig. Fig.2A.2A. No negative control to test for gDNA contamination is needed since the tag in the reverse primer can be amplified only from cDNA. In RT-qPCR 1 μg total RNA from each sample was used for cDNA synthesis using RevertAid Moloney murine leukemia virus (M-MuLV) reverse transcriptase according to the manufacturer's instructions (Fermentas). To verify primer specificity, PCR products were cloned using the TOPO-TA cloning kit (Invitrogen) and sequenced. To ensure single DNA fragment amplifications, a determination of the melting point was done at the end of every RT-qPCR experiment. Amplification data were analyzed by the iQ5 optical system software (Bio-Rad), all samples were run in technical duplicates, and standard deviations were calculated. The relative expression was calculated using the threshold cycle (ΔΔCT) algorithm and normalized against the wild-type strain, and the 23S transcript was used as a reference. Relative gene expression calculations were performed as described previously (37).
A putative transcription start point (TSP) was located with the 5′ RACE (rapid amplification of cDNA ends) system, version 2.0 (Invitrogen), using 2.5 μg total RNA and gene-specific primers (Table (Table1)1) following the manufacturer's instructions. The resulting PCR product was cloned into the pCR2.1-TOPO vector (Invitrogen) according to the manufacturer's instructions before being sequenced.
calA was amplified from genomic DNA with PCR by using oligonucleotides including the restriction sites for BamHI and HindIII (Table (Table1),1), before being cloned into pCR2.1-TOPO according to the manufacturer's instructions (Invitrogen). The product was confirmed by sequencing and ligated into pQE-30 (Qiagen) and transformed into the M15(pREP4) strain (LB Amp100 Km25) (Qiagen). After confirmation of the successful transformation by sequencing of the overexpression, purification procedures (Ni-NTA Superflow Resin) were performed according to the manufacturer's instructions (Qiagen). The purity of the obtained CalA His-tagged protein was examined on a 12% SDS-polyacrylamide gel with staining by Coomassie blue.
calA was amplified from genomic DNA with PCR by using oligonucleotides including the restriction sites for BamHI and EcoRI (Table (Table1)1) and cloned into pCR2.1-TOPO according to the manufacturer's instructions (Invitrogen). The product was confirmed by sequencing and subcloned into the self-replicating plasmid pnir4 (kindly provided by Dominique Desplancq, Ecole supérieure de Biotechnologie de Strasbourg, France) at EcoRI and BamHI sites (12) and transformed by triparental mating into Nostoc PCC 7120 (12, 15). An empty pnir vector was transferred in the same way as a control (N7120 EV). Selection of mutants was carried out on BG11 plates supplemented with 25 μg/μl neomycin, and single colonies were transferred to liquid BG11 medium supplemented with the same concentration of neomycin. In order to confirm that the cells were transformed with the vector, DNA extractions, followed by PCR analysis, were carried out by using oligonucleotides designed at both sides of the multiple cloning sites (Table (Table1).1). The obtained overexpressing strain used to express CalA was named N7120 OV. The overexpression was induced by changing the nitrogen source of the medium from 5 mM NH4Cl (5 mM MOPS, pH 7.8) to BG110 supplemented with 6 mM NaNO3.
DNA fragments used were obtained by PCR (Table (Table1)1) and end labeled with [γ-32P]ATP (GE Healthcare) according to the manufacturer's instructions. Twenty femtomoles of labeled DNA fragment was incubated with different amounts of purified CalA each in a 20-μl reaction mixture, in a buffer described previously (31). After incubation at 30°C for 30 min, the reaction mixtures were separated on a 6% (vol/vol) polyacrylamide nondenaturing gel by electrophoresis and the relative positions of the isotope were detected with a Pharos FX Plus molecular imager (Bio-Rad).
The activity of the bidirectional hydrogenase in Nostoc PCC 7120 was determined by measuring the amount of H2 evolved from 3 ml cells with a chlorophyll a (Chla) concentration of 30 μmol/ml in 8-ml glass vials with addition of methyl viologen reduced by sodium dithionite as described previously (49). The Chla measurements were performed as described previously (28). The H2 content was quantified using a gas chromatograph (Clarius 500) with a Molecular Sieve 5A 60/80 mesh column (Perkin-Elmer) and argon as the carrier gas. Two biological samples were used, each with technical duplicates. To obtain one curve for N7120 OV and N7120 EV, respectively, the mean values of the technical duplicates were calculated, then the mean values of the biological duplicates, and finally the standard deviations. After a lag phase of 20 min, the hydrogen evolution was linear in both N7120 OV and N7120 EV during the measured period of time (80 min). The values presented are relative values of N7120 OV divided by N7120 EV at each time point.
Sequence homology searches were made in the BLAST program (2), and sequence analyses were done with CLUSTAL W (51). Searches for conserved domains in proteins annotated as proteins with unknown function were performed in Cyanobase (40), SwissProt/TrEMBL, and NCBI (sequences available on http://www.expasy.org/sprot/ and in the GenBank database, http://www.ncbi.nlm.nih.gov/GenBank/index.html). 3D models of CalA (23) were constructed by using the online program SWISS-MODEL (http://swissmodel.expasy.org//SWISS-MODEL.html) and with AbrB from Bacillus subtilis (1Z0R, Protein Data Bank [PDB]) (4) as the template. Only the cyanobacterial AbrB (CyAbrB) domain sequence (amino acids [aa] 73 to 125) was used for the alignment and the 3D model. The results were visualized using the online program Swiss-PDB-viewer (http://www.expasy.org/spdbv ). For protein sequence analysis CalA from Nostoc PCC 7120 (GI|17130292, protein accession number BAB72903) was used together with AbrB from Bacillus subtilis (GI|113009, protein accession number P08874, PDB number 1Z0R). The consensus sequences of the AbrB family (PF04014) used in this study are as presented in Pfam (http://pfam.sanger.ac.uk/) and are made up either of 145 randomly selected sequences (done by Pfam) or with all 893 sequences in the database. All alignments were performed in BioEdit version 220.127.116.11 (18) using CLUSTAL W multiple alignment (51). For pattern and profile searches the MyHits Motifscan and InterPro Scan online programs were used (30, 35).
To identify proteins interacting with the upstream region of the hydrogenase maturation gene hypC and the operon starting with hypC, a protein-DNA affinity assay was performed with crude protein extract from Nostoc sp. strain PCC 7120 (Nostoc PCC 7120) grown in BG110 supplemented with 5 mM NH4Cl and 5 mM MOPS, pH 7.8 (38). In the cell extracts from the non-nitrogen-fixing culture five proteins, with molecular masses ranging from approximately 11 to 20 kDa, were found to interact with the 793-bp upstream region of hypC (Fig. (Fig.1B).1B). All five bands were excised and analyzed by mass spectrometry. The peptides were identified as a biotin carboxyl carrier protein, the phycocyanin β-subunit, the phycocyanin α-subunit, CalA (Alr0946 in the genome), and streptavidin. All identified DNA binding proteins with the exception of CalA are artifacts from the experimental procedure; see reference 31.
CalA is annotated as a hypothetical protein, although it has been identified before in proteomic studies (46), consisting of 144 aa and with a predicted molecular mass of 16.06 kDa (see Fig. S1A in the supplemental material) (40). Bioinformatic analysis showed that it contains a conserved DNA binding domain (aa 73 to 125) (IPR006339, TIGR01439), which classifies it as a member of the protein family of transcription factors called AbrB (antibiotic resistance) (38) (sequences available on http://www.expasy.org/sprot/). The conserved domain in CalA is located in the C-terminal part of the protein, which is typical for AbrB homologues in cyanobacteria (see Fig. S1A) (32, 33). CLUSTAL W alignments of CalA and homologues show that the protein is highly conserved in cyanobacteria, with similarities ranging from 60 to 99%. In all sequenced cyanobacterial strains, with the only exception being Trichodesmium erythraeum, the respective homologues to calA are followed by ORFs being homologues to alr0947. The Alr0947 homologues are also highly conserved proteins with similarities ranging from 41 to 92% on the protein level (see Fig. S1B).
To determine whether there are more genes transcribed together with calA and to identify putative TSPs, RT-PCR, Northern blotting, and 5′ RACE experiments were performed. calA and the downstream ORF alr0947 were shown to form an operon with two TSPs identified upstream of calA, TSP1 and TSP2 (−599 bp and −42 bp with respect to the translational start point of calA), and in the near vicinity both TSPs have putative extended −10 boxes, in the format TGNTAN3T (Fig. 2A and B) (54). Northern blotting results show a very prominent band of approximately 2,100 nucleotides (nt), which in size matches a transcript from TSP1 (Fig. 2A and C). The transcript level of calA was investigated in N2-fixing and non-N2-fixing cultures which were either constantly grown in light and oxic environments or transferred to darkness and sparged with argon 12 h before harvest. Under all four growth conditions examined, the calA transcript was abundant. However, the expression levels were qualitatively lower in cultures, both N2 fixing and non-N2 fixing, which were exposed to darkness and anoxic conditions 12 h before being harvested.
The cotranscribed alr0947 encodes a protein annotated as a hypothetical protein of unknown function. It has a conserved domain located in the C-terminal part of the protein and belongs to the Abi (abortive infection) superfamily, and many of its homologues in other cyanobacteria are annotated as abortive infection proteins (sequences available in the GenBank database [http://www.ncbi.nlm.nih.gov/GenBank/index.html] and on http://www.ebi.ac.uk/interpro/). Members of this protein family are metal-dependent proteases related to CAAX prenyl proteases, which are a large and diverse family of putative membrane-bound proteins (IPR003675/PF02517) with eight membrane-spanning regions (sequences available at http://www.expasy.org/sprot/ and http://www.ebi.ac.uk/interpro/). Characteristic for proteins belonging to this family is the much-conserved Glu-Glu motif in the end of the sequence as well as the two highly conserved histidine residues putatively involved in zinc binding. Zinc ions are often part of zinc finger motifs which interact with DNA and are found in proteins with regulatory functions (sequences available in the GenBank database [http://www.ncbi.nlm.nih.gov/GenBank/index.html]) (9, 27). Available experimental data and analysis of conserved CAAX motifs suggest that these metal-dependent proteases are involved in protein and/or peptide secretion and modification (36). The usual function of the protease is a proteolytic cleavage between the Cys and the three remaining residues, AAX, on the C-terminal end of a farnesylated protein (36).
To confirm the direct interaction between CalA and the upstream region of the hydrogenase maturation gene hypC, EMSAs were performed. His-tagged CalA was overexpressed in Escherichia coli, purified (Fig. (Fig.1C),1C), and mixed with a DNA fragment (F1R2, 717 bp) identical to the one used in the protein-DNA assay and an unrelated DNA fragment (346 bp) from PUC19 amplified by PCR (Table (Table1).1). CalA was detected to bind specifically to the F1R2 segment (Fig. (Fig.1D,1D, lanes 4 to 6) with no affinity of CalA to the unrelated DNA (Fig. (Fig.1D,1D, lanes 1 to 3). In order to identify the specific binding region, F1R2 was further divided into two segments, F1R1 (406 bp) and F2R2 (352 bp). EMSAs showed a clear retardation signal when F1R1 was used (Fig. (Fig.1E,1E, lanes 4 to 6), whereas no interaction was observed with F2R2 (Fig. (Fig.1F,1F, lanes 4 to 6).
To examine whether CalA interacts with its own upstream region, a 709-bp DNA fragment directly upstream of calA was amplified by PCR (Fig. (Fig.2A;2A; Table Table1).1). EMSAs were performed with increasing amounts of CalA with a clear shift already apparent after the addition of 50 ng CalA (Fig. (Fig.2D2D).
In order to study the regulation of CalA, both deletion mutants and strains overexpressing CalA were constructed. Although several attempts were made, we were unsuccessful in creating viable colonies of a fully segregated deletion mutant. However, the overexpression strain in Nostoc PCC 7120 (N7120 OV) under the control of the nirA promoter was viable and functioning. A strain containing an empty pnir vector without calA (N7120 EV) was constructed and used as a negative control in addition to wild-type Nostoc PCC 7120.
The transcription levels of calA, alr0947, hypC, and hoxE in the overexpression strain N7120 OV were studied with RT-qPCR at four different time points, −2, 0, +2, and +4 days in relation to the induction of the nirA promoter with NO3− (Fig. (Fig.33 and and4).4). The expression of calA in N7120 OV was strongly upregulated after 2 days of NO3− induction, and after 4 days the expression level was 25 times higher than on day −2 in the wild-type control (Fig. (Fig.3).3). The expression patterns of calA in the two control strains were steady when grown with either NH4+ or NO3−. The transcription of alr0947 was clearly inhibited after 2 days of induction by NO3− in N7120 OV and was steady in the two control strains. With increasing expression levels of calA the relative expression level of alr0947 decreased to 18% and 12% after the 2nd and 4th days, respectively, compared to the expression at day −2 in the wild-type control. The expression levels of hypC also decreased to 25% after 2 days and 14% after 4 days of induction compared to the expression level at day −2 in the wild-type control (Fig. (Fig.4).4). For hoxE no clear transcriptional regulation related to the overexpression of CalA could be detected. However, the transcription of hoxE seems to be generally upregulated when the nitrogen source is changed from NH4+ to NO3−.
A functional hydrogenase requires the action of several so-called Hyp proteins. Since the transcription of hypC is negatively affected by additional CalA, we analyzed the effect of the hypC downregulation on the bidirectional hydrogenase activity. Interestingly, CalA reduces substantially the hox activity (Fig. (Fig.5),5), demonstrating that there is a correlation between the transcript level of hypC and the bidirectional hydrogenase activity under the growth condition examined. The relative activity of N7120 OV compared to N7120 EV decreased from 120% at day 0 to 53% and 68% at days +2 and +4, respectively.
Bioinformatic analysis showed that CalA in Nostoc PCC 7120 contains an AbrB domain (aa 73 to 125) (see Fig. S1A in the supplemental material). However, several differences were observed compared to the homologue in Bacillus subtilis, the most notable being that the DNA binding AbrB domain, usually positioned in the N-terminal sequence, is present in the C-terminal sequence of CalA (see Fig. S1A). Several of the functionally important arginines found in AbrB from Bacillus subtilis are also missing in CalA (Fig. (Fig.6A)6A) (4). Despite the differences concerning the location of the domain within the protein, several similarities regarding the structure could clearly be identified. When considering the 3D model, which is based on the amino acid sequence of the CyAbrB domain in CalA and the crystal structure 1Z0R from Bacillus subtilis, it seems apparent that CalA acts as a dimer (Fig. 6B and C). This is supported by the fact that the AbrB domain can be found only once within the protein and that the total amino acid sequence of CalA is too short to allow the protein to form the DNA binding region as a single protein (10). Furthermore, the loop-hinge regions typical for the AbrB superfamily (Lys9-Val10 and Lys31-Asp32) also seem to be conserved to some extent, which enables a flexible binding to DNA (Fig. (Fig.6B)6B) (4, 55). Analysis of the surface electrostatic potential of the putative DNA binding domain in CalA indicates that the surface is mainly positively charged (Fig. (Fig.6C6C).
In this study we present evidence that the DNA binding protein CalA (Alr0946 in the genome) in Nostoc sp. strain PCC 7120 (Nostoc PCC 7120) interacts specifically with the upstream region of hypC, encoding an auxiliary protein necessary for maturation of hydrogenases (Fig. (Fig.1B).1B). The other four proteins picked up in the protein-DNA affinity assay are either artifacts from the method, as is the case with streptavidin, which covers the magnetic beads; results of sample contamination by phycobilisome proteins (phycocyanin β-subunit and phycocyanin α-subunit), which are very abundant in cyanobacteria (8, 11, 58, 60); or a result of the natural specific binding between streptavidin and biotin (biotin carboxyl carrier protein).
That the CyAbrB protein CalA is involved directly or indirectly in regulating cyanobacterial hydrogen metabolism is supported by results from studies of Synechocystis sp. strain PCC 6803 where its homologue Sll0359 was found to act as an activator of the hox operon which encodes the bidirectional hydrogenase (32, 33). AbrB is a well-studied DNA binding protein in Bacillus subtilis, where the AbrB protein directly or indirectly is regulating more than 60 genes involved in different physiological and metabolic pathways (47). Interestingly, homologues to CalA in cyanobacteria have been shown to regulate genes encoding key proteins in carbon metabolism, nitrogen fixation, and production of the toxin cylindrospermopsin, indicating that CyAbrB proteins also are involved as transcription factors in several physiological and metabolic processes (22, 24, 43). CyAbrB homologues usually exist in multiple copies in the genome and are classified into two distinct clades, A and B (22). The Nostoc PCC 7120 homologue of CalA, All2080 (CalB), is also annotated as a hypothetical protein, but slightly shorter, 137 aa, truncated in the last 7 aa of the C-terminal end.
In EMSA the purified His-tagged CyAbrB protein CalA was shown to interact specifically with the upstream region of hypC. CalA was found to bind specifically to the first part of the examined upstream region, F1R1. When F1R1 was further divided into shorter parts, it was not possible to detect any specific binding. Attempts to find a specific binding sequence for CyAbrB proteins in cyanobacteria also failed (22, 32). Our result is supporting the hypothesis from studies of Bacillus subtilis which suggests that AbrB, instead of binding to a specific DNA sequence, recognizes a specific stretch of the three-dimensional structure of the DNA helix (5, 6). In contrast, the AbrB homologue in Bacillus amyloliquefaciens inhibits the transcription of phytase by binding to two specific DNA sequences located in the upstream region of the corresponding structural gene phyC (29).
hypC is cotranscribed with the downstream genes, and it is therefore likely that CalA is regulating the full operon. The hypothesis that CalA indeed functions as one of the key players in regulating hydrogen metabolism in cyanobacteria is in line with our results. The CalA homologue in Synechocystis PCC 6803, Sll0359, was shown to interact with the hox upstream region as well as with its own upstream region and acts as an activator of hox transcription (32). In our study of Nostoc PCC 7120 the expression level of hoxE, encoding one of the diaphorase subunits of the bidirectional hydrogenase, first decreases in the overexpression strain in parallel with a lower NH4+ concentration and then increases strongly when NO3− is added. This increase is also observed in the control strains and is likely a result of the change of nitrogen source and corresponding nitrogen metabolisms. On the protein level overexpressed CalA is present in low levels before day 0, demonstrating that pnir is not completely silent in the presence of NH4+ (data not shown). The larger amount of hoxE transcript before induction with NO3− at day 0 in N7120 OV might be a result of this. From the relative expression level in Nostoc PCC 7120, we cannot observe any direct positive or negative regulation of CalA on hoxE when moved to NO3−, and a possible regulation by CalA will probably be masked by the strong increase in expression created by the NH4+/NO3− shift. Results from semiquantitative RT-PCR show that the expression pattern of hoxY, encoding the small subunit of the hydrogenase, is decreasing in N7120 OV following the same trend as hypC (data not shown). In Synechocystis PCC 6803 the activity of the bidirectional hydrogenase was shown to increase with additional amounts of CyAbrB (32). Therefore, it is important to note two differences between the model organisms, Nostoc PCC 7120 and Synechocystis PCC 6803. In Nostoc PCC 7120 the hox genes are present in two clusters with several transcriptional start points (45) which may be regulated individually, whereas in Synechocystis PCC 6803 there is only one hox cluster (17, 31). The second difference is that Nostoc PCC 7120 is an N2-fixing strain with an H2 metabolism involving both the bidirectional hydrogenase and an uptake hydrogenase, while Synechocystis PCC 6803 is a non-N2-fixing strain with only the bidirectional hydrogenase. One should also bear in mind that except from CyAbrB proteins CyLexA has been found to interact with the upstream regions of the hox operon in Synechocystis PCC 6803 and Nostoc PCC 7120 (31, 32, 33, 45).
Furthermore, CalA also binds to the upstream region of the calA-alr0947 operon. To investigate the effects of the increased level of CalA on the operon in N7120 OV, it is not possible to study the expression level of calA since the transcript level originating from the chromosomal copy will be masked by the strong increase originating from the overexpressing vector pnir harboring calA. In N7120 OV a strong increase in transcript level is observed +2 days after induction of pnir with NO3−, resulting in a decreased alr0947 transcript level. Thus, overexpression of CalA downregulates the expression of alr0947, which suggests that CalA acts as a repressor of its own operon. Negative feedback of transcriptional regulation is a well-documented mechanism and is for example used to regulate the concentration of the transcription factor so that it stays within certain levels (59).
calA is cotranscribed with the downstream ORF alr0947, and the operon has two identified TSPs, −42 bp and −599 bp, relative to the translational start point of calA. The prominent band of approximately 2,100 nt in the Northern blot results with a calA probe matches the size of a transcript from the TSP positioned furthest away from calA (−599 bp). The band is very strong, indicating a high abundance of transcript, which is in compliance with proteomic studies where CalA is found in large amounts (46; M. Ekman and K. Stensjö, personal communication). The band of approximately 2,100 nt is stronger for the cultures grown in an N2-fixing environment than for the cultures grown in the non-N2-fixing one. This is in accordance with two quantitative proteomic iTRAQ studies of Nostoc PCC 7120 and Nostoc punctiforme ATCC 29133, showing a relative higher level of CalA in N2-fixing filaments than in the non-N2-fixing filament (46) and an upregulation of 1.67 times of the homologue to CalA, NpR5944, in the heterocysts of Nostoc punctiforme (34), respectively. Generally the transcript levels of calA are high for all conditions investigated, but they are relatively lower when induced in darkness and an anoxic environment overnight compared to cultures grown in a light and oxic environment, independent of the nitrogen status of the culture. This may indicate additional functions for CalA coupled to activities in light. The second transcript originating from TSP2 is present at much lower levels under the growth conditions examined.
BLAST searches of cyanobacterial homologues of CalA show that all, with the exception of Trichodesmium erythraeum, are followed by a homologue to Alr0947. This conserved pattern and the fact that the DNA binding protein CalA and the putative protease Alr0947 are transcribed as one operon lead to the question if they also work together in the same regulatory network. Since CalA is binding to its own upstream region, it is possible that the protease Alr0947 is regulating the concentration of CalA by proteolytic cleavage, but if so, this might function in a yet-unknown way since no CAAX motif is found in CalA.
CalA has homologues in all sequenced cyanobacteria, and CLUSTAL W alignments show that they are all conserved (22). The homologues in filamentous strains are almost 100% identical on the amino acid level, but even in unicellular cyanobacteria there is a high similarity ranging from 60 to 70%. This conserved pattern and the fact that we and others (22, 32, 33) did not manage to produce knockout mutants indicate that the protein may be essential for the organism. In Synechocystis PCC 6803 only a partly segregated knockout mutant of sll0359 has been able to survive, supporting the importance of the protein for the organism (32). In contrast, it was possible to create a knockout mutant of the second CyAbrB protein encoded by sll0822 in Synechocystis PCC 6803. It was found playing an important role regulating some nitrogen-related genes (22). In this study we focus only on CalA since it was this homologue that was found interacting with the upstream region of hypC in Nostoc PCC 7120.
In CalA and its homologues in other cyanobacteria, the DNA binding domain is located in the C-terminal part of the protein (aa 73 to 125). This location makes CyAbrBs different from the AbrB homologue found in Bacillus subtilis (sequences available in the GenBank database [http://www.ncbi.nlm.nih.gov/GenBank/index.html] and on http://www.expasy.org/sprot/), where the DNA binding domain is located in the N-terminal part (aa 1 to 53) (47). Of the four arginines (R) existing in AbrB from Bacillus subtilis (Arg8, Arg15, Arg23, and Arg24) and which are considered crucial for the DNA binding, only Arg8 might have a counterpart, Arg7, in the CyAbrB domain in CalA (Fig. (Fig.6A)6A) (4, 55). The missing arginines obviously do not hinder CalA from binding to DNA but rather might alter the way that it is done. The arginine residues Arg15, Arg23, and Arg24 are in cyanobacteria exchanged for glutamine, threonine, and lysine, respectively. In comparing the amino acid sequence of CalA with several other sequences of proteins from the AbrB family, it becomes apparent that both glutamine (Q) and lysine (K) are common substitutes for arginine in several proteins.
The 3D model shows that CalA acts as a dimer. This is in agreement with results from studies made on one of its homologues, Sll0822, in Synechocystis PCC 6803 (22) and is further supported by 3D modeling with the solution structure of the monomer PHS018 from Pyrococcus horikoshii (2GLW.pdb) (data not shown). Since the loop-hinge regions seem to be conserved to some extent, CalA might bind to DNA in a flexible way and thus to several different regulatory regions (10, 14). Since no consensus sequence for DNA binding has been identified in CalA or in the AbrB homologue in Bacillus subtilis, the ability of the DNA to adopt a particular conformation or the topology of the DNA region itself seems to be the way that the protein finds the right spot (4). The surface electrostatic potential of the putative DNA binding domain of CalA is mostly positively charged, which is interesting since an opposite charge to the negative DNA helix is usually considered beneficial for establishing strong DNA binding. This differs from AbrB from Bacillus subtilis, which has both negatively charged and positively charged areas.
The observation that while the transcript level of hypC decreased, the bidirectional hydrogenase activity decreased to 53% in N7120 OV compared to N7120 EV 2 days after induction with NO3− demonstrates that the bidirectional hydrogenase is affected by the overexpression of CalA. Four days after induction the bidirectional hydrogenase activity has increased in N7120 OV to 68% of N7120 EV, which might be an effect of the higher transcript level of the bidirectional hydrogenase in all strains under the growth conditions examined. The downregulation of the bidirectional hydrogenase activity might be indirect due to smaller amounts of maturation proteins and/or might be through a direct regulation of CalA on the hox genes. The result is interesting in the aspect of hydrogen metabolism and possibilities to increase H2 production by the regulation of specific transcription factors.
This work was supported by the Swedish Energy Agency, the Knut and Alice Wallenberg Foundation, the Nordic Energy Research Program (project BioH2), the EU/NEST FP6 project BioModularH2 (contract 043340), and the EU/Energy FP7 project SOLAR-H2 (contract 212508).
Published ahead of print on 18 December 2009.