hetR and patS, two genes necessary for heterocyst pattern formation, are widespread in filamentous nonheterocyst-forming cyanobacteria

Many cyanobacteria can fix N₂, a process catalysed by nitrogenase, an enzyme complex highly sensitive to oxygen, whereas oxygen is produced by cyanobacteria as byproduct of photosynthesis. Cyanobacteria adopt two major strategies to resolve the problem of incompatibility between oxygenic photosynthesis and oxygen-sensitive nitrogen fixation, by separation of these two activities either in time or in space, or a combination of both strategies (Berman-Frank et al., 2003). In the model organism Anabaena sp. strain PCC 7120 (hereafter Anabaena 7120), the spatial separation of these two activities is achieved by the differentiation of 5–10 % of the cells in each filament into heterocysts, cells that specialize in nitrogen fixation (Meeks & Elhai, 2002; Zhang et al., 2006a). A micro-oxic environment within heterocysts protects nitrogenase against inactivation by oxygen (Wolk et al., 1994).

Heterocyst differentiation requires a number of genes involved in the different steps of the developmental process (Meeks & Elhai, 2002; Zhang et al., 2006a). At least in Anabaena 7120, heterocyst differentiation is triggered by the accumulation of 2-oxoglutarate, which acts as a signal of deprivation of combined nitrogen (Chen et al., 2006; Laurent et al., 2005; Li et al., 2003). One of the receptors of the 2-oxoglutarate signal could be NtcA, a transcription factor highly conserved in cyanobacteria (Herrero et al., 2004; Laurent et al., 2005). Another critical factor for the initiation of heterocyst differentiation is HetR, a protease with DNA-binding activity (Buikema & Haselkorn, 1991; Huang et al., 2004; Zhou et al., 1998b). HetR has been reported so far in heterocyst-forming strains, in filamentous strains that fix N₂ and in Arthrospira platensis, a nondiazotrophic filamentous strain (Zhou et al., 1998a). At the maturation stages of heterocyst differentiation, two envelope layers, composed of polysaccharides and glycolipids, are formed; genes encoding nitrogenase are expressed, and heterocysts become functional (Meeks & Elhai, 2002). In many heterocyst-forming strains, heterocysts are spaced semi-regularly along each filament, corresponding to a one-dimensional pattern (Meeks & Elhai, 2002; Zhang et al., 2006a). The formation of the heterocyst pattern possibly requires interaction between HetR and PatS (Huang et al., 2004). PatS, or a pentapeptide of PatS (PatS-5), has been proposed to be a diffusible molecule acting as an inhibitor of heterocyst differentiation (Yoon & Golden, 1998). According to the current model, HetR activates the expression of patS, and PatS or a derivative of PatS diffuses laterally to inhibit the differentiation of the neighbouring cells by acting negatively on the DNA-binding activity of HetR (Zhang et al., 2006a). Another gene, hetN, is required for the maintenance of heterocyst pattern (Callahan & Buikema, 2001). Both patS and hetN are necessary to prevent heterocyst differentiation in the presence of a combined nitrogen source (Borthakur et al., 2005).

How different forms of cells and organisms emerged during evolution remains a challenging question in biology. Evolutionary developmental biology (known as evo-devo biology) explores the molecular basis of the emergence and evolution of the form and shape of living organisms (Goodman & Coughlin, 2000). Heterocysts are interesting for evo-devo studies because they may have appeared in evolution between 2450 and 2100 Ma based on indirect fossil record and phylogenetic analysis (Tomitani et al., 2006), making them one of the oldest forms of differentiated cells to have appeared on the Earth. Although we have an idea about when heterocysts appeared during evolution (Tomitani et al., 2006), it remains unknown how they appeared. In this study, we show that hetR and patS are found not only in nitrogen-fixing filamentous cyanobacteria, but also in filamentous strains that do not fix N₂. The implication of these findings is discussed.

Sequence analysis.
Genome data of cyanobacteria were collected from the Integrated Microbial Genomes (IMG) website (). Genomic data of Arthrospira platensis were kindly made available by Beijing Genomic Institute. The 16S rRNA sequences and HetR sequences used for phylogenetic analysis were obtained from GenBank. Phylogenetic analyses were performed with MEGA 4.0 (Tamura et al., 2007) by using the minimum evolution (ME) method (Rzhetsky & Nei, 1992) together with the close-neighbour-interchange (CNI) searching algorithm (Nei & Kumar, 2000). The 16S rRNA tree was inferred with the Kimura two-parameter substitution model (Kimura, 1980), while the HetR tree was inferred with the Poisson correction substitution model (Zuckerkandl & Pauling, 1965).

Bacterial strains and culture conditions.
All cyanobacterial strains related to this study are listed in Table 1. Schizothrix calcicola, Oscillatoria lutea, Lyngbya sp. and Phormidium mucicola were obtained from the culture collection of the Institute of Hydrobiology, Chinese Academy of Sciences; Arthrospira platensis was obtained from the culture collection of the Institute of Botany, Chinese Academy of Sciences and was grown in modified SAG medium as described by Zhang et al. (2006b). Anabaena 7120 was grown in BG11 with nitrate as source of combined nitrogen. For cultures under conditions of combined-nitrogen starvation, nitrate was omitted from the BG11 medium (BG11₀) or SAG medium. All cyanobacterial strains were grown at 28 °C.

Table 1. Relevant features of cyanobacterial strains related to this study

PCR for hetR and immunodetection of the HetR protein.
Two degenerate PCR primers, PDhetR-up and PDhetR-dw (Table 2), were deduced from conserved motifs of HetR proteins and were used to amplify hetR sequences. The procedure for immunodetection was as described previously (Kuhn et al., 2000), using a polyclonal antibody prepared against HetR of Anabaena 7120.

Table 2. DNA primers used in this study

Plasmid constructions.
We first constructed pRL25Z, a modified version of the replicative vector pRL25c in Anabaena 7120 (Wolk et al., 1988). For this purpose, pRL25c was digested with BamHI and EcoRI. The larger DNA fragment was purified and ligated with an artificial fragment, which was generated by annealing two oligonucleotides (GATCCCCGGG and AATTCCCGGG), and thus had a BamHI site at one end and an EcoRI site at the other end and a SmaI site in the middle. The resulting vector pRL25Z has a size of 9507 bp and a linker with unique cloning sites for BamHI, SmaI and EcoRI. All PCR primers are listed in Table 2. The hetR gene of Arthrospira platensis (hetR_ar) was obtained by PCR with the primers ParhetR498m and ParhetR1285, and the patS_ar-hetR_ar gene cluster of Arthrospira platensis was obtained with primers ParhetR979m and ParhetR1285. The PCR products were digested with BamHI and inserted into the BamHI site of pRL25Z, leading to pRar and pSRar, respectively (see Fig. 3). To obtain the construct pSar (Fig. 3), the hetR_ar corresponding sequence in pSRar was removed by digestion with EcoRI. The promoter region of hetR_ar was amplified with primers ParhetR498m and ParhetR9 and cloned at the SmaI site of pGFPTX (derived from pGFPuv by replacing the region of the lac promoter on pGFPuv with the T4 transcriptional terminator) to produce pGFPparhetR. pGFPparhetR was linearized with XmnI and inserted into the SmaI site of pRL25Z, resulting in pRarGfp (Fig. 3). The promoter region of patS_ar together with the first 100 bp of its coding region was amplified with primers ParpatS434m and ParpatS100, digested with HindIII and KpnI, and ligated into the same sites of pGFPTX; the resulting plasmid was then linearized with EcoRI and inserted into the EcoRI site of pRL25Z, leading to pSarGfp (Fig. 3). All pRL25Z-derived plasmids were introduced into Anabaena 7120 by conjugation (Wolk et al., 1988).

(62K): Fig. 3. (a) The genomic region including patSar-hetRar in Arthrospira platensis (top) and various plasmids carrying different regions. Green colour represents the gfp gene fused to the promoter region of hetRar or patSar (see Fig. 4 for GFP fluorescence). (b) Phenotypes of the Anabaena strains expressing hetRar and/or patSar after 24 h of combined-nitrogen deprivation. Three constructs were transferred by conjugation into various strains as shown: pSRar (b1, b2), pSar (b3, b4) and pRar (b5, b6). The wild-type (b1, b3, b5), a hetR mutant (b2, b6), and a patS mutant (b4) were used in this analysis. Arrowheads indicate heterocysts. wt, wild-type.

Microscopy.
A Nikon DXM 1200 digital camera mounted on a Nikon Eclipse E800 microscope was used to capture images as described before (Sakr et al., 2006). The same microscopic setting was used to acquire series of images. For fluorescent images, areas under the microscope were chosen under bright-field and were excited at 465 nm only during the time of image acquisition to avoid unnecessary bleaching. HetR orthologues are widespread among both diazotrophic and nondiazotrophic filamentous cyanobacteria
The hetR gene was first identified as a key regulator of heterocyst differentiation and found also in filamentous cyanobacteria that are able to fix molecular nitrogen without heterocyst formation (Buikema & Haselkorn, 1991). It was used as a marker for phylogenetic analysis of nitrogen-fixing filamentous cyanobacteria (Mes & Stal, 2005). Similar genes are absent in all unicellular cyanobacteria whose genomes have been sequenced. One report also mentioned the existence of a protein in Arthrospira platensis that could react with antibody raised against the HetR of Anabaena PCC 7120 (Zhou et al., 1998a), but it remained unknown whether it represented an isolated case or a general observation in filamentous cyanobacteria. In this study, we used a polyclonal antibody raised against HetR of Anabaena 7120 to assess the presence of HetR in five nondiazotrophic filamentous cyanobacteria (Table 1), Schizothrix calcicola, Oscillatoria lutea, Lyngbya sp., Phormidium mucicola and Arthrospira platensis. All five strains displayed a cross-reactivity with the antibody (Fig. 1). To further confirm these data, partial sequence of hetR was successfully amplified from S. calcicola, using degenerate oligonucleotide primers designed according to conserved regions of known HetR sequences. A HetR-encoding sequence was also identified from the genome of A. platensis. These data clearly indicate that HetR homologues are widespread in both diazotrophic and nondiazotrophic filamentous cyanobacteria.

(33K): Fig. 1. Immunoblot analysis using antibodies against HetR of Anabaena 7120. (a) Detection of HetR from several cyanobacteria. Lanes 1 to 4: protein samples extracted from Schizothrix calcicola, Oscillatoria lutea, Lyngbya sp. and Phormidium mucicola, respectively; lane 5, protein extract from Anabaena 7120. (b) Increase in HetR level in Arthrospira platensis following nitrogen starvation at 0 h, 18 h, and 42 h respectively. A similar amount of total protein was loaded on the first three lanes. rHetR, recombinant HetR of Anabaena 7120 as control.

In Anabaena 7120, the level of the HetR protein is increased following limitation of combined nitrogen (Buikema & Haselkorn, 1991; Zhou et al., 1998a). Using immunoblotting, we found that the levels of HetR_ar (HetR of Arthrospira platensis) increased in cells of A. platensis incubated in the absence of a combined nitrogen source for 18 or 42 h (Fig. 1b). The level of HetR_ar was found to be highest in filaments at 48 h of incubation in the absence of combined nitrogen, and then decreased (data not shown). These results show that although A. platensis is not able to fix N₂ or to form heterocysts, the level of HetR_ar is upregulated similarly as in heterocyst-forming cyanobacteria.

The HetR sequences identified here, together with some representative HetR sequences previously published, were used to perform a phylogenic analysis (Fig. 2). This analysis shows that HetR from heterocyst-forming cyanobacteria forms a monophyletic clade distinct from that of non-heterocyst-forming cyanobacteria, suggesting that hetR genes found in the latter strains are unlikely to be the results of horizontal gene transfer from heterocyst-forming cyanobacteria.

(30K): Fig. 2. Phylogenetic relationship of HetR sequences analysed by the minimal evolution (ME) method. The corresponding gi numbers of sequences are labelled on the left of the genus names. The numbers in the major branches are the bootstrap values calculated from 1000 replicate trees.

hetR_ar is adjacent to a putative patS sequence in Arthrospira platensis and the corresponding genomic region is functional in regulating heterocyst development in Anabaena 7120
The amino acid sequence of HetR_ar shares 76 % identity with HetR of Anabaena 7120. Since no reliable genetic system exists for filamentous nondiazotrophic cyanobacteria such as Arthrospira platensis, we used Anabaena 7120 as a heterologous system to get insight into the possible function of hetR_ar. A genomic region including hetR_ar was cloned in pRL25Z, a replicative vector in Anabaena 7120. The construct, pSRar (Fig. 3), was then transferred into both the wild-type and the hetR mutant of Anabaena 7120 by conjugation. Surprisingly, the presence of this plasmid suppressed heterocyst differentiation in the wild-type and was unable to complement the hetR mutation (Fig. 3). As controls, a plasmid bearing a copy of hetR of Anabaena 7120 could complement the defect of the hetR mutant in heterocyst differentiation, whereas the same plasmid in the wild-type strain resulted in Mch (multiple contiguous heterocysts) phenotypes as reported previously (Buikema & Haselkorn, 1991).

Further examination of the DNA fragment in pSRar revealed that the ORF upstream of hetR_ar could encode a small protein of 84 amino acid residues with the sequence of the last five residues identical to that of the PatS pentapeptide of Anabaena 7120 (Fig. 3). To test whether this ORF could confer a PatS-like function, two constructs were made, one with hetR_ar alone (pRar, Fig. 3), and another with the ORF corresponding to the putative patS gene (pSar). In the presence of pSar, neither the wide type nor the patS mutant of Anabaena 7120 formed heterocysts in the presence or absence of a combined-nitrogen source, whereas the patS mutant gave rise to Mch phenotypes as expected (Fig. 3). Thus the presence of the upstream ORF had a heterocyst-suppressing activity. We therefore designated this ORF patS_ar. Similarly, the construct with hetR_ar alone led to a Mch phenotype when present in the wild-type of Anabaena 7120, and could complement the hetR mutant of the same organism (Fig. 3). Based on these results, we concluded that hetR_ar and patS_ar from Arthospira platensis could regulate heterocyst differentiation in Anabaena 7120.

Although the products of hetR_ar and patS_ar could function in Anabaena 7120, their promoter regions show no obvious similarity to those of hetR and patS in Anabaena 7120, raising the question of whether the transcriptional control could operate on hetR_ar and patS_ar in Anabaena 7120. Therefore, the putative promoter region of hetR_ar and that of patS_ar were fused to a gfp reporter. The two corresponding constructs, pRarGfp and pSarGfp (Fig. 3), were transferred into Anabaena 7120 by conjugation. After induction by nitrogen starvation, green fluorescence emitted by GFP could be detected more strongly in heterocysts than in vegetative cells for Anabaena 7120 bearing pSarGfp (Fig. 4). GFP fluorescence observed with the strain bearing pRarGfp displayed also a strong upregulation after combined-nitrogen stepdown; this upregulation was not confined specifically to heterocysts, and became less strong after 8 h of incubation under conditions of combined nitrogen starvation (Fig. 4). These results showed that both patS_ar and hetR_ar displayed activation by combined-nitrogen deprivation in Anabaena 7120, but only patS_ar displayed a cell-type-specific upregulation. This result is consistent with the observation obtained with Anabaena 7120 that cell-type-specific expression of hetR is not necessary for heterocyst formation (Black et al., 1993; Buikema & Haselkorn, 2001).

(67K): Fig. 4. Analysis of promoter regions of patSar and hetRar fused to gfp reporter in Anabaena 7120. The corresponding constructs are depicted in Fig. 3. (a, b) pSarGfp in a patS mutant at 24 h after nitrogen stepdown. (c–h) pRarGfp in the wild-type at time 0 h (c), 3 h (d), 8 h (e), 24 h (f) or 48 h (g, h) after combined-nitrogen stepdown. (a) Filament pictured under bright-field; (b, c, d, e, f, h) green epifluorescence emitted by GFP; (g) red epifluorescence emitted from photosynthetic pigments of the same filaments as in (h), to highlight heterocysts that display a dimmer red fluorescence as compared to vegetative cells. Arrowheads indicate heterocysts.

Putative patS-like sequences are widespread among filamentous cyanobacteria
Potential PatS-like peptides were scanned bioinformatically in 18 cyanobacterial strains whose genomes have been completely sequenced. These strains include (Table 1): Anabaena variabilis ATCC 29413, Nostoc punctiforme PCC 73102, Trichodesmium erythraeum IMS 101, Arthrospira platensis, Crocosphaera watsonii WH8501, Synechocystis PCC 6803, Gloeobacter violaceus PCC 7421 and Thermosynechococcus elongatus BP-1, Synechococcus PCC 6301, Synechococcus PCC 7942 and 8 Prochlorococcus strains (Prochlorococcus marinus SS120, Prochlorococcus marinus MED4, Prochlorococcus marinus NATL2A, Prochlorococcus marinus MIT 9313, Prochlorococcus marinus MIT 9312, Synechococcus WH8102, Synechococcus CC9902 and Synechococcus CC9605). Those sequences with an RGSGR motif embedded within the protein sequence were excluded. Several possible patS-like genes were identified. Such putative patS-like genes were not detected in the unicellular cyanobacterial genomes, but were present in all filamentous cyanobacteria whose genome sequences were available (Fig. 5). While, in the heterocyst-forming cyanobacteria such as Anabaena 7120, Anabaena variabilis or N. punctiforme, the patS gene could encode a peptide of 17 or 13 amino acids depending on the initiation codon that would actually be used, those deduced from non-heterocyst-forming cyanobacteria could encode a much larger peptide (Fig. 5). The patS-like gene in Arthrospira platensis could encode a peptide of 84 residues, and that in T. erythraeum could encode a peptide of 90 residues. The latter one was annotated as hypothetical (GenBank accession no. ABG51177). Fig. 5 shows the position of the hetR locus relative to that of patS on different genomes. In A. platensis and T. erythraeum, the patS gene and hetR genes form a cluster and are separated by 459 bp and 1284 bp, respectively, with no other ORF detected in the intergenic regions. In Anabaena 7120 and A. variabilis, these two genes are located on opposite DNA strands and are separated by about 50 kb. In N. punctiforme, they are located very far away from each other, at a distance of about 4.5 Mb.

(18K): Fig. 5. (a) PatS or PatS-like sequences deduced from the genomes of several filamentous cyanobacteria (Np, Nostoc punctiforme; A7120, Anabaena 7120; Av, Anabaena variabilis; Ap, Arthrospira platensis; Te, Trichodesmium erythraeum). The last five residues conserved among the sequences are in bold. (b) Relative location of patS and hetR on the chromosomes of different strains.

While this paper was in preparation, several partially sequenced cyanobacterial genomes became available. In three of them, namely Lyngbya sp. PCC 8106, Microcoleus chthonoplastes PCC 7420 and Synechococcus sp. PCC 7335, ORFs similar to both hetR and patS, clustered on the chromosome similarly as in Arthrospira platensis, could also be identified. The putative hetR gene and the putative patS gene are separated by 435 bp, 277 bp and 401 bp in Lyngbya sp. PCC 8106, Microcoleus chthonoplastes PCC 7420 and Synechococcus sp. PCC 7335, respectively (data not shown). While both Lyngbya sp. PCC 8106 and Microcoleus chthonoplastes PCC 7420 are filamentous, non-diazotrophic strains, the position of Synechococcus sp. PCC 7335 in the classification of cyanobacteria is still ambiguous (Table 1). Even though Synechococcus sp. PCC 7335 is considered as a unicellular strain, its 16S rRNA sequence belongs to a lineage of filamentous strains (Honda et al., 1999; Wilmotte & Herdmann, 2001). The hetR gene has been extensively studied in Anabaena 7120 as a key regulator of heterocyst development. The presence of hetR in all filamentous strains examined in this study gives the possibility that it might have been laterally transferred from heterocyst-forming cyanobacteria. However, this is not supported by phylogenetic analysis and the fact that its presence among filamentous strains appears to be general rather than exceptional. Consistently, the monophyletic clade of heterocyst-forming cyanobacteria as based on HetR sequences also suggests that the hetR gene originated before the divergence of heterocyst-forming and non-heterocyst-forming cyanobacteria but at or after the time of the appearance of filamentous cyanobacteria, because hetR is not found in any of the unicellular cyanobacteria so far examined. The only exception is Synechococcus PCC 7335, which appears to be more related to filamentous strains than unicellular strains based the phylogenetic trees obtained with 16S rRNA sequences (Honda et al., 1999; Wilmotte & Herdmann, 2001). A recent study suggests that heterocyst-forming cyanobacteria probably emerged between 2450 and 2100 Ma (Tomitani et al., 2006), so the hetR gene may have originated at least before this lower time limit. Having experienced such a long period of evolutionary history, the HetR proteins remain highly conserved throughout the whole sequence, in particular those residues critical for various HetR functions (Risser & Callahan, 2007), implying that HetR must play a very critical role, not limited to heterocyst differentiation as earlier studies suggested (Buikema & Haselkorn, 1991).

We found that in several filamentous cyanobacteria the last five residues of a peptide encoded by a gene adjacent to hetR were identical to the functional pentapeptide of PatS. Three arguments support the idea that these genes correspond to patS. First, similar sequences were found in several filamentous cyanobacteria, often with similar genetic organization relative to hetR. Second, the patS-like sequence of Arthrospira platensis has heterocyst-suppressing activity when cloned in Anabaena 7120. Finally, patS_ar displayed a cell-type-specific expression in heterocysts in Anabaena 7120. These results suggest that the two alien genes, hetR_ar and patS_ar, can interact with their native partners in the control of heterocyst development.

In Anabaena 7120, the expression of hetR was upregulated after combined-nitrogen deprivation. In two non-heterocyst-forming diazotrophic strains Trichodesmium and Symploca, the expression of hetR also seemed to be affected by nitrogen status (El-Shehawy et al., 2003; Janson et al., 1998). We found that the level of HetR increased in Arthrospira platensis following combined-nitrogen removal. Unlike unicellular strains, filamentous cyanobacteria may need a certain degree of intercellular cooperation for a better control of nitrogen distribution along filaments under nitrogen-limiting conditions. The semi-regular distribution of nitrogen-fixing heterocysts along filaments in some modern cyanobacteria such as Anabaena 7120 may be considered as the best example of such intercellular interactions. One observation in support of such a hypothesis is that Arthrospira platensis could survive nitrogen limitation for at least a month, with most cells showing weak red fluorescence indicative of extensive pigment degradation, while a few single cells along the filament retained a high level of pigment fluorescence (data not shown). This result could imply that a filament may have evolved a mechanism to ensure the survival of at least some of the cells under adverse conditions. Since genetic manipulation of filamentous nondiazotrophic cyanobacteria remains difficult, whether hetR is involved in such a mechanism remains unknown.

We would like to thank Professor Huanming Yang for the access to the genomic data of Arthrospira platensis, and Dr Annick Wilmotte for helpful discussion. This study was supported by the CNRS and the Agence Nationale de la Recherche (ANR), through the PCV (Physique et Chimie du Vivant) program.

Edited by: A. Wilde