Background

In cyanobacteria, the ancestors of plant chloroplasts, the role of calcium ions (Ca2+) in abiotic stress response was identified by the recombinant expression of bioluminescent Ca2+ sensors. A transient increase in the intracellular Ca2+ concentration ([Ca2+]i) from a resting level of about 0.2 μM up to 4 μM free Ca2+ was observed upon light-to-dark transitions and during temperature, salt and osmotic stress [1,2,3]. The involvement of Ca2+ has also been established in cyanobacterial growth [4], regulation of reactive oxygen species (ROS) [5], fine-tuning of the carbon (C) and nitrogen (N) balance [6, 7], heat stress acclimation [8], exopolysaccharide production [9], and fatty acid and hydrocarbon composition [10].

According to their morphology, cyanobacteria are classified into unicellular and multicellular (filamentous) species [11]. The filamentous Nostoc sp. PCC 7120 (hereafter designated Anabaena) represents a group of organisms of special interest because of their ability to fix atmospheric N under combined N-limiting growth conditions, which occurs within specialised cells called heterocysts (reviewed in [12]). A prominent role of Ca2+ in Anabaena is in the regulation of heterocyst formation [13], which is thought to occur through the activity of the cyanobacterial Ca2+-binding protein (CcbP), which binds Ca2+ via negative surface charges [14, 15]. In N-limiting conditions, CcbP was reportedly strongly downregulated, both at the expression level by the transcriptional regulator NtcA, and at the protein level through HetR-mediated proteolysis. The decrease in CcbP abundance resulted in an increase in free Ca2+ in differentiating cells 5 to 6 h after removal of combined N [16]. HetR also acts as a transcription factor, which regulates expression of several genes involved in the commitment of a vegetative cell to differentiation into a proheterocyst and maturation into a functional heterocyst. This genetic reprogramming includes inhibition of cell division and formation of the heterocyst envelope, comprising a gas-impermeable glycolipid layer and outer polysaccharide layer [17,18,19]. This developmental process occurs in about every tenth cell of a filament under N-deprived conditions, due to the action of heterocyst pattern formation proteins such as the small peptide PatS, which is expressed mainly in heterocysts [20, 21] and diffuses into adjacent cells where it inhibits the activity of HetR [22].

Proper heterocyst development in Anabaena is closely coupled with filament integrity. Several mutant strains of Anabaena that are defective in heterocyst differentiation and function exhibit a fragmented filament phenotype upon N deprivation [23,24,25,26,27,28,29,30,31]. Proteins required for both filament integrity and heterocyst development in the absence of combined N include cell envelope components [31], in particular the SepJ (also called FraG) protein, which is localised to the septum structure between cells, as well as a series of other “Fra” proteins [25, 28, 32,33,34]. Recently the gene cluster fraCDEF was identified, from which the corresponding proteins FraCDE promote filament elongation, whereas FraF restricts filament length [29]. FraCD and SepJ were shown to be involved in the formation of septum-localised channels for communication, and for exchange of reduced C (sugars) and combined N metabolites (amino acids) between CO2-fixing vegetative cells and N-fixing heterocysts. Mutants lacking these proteins fragmented and became unviable upon the shift to N-deficient media due to the malformation of septal structures [35,36,37].

We recently identified the Ca2+-binding protein Ca2+ Sensor EF-hand (CSE) to be highly conserved in filamentous, heterocystous cyanobacteria [7]. Here we describe mutant strains of Anabaena lacking CSE, which demonstrated severe filament fragmentation and were impaired in heterocyst formation and function. We propose a role for CSE in transducing the Ca2+ signal required for early heterocyst differentiation, which implicates Ca2+ and CSE in responding to and restoring the C/N balance in N-fixing cyanobacteria.

Results

Deletion of CSE leads to filament fragmentation and compromised N-fixing ability

Two independent asr1131 deletion clones (Δcse1 and Δcse2) were produced and used to inoculate two separate cultures. Complete segregation of both cultures was confirmed at both DNA and RNA levels (Fig. 1). Both Δcse cultures grown in BG11 were composed of predominantly short filaments, compared to the primarily long filaments in the WT strain (Fig. 2a and b). In addition to the short filaments, Δcse cultures grown in BG11 in 3% CO2 contained a small population of long filaments (Fig. 2b) that resembled WT filaments, including occurrence of heterocyst cells. Growth for 4 days in BG110 medium lacking any combined N source led to an increase in the relative abundance of long filaments containing heterocysts in Δcse (Fig. 2d and f), in contrast to the short filaments that were prevalent in BG11 medium. Aggregated clusters of long Δcse filaments were harvested from BG110, transferred to either BG110 or BG11 and incubated as previously. After 5 days, short filaments began to appear in the BG11 cultures, and after a further 2 to 3 days, these became the dominant phenotype in the culture, whereas short filaments were detected in BG110 after 10 days (results not shown). Transformation of the Δcse mutant with a plasmid expressing asr1131 under the control of its native promoter abolished the short filament phenotype, with filaments in the complemented strain resembling those in the WT (Fig. 2g). Filament length counts revealed that 93% of the “filaments” in Δcse in BG11 comprised only 1–5 cells (Fig. 3), while over 98% of filaments contained less than 20 cells. Only 0.5% of filaments in Δcse had more than 80 cells. In contrast, about 70% of WT filaments constituted > 20 cells and 17% comprised > 80 cells in BG11 medium. Under N-fixing conditions (BG110 medium), the majority of WT filaments comprised 16–30 cells, while the proportion of Δcse filaments comprising 1–5 cells decreased in BG110 to 52% of total filaments. Conversely, long filaments were more abundant in Δcse cultures growing in BG110, in comparison to BG11 medium (Fig. 3).

Fig. 1
figure 1

Construction and verification of the Δcse mutant. a Scheme of the Δcse vector construct, with a kanamycin/neomycin resistance cassette (KmR/NmR) replacing the cse gene. b Polymerase chain reaction (PCR) confirmation of two independently-obtained, fully segregated Δcse clones. Amplification of the cse gene with its upstream and downstream flanking regions in Anabaena wild-type (WT) resulted in a PCR product of 3593 bp. In the Δcse clones, replacement of cse with KmR/NmR resulted in a larger PCR product of 4075 bp. No traces of the WT PCR product in the Δcse clones indicated full segregation of the mutant. c Expression of cse in WT and Δcse clone 1 normalised to the expression of the reference gene rpoA. Error bars indicate the standard deviation from three biological replicates (n = 3)

Full size image
Fig. 2
figure 2

Phenotype of the Δcse mutant. Bright-field micrographs of four-day old cultures of wild-type (WT; left), Δcse (right) and Δcse::pBG2089 (g) growing in 3% CO2 in regular BG11 medium (a, b, g) or in BG110 medium lacking combined nitrogen (c and d). Alcian Blue stains (e and f) were used to visualise heterocysts and proheterocysts, indicated by carets, in long filaments

Full size image
Fig. 3
figure 3

Filament length counts. Range and frequency of filament lengths in wild-type (WT) and Δcse cultures grown in BG11 or BG110 in 3% CO2, expressed as a proportion of the total number of filaments counted

Full size image

The Δcse mutant strains were grown alongside WT in BG11 growth media, as well as in BG110 lacking any combined N source (Fig. 4). Protein, chlorophyll and total pigment contents were equivalent in both strains after 5 days in BG11 (Fig. 4a, b and e). In contrast, Δcse cultures had very poor growth rates in BG110, compared to the WT (Fig. 4c and d), and developed a yellow colouration after 2 to 3 days that corresponded with decreased contents of chlorophyll (peak at 680 nm) and phycocyanin (peak at 635 nm) relative to absorption at wavelength 750 nm (Fig. 4e and f).

Fig. 4
figure 4

Growth phenotype of the Δcse mutant under different growth conditions. Growth curves a-d of wild-type (WT) and Δcse cultures monitored using total proteins (a and c) and chlorophyll concentrations (b and d) under 3% CO2. Cultures were grown either in regular BG11 medium (a and b), or in BG110 medium lacking combined nitrogen (c and d). Data points represent mean values from three biological replicates (n = 3), error bars show standard deviations. Significant differences between WT and Δcse samples are indicated with asterisks (t-test P <  0.05). Absorption spectra of WT and Δcse cultures were recorded after growing cultures for 5 days in BG11 (e) or BG110 (f)

Full size image

Nitrogenase activity was assayed by the conversion of acetylene to ethylene. In N-replete medium, nitrogenase activity in Δcse cultures was barely detectable at around 6-fold less than the WT (Fig. 5). After 48 h in N-fixing conditions, nitrogenase activity increased from 0.36 to 1.11 μmol h− 1 mg proteins− 1 in WT and 0.06 to 0.35 μmol h− 1 mg proteins− 1 in Δcse, with Δcse demonstrating around 3-fold lower activity than that of WT in N-fixing conditions.

Fig. 5
figure 5

Acetylene reduction assay. Nitrogenase activities of wild-type (WT) and Δcse cultures grown for 2 days in BG11 or BG110 in 3% CO2. Significant differences between WT and Δcse samples (n = 3) are indicated with asterisks (t-test P <  0.05)

Full size image

Particles adhere to the outer surface of Δcse cells

SEM micrographs of WT and Δcse cultures highlighted the filament fragmentation phenotype of the mutant (Fig. 6a and b), and also revealed the occurrence of disorganised clumps of Δcse cells that appeared to be surrounded by an extracellular membrane or matrix (Fig. 6e). Another striking feature of Δcse apparent in the SEM images was the presence of particles of an unidentified substance on the exterior surface of the cells, which was in contrast to the smooth cell surface of WT vegetative cells (Fig. 6a-c). SEM and TEM micrographs revealed the frequent occurrence in Δcse cultures of two vegetative cells, which appeared to be newly divided, encapsulated in a mutual outer layer that resembled a heterocyst envelope (see arrows in Fig. 6f and h). Staining of these dividing cells with Alcian Blue confirmed the presence of heterocyst-specific polysaccharides (see Additional Fig. 1).

Fig. 6
figure 6

Phenotypic features of the Δcse mutant. Electron micrographs of wild-type (WT; left) and Δcse (right) cells grown in BG11 medium in 3% CO2. a-f Scanning electron microscopy (SEM) images. g-h Transmission electron microscopy (TEM) images. Normal heterocysts (carets) and partially-differentiated proheterocysts (arrows) are indicated in (d-h)

Full size image

Δcse has higher concentrations of total sugars

Total sugars were measured in WT and Δcse cell pellets, and in the supernatant fractions obtained after centrifugation of cultures grown in BG11. These analyses showed that Δcse cells comprised 32% sugars in comparison to 23% in WT cells, relative to their dry biomass (Fig. 7). The total sugars content of the growth media of Δcse cultures was approximately twice that isolated from WT growth media (14 and 8%, respectively, relative to the dry biomass).

Fig. 7
figure 7

Distribution of accumulated and secreted carbohydrate hexoses of cultures. Total sugars measurements in wild-type (WT) and Δcse cell pellets (cellular) and their respective supernatants (growth media; extracellular) relative to the dry biomass. Error bars indicate the standard deviation of three biological replicates (n = 3) grown in BG11 in 3% CO2. Significant differences between WT and Δcse samples are indicated with asterisks (t-test P <  0.05)

Full size image

Heterocyst differentiation is downregulated at the transcript level

Comparison of the transcriptomes of the Δcse mutant strain and the WT grown in BG11 revealed that a majority of differentially expressed genes were related to the formation and function of heterocyst cells. Among the strongest upregulated genes in Δcse were patU5/3, asr1734, patS, hetZ, hetP, patA, which are involved in the regulation of heterocyst differentiation (see Table 1). Other heterocyst regulators ntcA and hetC were only slightly downregulated (FC = − 1.3). Expression of factors responsible for the biosynthesis of the heterocyst envelope and cytoplasmic differentiation was predominantly downregulated. For instance, gene cluster alr2822 - alr2841, which is called the “Heterocyst Envelope Polysaccharide (HEP) island” [38], was downregulated 2 to 3.5-fold, compared to the WT, while gene cluster all5341 - all5359, encoding proteins for glycolipid biosynthesis, was even more strongly downregulated (see Table 1). Genes encoding heterocyst-specific proteins such as flavodiiron proteins (flv1B/3B), ferredoxin (fdxH), both cytochrome c oxidase operons (cox2/3), the devBCA transporter, patB and the uptake hydrogenase cluster were also strongly downregulated. The expression of nif genes was mildly to highly repressed (1.1 to 6.5-fold), with nifH, nifK, nifB, nifS and nifW being the most downregulated. Other clusters associated with N fixation such as all1424 – all1427 and all1431 – all1440 were also strongly downregulated.

Table 1 Transcription changes in the Δcse mutant
Full size table

Among the most strongly upregulated genes in Δcse were enzymes of the chlorophyll and pyrimidine biosynthesis pathways, including coproporphyrinogen III oxidase (FC = 17.9), dihydroorotate dehydrogenase (FC = 6.5) and the magnesium-protoporphyrin IX monomethyl ester [oxidative] cyclase 1 (FC = 4.2). The bidirectional hydrogenase complex subunit hoxH was also upregulated 5.9-fold.

Knockout of the cse gene affects expression of N-regulated genes

To investigate the impact of CSE on the expression of genes involved in heterocyst differentiation, we analysed the expression of ntcA and hetR genes in WT and Δcse during the course of N step-down. In WT, ntcA and hetR expression peaked after 8 h in N-deficient media, and afterwards returned to pretreatment levels. In Δcse, however, ntcA expression peaked at 24 h, while hetR expression peaked 1 h after the N shift (Fig. 8).

Fig. 8
figure 8

Expression of heterocyst differentiation regulator genes during nitrogen step-down. Expression of ntcA (a) and hetR (b) in wild-type (WT) and Δcse after nitrogen step-down. Cultures were grown in BG11 medium in 3% CO2 and then refreshed in BG110 medium lacking any source of combined nitrogen for a nitrogen step-down. RNA samples were taken at the timepoints indicated. Gene expression values were normalised to the reference gene rpoA and the timepoint 0 h values. Error bars indicate the standard deviation of mean values from three biological replicates (n = 3). Significant differences between WT and Δcse gene expression are indicated with asterisks (t-test P <  0.05)

Full size image

Discussion

CSE is a small Ca2+-binding EF-hand protein, which undergoes a conformational change upon binding of Ca2+ [7]. This is a typical feature of Ca2+ sensor proteins for the interaction with protein partners, thus translating a Ca2+ signal into a physiological response [39]. In Anabaena, CSE is the only known Ca2+ sensor [7], putatively containing two Ca2+-binding EF-hand domains of low and high affinity, similar to calcineurin B, the regulatory subunit of the mammalian serine/threonine protein phosphatase calcineurin A [7, 40,41,42,43,44]. It was shown that cse expression responds to changes in the intracellular C/N balance, which is an indicator of the metabolic status of the cell. Expression of cse was strongly downregulated 1 h after the shift to N-depleted media, while conversely a low C/N ratio caused a dramatic increase in cse expression. Downregulated photosynthetic activity under increased CSE abundance indicated that CSE and Ca2+ signalling may link photosynthetic activity with the metabolic status of the cell [7]. In the current study, CSE knockout mutants displayed a striking phenotype of predominantly very short filaments comprising up to five cells, which was restored to the WT phenotype when the Δcse mutant was complemented with the asr1131 gene (Fig. 2g). Several independently conjugated Δcse clones were created, and in every case the short filament phenotype came to dominate the culture, although some variation was observed in development of the phenotype over time, ranging from days to weeks. Despite full segregation of Δcse cultures, a small proportion of WT-like long filaments persisted. These heterocyst-containing filaments supported the growth of Δcse cultures in N-deficient conditions, albeit more slowly and with lower nitrogenase activity than WT cultures and substantial catabolism of N-containing pigments (Figs. 4 and 5). However, the WT-like filaments of the Δcse mutant were also susceptible to fragmentation in both N-replete and N-deficient conditions (Fig. 3). These results implicate CSE in a Ca2+ signalling pathway that influences filament integrity in Anabaena. Filament fragmentation has been previously reported to result from abiotic disturbances, such as osmotic stress through UV radiation, or nutrient deficiencies [45], while high [Ca2+] introduced into growth media also induced filament fragmentation [4]. The role of Ca2+ signalling in abiotic stress response in Anabaena has been demonstrated [1,2,3], suggesting that fragmentation of Δcse mutant filaments may be due to disruption of a stress-responsive Ca2+ signalling pathway that leads to an inability to prevent disconnection of filaments. In this case, the WT-like filaments present in Δcse cultures may be unstressed cells in which the Ca2+ signalling pathway is not activated.

Genes involved in later stages of heterocyst development and nitrogen fixation activity were among the most strongly downregulated in Δcse cultures growing in BG11 (Table 1), correlating with a strong decrease in the frequency of heterocysts in these cultures, compared to WT cultures. At the same time, abnormally high expression of hetP, patA, patS, patU3/5 and other differentiation inhibitors and heterocyst patterning genes in Δcse (Table 1) suggested that more Δcse cells, compared to WT, commit to enter the “proheterocyst” stage during early stages of heterocyst development, but then fail to develop into mature heterocysts. This indicates that heterocyst maturation was suppressed at the transcriptional level. This may be linked to abnormal activity of the HetR master regulator of heterocyst development due to increased abundance of the PatS peptide in Δcse, which is likely to interfere with HetR function [46,47,48]. Evidence of dysfunctional cell differentiation in the Δcse mutant was also apparent in the frequent occurrence of dividing cells enveloped in a heterocyst-like outer layer, revealed by electron and bright-field microscopy (Fig. 6f, h and Additional Fig. 1). These phenomena have been previously identified in short filament mutants of Anabaena and are described as partially-differentiated proheterocysts that continued dividing [25, 28, 32, 33], whereas cell division is normally arrested in mature heterocysts [49]. Failure to complete heterocyst commitment and maturation may have been linked to abnormal abundance of differentiation inhibitors described above [50, 51], or interrupted Ca2+ signalling during heterocyst development (discussed below). Both NtcA and the signalling molecule 2-oxoglutarate [52] were suggested to control Ca2+ signals required for heterocyst differentiation in multicellular cyanobacteria and acclimation to N starvation in unicellular cyanobacteria, defining Ca2+ as an early signal of N deprivation [53, 54].

Misdeveloped heterocysts in Δcse may have become weak points in the filaments, causing disconnection and disrupting filament integrity. Disrupted dispersion of the inhibitors may have resulted in a higher number of differentiating cells that accumulate inhibitors and thus fail heterocyst maturation, leading to the observed short filament phenotype [51, 55]. Short filament phenotypes similar to that observed in the Δcse mutant have been reported in numerous heterocyst formation-impaired mutants [23,24,25,26,27,28,29,30,31]. The majority of previously reported filament fragmentation mutants have been defective in the Fra proteins [34], of which the septum-localised SepJ/FraG [32, 33] is the most important for filament integrity in N-depleted environments. No significant differences in expression of Fra-encoding genes were observed in our RNAseq data in Δcse in comparison to the WT; however, we cannot rule out possible functional interactions between CSE and Fra proteins that may have induced fragmentation in the Δcse mutant.

Ca2+ signalling plays a vital role in heterocyst differentiation [13], and the evidence presented here suggests that the Ca2+-binding CSE protein may be involved in this signalling pathway. Earlier studies have reported a strong and sustained increase in [Ca2+]i in Anabaena five to six hours after N step-down, which is attributed to downregulation of the Ca2+-binding protein CcbP by the master regulators HetR and NtcA [16, 56,57,58,59,60]. However, CSE appears to operate at an earlier point in the heterocyst differentiation process. Indeed, cse (asr1131) in Anabaena WT was included in a group of genes shown to be downregulated by combined N deprivation [61], and cse transcription decreased sharply within the first hour of N step-down [7]. This timing appears to correspond with a spike in Ca2+ uptake and a small increase in [Ca2+]i that have been observed soon after N deprivation in Anabaena [13, 16]. This early Ca2+ signal of N step-down, which could also arise through uptake of extracellular Ca2+ through channels in the plasma membrane [1,2,3, 62], may activate downstream processes that lead to heterocyst differentiation. In the current work, expression of hetR during N step-down was misregulated in Δcse, peaking after 1 h compared to 8 h in the WT (Fig. 8b). In addition, HetR- and NtcA-binding sites have been identified in the promoter and terminator region of the cse coding sequence [7, 63, 64]. Taken together, these data suggest that CSE may be involved in a Ca2+ signalling pathway during the initial stages of heterocyst differentiation, with a decrease in CSE abundance being important for the distinct early transient rise in [Ca2+]i that influences the transcriptional regulation activity of HetR [14, 16]. Complete deletion of the cse gene would interrupt this early Ca2+ signal, leading to unregulated heterocyst development and filament fragmentation.

Conclusions

CSE appears to be important for Ca2+ signalling during changes in cellular C/N balance. In our previous study, downregulation of photosynthesis was linked to CSE upregulation upon C decrease [7], while the current work has implicated CSE in filament integrity and proper cell differentiation in response to low N. The exact mechanism by which the newly discovered CSE protein operates is still not clear, but likely involves a conformational change upon Ca2+ binding that induces interaction with an unidentified protein partner. CSE may also regulate the abundance of free Ca2+ in the cell, thereby influencing Ca2+-sensitive processes like stress response, gene expression and heterocyst development.

Methods

Generation of an asr1131 deletion mutant and complementation

Asr1131 deletion mutant (Δcse) strains of Anabaena were generated by replacing the 234 bp asr1131 coding region [65] as well as putative regulatory regions of non-coding DNA 93 and 266 bp up- and downstream, respectively, with a neomycin/kanamycin-resistance cassette (Fig. 1a). 1.5 kb sequences up- and downstream of asr1131 were amplified from Anabaena wild-type (WT) genomic DNA by PCR for homologous recombination in Anabaena, using the oligonucleotides cse_upst-PstI-S, cse_upst-XbaI-AS, cse_dwst-BamHI-S, and cse_dwst-SalI-AS (Additional Table 1). The upstream PCR product was digested with PstI and XbaI, and the downstream PCR product was digested with BamHI and SalI. Both fragments were ligated to a vector containing a neomycin/kanamycin antibiotic cassette and the sacB gene for selection of double recombinants obtained from digestion of the plasmids pRL448 with XbaI and BamHI, and pRL271 with PstI and SalI, respectively. The resulting plasmid consisted of a neomycin/kanamycin resistance cassette flanked by asr1131 upstream and downstream sequences, which was used for triparental conjugation of Anabaena WT as described earlier in [66]. Double recombinants were selected on growth media containing 5% sucrose and 200 μg μl− 1 neomycin. Full segregation of two independent Δcse clones was shown by PCR amplification using the oligonucleotides cse_upst-PstI-S and cse_dwst-SalI-AS. The Δcse mutant strain was complemented by conjugation with RSF1010-based replicative vector pBG2089 expressing asr1131 under the control of its native promoter and terminator sequences (Additional Table 1). Transformants were selected on media supplemented with 200 μg μl− 1 neomycin and 5 μg μl− 1 erythromycin.

Growth conditions and treatments of Anabaena and E.coli cultures

Anabaena WT, Δcse and the complemented Δcse strain were grown in BG11 [11] buffered with 10 mM TES-KOH (pH 8.0) under constant illumination of 50 μmol photons m− 2 s− 1 at 30 °C. All cultures were grown in air enriched with 3% CO2 with gentle agitation (120 rpm). For the selection of transformants, 40 μg μl− 1 neomycin was added to liquid cultures of Δcse, and 40 μg μl− 1 neomycin and 5 μg μl− 1 erythromycin were added to the complemented Δcse strain. E.coli strains used for cloning were grown in Luria-Bertani (LB) medium supplemented with the antibiotics indicated in the Additional Table 1.

Unless otherwise stated, fresh cultures of Anabaena were started at a chlorophyll concentration of 0.1 μg ml− 1 in BG11 or BG110 (BG11 lacking NaNO3 in the macronutrients, and with CoCl2 ∙ 6 H2O substituting Co(NO3)2 ∙ 6 H2O in the trace metals). Growth of cultures was monitored over 5 days by measuring total proteins according to a modified Lowry protocol described in [6] and chlorophyll a absorption (OD665) in 90% methanol. The optical densities and total absorption spectra of cultures were measured with a Thermo Scientific Genesys 10S UV-Vis Spectrophotometer. The dry biomass of cells was determined according to the method described in [6].

Microscopy techniques

Bright-field images taken with a Zeiss Axiovert 200 M inverted microscope on × 200 magnification were used for filament length counts. At least 200 filaments were counted from each culture grown for 4 days in BG11 starting from OD750 = 0.1 under standard growth conditions. Heterocysts were stained with Alcian Blue stain according to the method described in [7].

For Scanning electron microscopy (SEM) and Transmission electron microscopy (TEM), 1 ml of culture grown in BG11 was collected during the exponential growth phase (OD750 = 1.0), centrifuged at 1 x g, and the cell pellet fixed in S-collidin buffer and 25% glutaraldehyde (4:1). The samples for SEM and TEM were prepared at the Laboratory of Electron Microscopy (University of Turku, Finland). TEM was performed at the Laboratory of Electron Microscopy (University of Turku, Finland) using a JEM-1400 Plus Transmission Electron Microscope. SEM was carried out at the Institute of Dentistry (University of Turku, Finland). For determination of the percentage of dividing cells, a pair of dividing cells was counted as one cell.

Determination of total sugars content

Samples of 1 ml of a culture grown for 3 days in BG11 were centrifuged at 6000 x g, and the supernatant and cell pellet were treated separately. After diluting the cell pellets 1:1 with milliQ water in glass tubes, the total amount of sugars in cell pellets and supernatants was determined with a colorimetric method according to [67]. Raw data were normalised to the dry biomass of the culture.

Nitrogenase activity measurements

Nitrogenase activity of liquid cultures grown in BG11 or BG110 for 2 days was detected using the acetylene reduction assay according to [68], described in [7].

N step-down experiment and RT-qPCR

WT and Δcse strains were grown for 3 days in BG11. During the exponential growth phase (OD750 = 1.0), 2 ml samples were frozen for RNA isolation. The cultures were adjusted to OD750 = 0.6 in fresh media and shifted to N limited conditions (BG110 medium) after washing once with BG110. Samples of 2 ml were collected and frozen for RNA isolation immediately before (timepoint 0 h) and 1, 4, 8, 12, 24 and 48 h after the shift. RNA isolation was performed as described in [6].

Five hundred nanograms of RNA was utilised for cDNA biosynthesis using the SuperScript III First-Strand Synthesis System (Invitrogen). Transcripts were amplified from 5-fold diluted samples from three biological replicates with the iQ SYBR Green Supermix and the iQ5 Multicolor Real Time PCR Detection System. The reference gene rpoA was amplified with the oligonucleotides rpoA_qPCR-S and rpoA_qPCR-AS. Oligonucleotides used for the analysis of gene expression during N step-down experiments are listed in the Additional Table 1. Normalised expression values were calculated using the Pfaffl method [69].

RNA isolation and transcriptome sequencing and analysis

WT and Δcse strains were grown for 3 days in regular growth conditions in BG11, and 2 ml of culture were taken from three biological replicates (n = 3) of each strain during the exponential growth phase (OD750 = 1.0). Total RNA was isolated as described in [6], re-extracted in lithium chloride overnight and submitted to the Beijing Genomics Institute (Shenzhen, China) for preparation of single-ended RNAseq libraries and sequencing using Illumina-HiSeq2500. RNAseq reads were aligned to the reference genome of Nostoc sp. PCC 7120, downloaded from Ensembl (EBI), using Strand NGS 2.7 software (Agilent, USA). Quantification of the aligned reads was performed using the DESeq R package. Significantly differentially expressed genes were identified by a 2-way ANOVA test using the Benjamini-Hochberg method for false discovery rate (FDR) correction of P values.

Bioinformatics methods

Gene descriptions were obtained from CyanoBase (Kazusa Genome Resources; genome.microbedb.jp/cyanobase), KEGG (www.genome.jp/kegg/), UniProt (www.uniprot.org) and the National Center for Biotechnology Information (NCBI; www.ncbi.nlm.nih.gov).