6S RNA supports recovery from nitrogen depletion in Synechocystis sp. PCC 6803

Background The 6S RNA is a global transcriptional riboregulator, which is exceptionally widespread among most bacterial phyla. While its role is well-characterized in some heterotrophic bacteria, we subjected a cyanobacterial homolog to functional analysis, thereby extending the scope of 6S RNA action to the special challenges of photoautotrophic lifestyles. Results Physiological characterization of a 6S RNA deletion strain (ΔssaA) demonstrates a delay in the recovery from nitrogen starvation. Significantly decelerated phycobilisome reassembly and glycogen degradation are accompanied with reduced photosynthetic activity compared to the wild type. Transcriptome profiling further revealed that predominantly genes encoding photosystem components, ATP synthase, phycobilisomes and ribosomal proteins were negatively affected in ΔssaA. In vivo pull-down studies of the RNA polymerase complex indicated a promoting effect of 6S RNA on the recruitment of the cyanobacterial housekeeping σ factor SigA, concurrently supporting dissociation of group 2 σ factors during recovery from nitrogen starvation. Conclusions This study reveals 6S RNA as an integral part of the cellular response of Synechocystis sp. PCC 6803 to changing nitrogen availability. According to these results, 6S RNA supports a rapid acclimation to changing nitrogen supply by regulating the switch from group 2 σ factors SigB, SigC and SigE to SigA- dependent transcription.


Background
Cyanobacteria are photoautotrophic prokaryotes essentially relying on the availability of sunlight and CO2 as their major energy and carbon source, respectively. Due to their autotrophic lifestyle and versatile metabolism, cyanobacteria are suited as economical cellular chassis for diverse biotechnological applications like energy feedstock accumulation [1], third generation biofuel production [2][3][4][5] and commodity product biosynthesis [6][7][8]. However, the long-term mass cultivation for these applications at high cell densities also causes stress conditions that arise from redirecting metabolism, accumulation of potentially toxic compounds and the concomitant tendency for nutrient limitation (reviewed by Dexter et al. [2]). Hence, the requirements for molecular tools and well-defined targets for manipulating the cellular program towards optimized nutrient utilization are of growing interest. Within the diversity of their natural terrestrial and aquatic habitats, cyanobacteria have certainly developed extensive regulatory systems to acclimate to nutrient and light limiting conditions, which are also involving small regulatory RNA molecules (sRNAs, e.g. [9][10][11][12][13][14]).
One of the best-characterized prokaryotic sRNA regulators is the widely conserved 6S RNA that responds to changes of the nutritional status [15][16][17][18]. When cells of Escherichia coli (E. coli) enter the stationary growth phase, this highly structured RNA specifically regulates transcription by promoter mimicry, as the RNA polymerase (RNAP) holoenzyme carrying the housekeeping sigma factor σ 70 binds to 6S RNA instead of the promoter regions of the household genes [19][20][21][22]. Upon nutrient-induced outgrowth from the stationary phase, which includes enhancement of NTP levels, 6S RNA acts as a template for the de novo synthesis of a ~20 nt product RNA (pRNA) [23,24]. pRNA synthesis triggers the release of 6S RNA from RNAP reverting 6S RNA-dependent inhibition [23,25].
Homologues of 6S RNA have been identified in several freshwater cyanobacteria, including the model organism Synechocystis sp. PCC 6803 (Synechocystis 6803) [15,26,27]. Whereas in vitro studies indicate a conservation of basic 6S RNA mechanisms, little is known about the functional relevance of the cyanobacterial 6S RNA in vivo [28].
Since phototrophic growth of cyanobacteria does essentially rely on atmospheric or enriched CO2 supplementation and -importantly -light, plain dilution experiments would rather evoke responses to changing light availability (and its relation to CO2 availability). Accordingly, a specified experimental setup was required for functional characterization of cyanobacterial 6S RNA during outgrowth from nutritional deprivation. Regarding the status of an inorganic nitrogen source -i.e. nitrate or ammonia -non-diazotrophic cyanobacteria like Synechocystis 6803 exhibit a wellcontrolled acclimation behavior that is also accompanied by a stationary growth plateau [35].
Acclimation of cyanobacteria to changing nitrogen availability is a tightly controlled process that has been well described in the literature. One of the most pronounced physiological responses is the remodeling of the photosynthetic machinery, leading to distinctly and visibly reduced amounts of the phycobilisome antenna complex and -delayed in time -chlorophyll a breakdown [35][36][37][38]. These biochemical mechanisms are influenced by elaborate underlying events of gene expression control.
This network essentially involves the collaborative action of the global transcriptional regulator NtcA and the proteins PII and PipX [39].
Moreover, several group 2 σ factors appear to play a role within the nitrogen regulatory network.
Besides the nitrogen stress response σ factor SigE, these are SigB, involved in the NtcA-dependent nitrogen-related gene expression during exponential growth and SigC, in the response of stationary phase cultures to nitrogen starvation [40][41][42].
Several reports described the transcriptomic characteristics of nitrogen-starving cyanobacteria [43,44] as well as the response upon nitrogen repletion in a time-resolved manner [38,45]. The recovery process after long-term starvation is initiated by the upshift of respiratory gene expression and switching on the translational machinery as well as nitrogen assimilation.
Further, until full physiological restoration, the cells successively reactivate photosynthetic complexes, carboxysomes and the cell division machinery [45]. Cells without any group 2 σ factors are not able to recover from nitrogen deficiency [42].
Here we describe the involvement of the sRNA regulator 6S RNA in the dynamics of nitrogen acclimation in Synechocystis 6803, elaborating the example of NO3 --mediated recovery from longterm nitrogen starvation.

Inactivation of 6S RNA affects photosynthetic parameters and storage of carbohydrates upon nitrogen starvation and recovery
To analyze the functional role of 6S RNA in cyanobacteria, we deleted the ssaA gene encoding 6S RNA from the genome of Synechocystis 6803. The ssaA gene was replaced with a kanamycin resistance cassette (Fig. 1a) and complete segregation of the resulting ΔssaA mutant was confirmed by colony PCR (Figure 1b) as well as by Northern Blot analysis (Fig. 1c).
Under standard growth conditions, the ∆ssaA strain did not show any clear phenotype.
Referring to 6S RNA function in E. coli and B. subtilis, where it regulates RNAP activity during the stationary growth phase and outgrowth [17,18,46], the following mutant characterization experiments were conducted under nitrogen depletion (which leads to a stationary state of cell culture) and subsequent recovery by repletion with 17.6 mM NaNO3 (corresponding to outgrowth).
The acclimation to nitrogen depletion proceeded similarly in the ΔssaA strain as in wild type (WT), demonstrated by similar whole-cell absorption spectra of ΔssaA and WT cells after seven days of nitrogen deficiency (Fig 2a, 0h +N samples). Furthermore, light-saturated photosynthetic activities ( Fig. S1a) and 77K fluorescence spectra (Fig. S1b) were similar in both strains in the middle of the nitrogen deficiency treatment.
Visible differences between the strains were detected during the recovery phase. After 22h of recovery, WT showed normal green color of a viable Synechocystis 6803 culture while the ΔssaA mutant still remained yellow-orange, which is a typical characteristic of nitrogen depleted cultures (Fig. 2a). The ΔssaA complementation strain (∆ssaA-c) recovered from the nitrogen deficiency similarly as WT (data not shown).
The color of nitrogen depleted cells is due to prominent loss of blue phycobilins (absorption maxima at 620 -635 nm), clear loss of green chlorophyll a (Chl a, absorption maxima at 680 nm), but only minor reduction of yellow-orange carotenoid pigments ( Fig. 2c; compare 0h +N and the fully recovered 144h +N samples). Re-synthesis of pigments takes hours after nitrogen supply. In both strains phycobilin (Fig. 2c) and Chl a ( Fig. 2b and 2c) contents remained low after 4h of recovery.
After 22h both phycobilins and Chl a showed significant retrieval in WT while in ∆ssaA, the recovery process was still in an early stage. A significant delay in phycobilisome reconstruction in ∆ssaA was also seen as reduced contents of the main phycobilisome proteins, phycocyanin and allophycocyanin, after 22h of nitrogen repletion ( Fig. S2a and c). However, ΔssaA cells completely recovered over extended periods of 48h or longer ( Fig. 2b and 2c).
77K fluorescence spectra, measured with orange light phycobilisome excitation 14h after nitrogen addition, showed a lower phycobilisome peak in ΔssaA cells than in WT confirming the slow recovery of phycobilisome antenna in the mutant (Fig. 3a). In addition, the ratio of the photosystem II (PSII) CP47 peak to CP43 peak was lower in the ΔssaA mutant than in WT, indicating reduced efficiency of electron transfer within PSII in ΔssaA (Fig. 3a). In accordance with that the lightsaturated PSII activity of the ΔssaA strain was 20% lower than that of WT after 22h of recovery in nitrogen replete conditions (Fig. 3b), and also the light saturated photosynthesis rate of the mutant strain was lower than that of WT (Fig. 3c). According to these results, both light-harvesting capacity and photosynthetic activity recovered more slowly in the mutant than in WT.
Since nitrogen starvation limits consumption of carbon skeletons, starved Synechocystis 6803 cells store carbon polymers like glycogen. The stored glycogen content remained high in both strains during the first four hours after nitrogen addition (Fig. 3d). Thereafter, glycogen decreased in both strains but the degradation proceeded more rapidly in WT than in ΔssaA (Fig. 3d). Our results indicate that although 6S RNA is not completely essential for the cells to recover from nitrogen starvation, the presence of 6S RNA accelerates the recovery process for hours.
Further investigations was carried out to analyze the 6S RNA transcript level under nitrogen starvation and recovery. Northern Blot analysis of 6S RNA revealed that the accumulation of 6S RNA increased in WT in the course of long term (189h -N) nitrogen depletion by ~40 -55% and returned to the original level, when normalized to 5S rRNA levels (Fig. 4a). However, since ribosomal RNA is rather expected to decrease during chlorosis [47] and consequently to increase again during recovery, 6S RNA accumulation is rather constant or slightly upregulated (Fig. 4a).

6S RNA accelerates expression of household genes upon recovery from nitrogen starvation
Comparative transcriptome analysis was performed to infer a functional signature from the mutantspecific transcript profiles (raw data is available in supplementary files 2 and 3). For this purpose, microarray analysis was used to probe total RNA from WT and ∆ssaA cultures that were starved for nitrogen for seven days and recovered over a period of 22 hours. Sampling time points were t1 = 0h (7d -N; reference), t2=1h +N, t3=4h +N and t4=22h +N. At t1=0h a total set of 65 features showed differential accumulation (log2 fold change (FC) of > 1 (adjusted p-value of ≤ 0.05) in ΔssaA, the majority of which were downregulated. Most of the latter comprised non-coding transcripts like asRNAs, but also eight mRNAs showed lower levels. Among them were several mRNAs encoding subunits of ATP synthase, the mRNA for cell division protein SepF as well as ferredoxin encoding petF and the high-light inducible hliA. The positively regulated genes included the heat stress responsive histidine kinase hik34 and the chaperone encoding genes dnaJ and groEL-2 (Fig. S3).
Temporally altered transcript levels in ΔssaA from t1= 0h to all three analyzed time points of recovery were plotted as a function of the corresponding temporal changes in WT, distinctively comparing the recovery in both strains (Figs. S4-6). Comparison of transcript profiles indicated that only few genes were activated or repressed in the early phase of recovery, whereas 22h after addition of nitrogen, numerous genes responded, most of which were upregulated (Fig. S6).
Absolute expression values of selected groups of functionally related genes are further depicted in time courses in Fig. 5a-f. In WT cells transcriptional upshift of ATP synthase genes was seen already 1h after nitrogen addition and continued for whole recovery phase (Fig. 5a). In ΔssaA, upregulation of atp transcripts was delayed, keeping atp transcripts at low levels for the first 4h.
However, these differences between both strains vanished after 22h of recovery (Fig. 5a). Likewise, the transcriptional upshift of ribosomal subunits was delayed in ΔssaA (Fig. 5b).
During the first 4h after nitrogen repletion many genes encoding proteins for the major t After 22h of recovery these effects wer not that clear anymore. For genomic context of apc and cpc operons, see Figure S7. Altogether, the transcriptomics data is in good agreement with the delayed physiological response of the mutant as reflected by the data shown in Figures 2 and 3.
Notably, levels of important 'marker' transcripts of nitrogen recovery encoding components of the nitrate uptake (nrtABCD) and early assimilation machinery (narB, nirA, moaACE) increased rapidly in both strains (Fig. 5d). Over the complete period, abundances of e.g. nitrate reductase encoding nirA kept stable in the mutant, whereas a moderate decline after the first hour was observed for all transcripts in the WT and likewise -for the majority of this group -in ΔssaA (Fig. 5d).
For possible transcriptional regulators, we focused on genes encoding the regulatory σ factors. The sigA transcript (encoding primary σ factor SigA) was upregulated in ΔssaA compared to WT after 4h of recovery, while transcripts for group 2 σ factors SigB and SigC were slightly up in all samples. Levels of sigG, encoding an ECF-type σ factor, were slightly down ( Fig. 6a-g). Note that the mRNA accumulation of σ factor encoding genes in cannot considered a measure of σ factor recruitment by RNAP, but rather an effect of the latter.
In addition to protein-coding genes, a few small non-coding RNAs (sRNAs) showed altered levels in the mutant, including the previously characterized PmgR [48], NsiR4 [49] and PsrR1 [11], as well as several hypothetical riboregulators that were identified by differential RNA sequencing (dRNA-seq) [50]. Among these transcripts, SyR11 exhibited the most diverging accumulation kinetics ( Fig. 6h; see also Northern Blot validation in Figure S8). Additionally, Fig. 6i-k illustrate the mRNA abundances of three exemplarily selected transcripts (psaM, psbK, psbM) of PSI and PSII, revealing the decelerated upregulation of components of the photosynthetic apparatus in ΔssaA in detail.

6S RNA determines the allocation of SigA, SigB, SigC and SigE in the RNA polymerase holoenzyme
We next studied if the absence of 6S RNA influences the content of the primary or group 2 σ factors and/or recruitment of different σ factors by the RNAP core to form the holoenzyme during nitrogen starvation and recovery phase. To that end, the γ-subunit of RNAP core was replaced with His-tagged γ-subunit in WT and ΔssaA as described recently for a non-motile Synechocystis 6803 strain [51]. The WT RNAP-His and ΔssaA RNAP-His strains were grown under standard conditions or subjected to nitrogen depletion for 6 days; recovery samples were taken 1h, 4h or 22h after nitrogen addition.
The His-tagged RNAP complexes were collected from one-half of each sample and the soluble proteins were isolated from the other half.
To measure possible changes in SigA content and/or in recruitment efficiencies by the RNAP core, isolated soluble proteins and collected RNAP complexes were separate on SDS-PAGE, respectively. Subsequently, the SigA content was detected using a SigA specific antibody [51]. Then membranes were re-probed with the antibody against the α (or β) subunit of the RNAP core and the SigA content was normalized to the RNAP core content. Finally, after each treatment, the amount of SigA was compared to the amount of SigA in cultures from standard (nitrogen replete) conditions. SigA almost vanished from the RNAP holoenzyme during nitrogen depletion in both strains (Fig. 7a), but SigA protein content decreased by only 65% (Fig. 7b) indicating that the recruitment of SigA by RNAP core reduced more drastically than SigA content. After nitrogen addition, RNAP core successively recruited more SigA (Fig. 7a). The recovery of the SigA recruitment occurred more slowly in ΔssaA than in WT (Fig. 7a), although the SigA content increased more rapidly during recovery in ΔssaA than in WT (Fig. 7b). Poor recruitment of SigA in ΔssaA is an obvious reason for slow activation of household genes in this strain.
When group 2 σ factors were analyzed, using the same method as described for the primary σ factor, the results showed that in nitrogen starved WT cells (0h +N) both recruited and total SigB remained at the same level as under standard conditions ( Fig. 7c and 7d). Although the total amount of SigB decreased moderately during the recovery period, this σ factor was recruited in particular during the early recovery phase ( Fig. 7c and 7d). In ΔssaA, SigB was more efficiently recruited by RNAP core both under standard conditions and during early recovery than in WT ( Fig. 7c and 7d).
Thus, 6S RNA seems to restrict recruitment of SigB by the RNAP core. Like SigB, SigC was highly recruited during recovery phase, however SigC was already recruited efficiently in nitrogen depleted cells ( Fig. 7e and 7f). Very high levels of SigC recruitment were observed in the mutant strain at early time points of recovery. The amount of SigD was very low or below detection limit in all samples and quantification of SigD results was not possible. Unlike the other σ factors, the SigE protein content was clearly lower in nitrogen depleted samples, which was also observed in samples from the early recovery phase. Increased SigE content resulted in enhanced recruitment of SigE during nitrogenrecovery, especially in early phase ( Fig. 7g and 7h).

6S RNA promotes the recovery process from N starvation
The acclimation processes of cyanobacteria to changing nitrogen availability are underlain by a complex regulatory network, involving signal-driven transcriptional regulons and differential σ factor activity, refined with a layer of sRNA-mediated post-transcriptional regulation [49,[52][53][54].
Here we demonstrate that the cyanobacterial 6S RNA homologue plays a role within the rapid transcriptional response during recovery from nitrogen depletion. The ΔssaA mutant of Synechocystis 6803 showed a more sustained bleaching phenotype linked with a generally decelerated transcriptional and physiological response during recovery. Thus, within this gradual acclimation process 6S RNA containing cells have a substantial selective advantage over cells missing 6S RNA. On the other hand, overproduction of 6S RNA in Synechocystis did not accelerate recovery from nitrogen starvation, suggesting that this riboregulator is not a limiting factor in the native acclimation system.
The behavior of 6S RNA during nitrogen-deficiency-induced bleaching and subsequent recovery in Synechocystis 6803 differs from that reported for stationary phase in heterotrophic bacteria. In Synechocystis 6803, 6S RNA increases only slightly during nitrogen starvation (Fig. 4a) and 6S RNA is quite abundant under standard conditions [50,55]. For the cyanobacterium Synechococcus sp. PCC 6301 it has been reported previously, that 6S RNA levels decrease during stationary phase [27]. In E. coli, 6S RNA molecules enrich 10-fold from early exponential growth phase to late stationary phase [19] and binding of RNAP-σ 70  It is possible that 6S RNA exhibits some functional diversity within different bacterial phyla.
In accordance with the physiological data, transcriptome profiling demonstrated that particularly genes encoding ATP synthase, PSI and PSII and phycobilisomes as well as the translational machinery were overall negatively affected by the absence of 6S RNA during recovery from nitrogen starvation. The resuscitation process from severe long-term nitrogen starvation has recently been described in detail for Synechocystis 6803 by Klotz and co-workers [45]. The recovery process from nitrogen starvation can be divided into two phases. Phase 1 is characterized by regeneration of basic cellular functions, including respiration, translation and nitrogen assimilation, which could be clearly confirmed by our transcriptome study. Here, certainly expression of ATP synthase and ribosomal genes showed a delay of general upshift from 1h to 4h after nitrogen addition in the ΔssaA mutant.
The kinetics of photosynthesis-related mRNA accumulation further emphasizes the generally delayed recovery: the early phase 2 response, which is partially initiated after 4h in WT, rather arises later in the mutant, largely approaching WT levels after 22h. Moreover, due to delayed transcriptional activation of phase 1 and phase 2 genes, recovery of physiological processes like phycobilisome reassembly, photosynthetic activity and glycogen degradation were delayed in ΔssaA.
Our results show that 6S RNA regulates recruitment of σ factors especially during early recovery from nitrogen depletion but does not affect expression of NtcA controlled genes like amt1, glnA, glnB or sigE, indicating an additional regulatory circuit that acts independently and putatively supplementary to the well-characterized PipX-PII-NtcA network [46,47]. Although none of the group 2 σ factors alone is essential for acclimation to nitrogen deficiency and subsequent recovery, cells without functional group 2 σ factors were more sensitive to nitrogen depletion [30].
During recovery from nitrogen depletion WT cells recruited SigA more efficiently than ΔssaA cells, indicating that SigA recruitment was promoted by 6S RNA when cells switched from nitrogen depletion to repletion.
The amount of RNAP-SigC increased in both strains during nitrogen deficiency and in the beginning of recovery phase. However, the SigC recruitment during the early recovery phase was considerably more pronounced in ΔssaA than in WT. Cells containing SigC as the only functional group 2 σ factor (the ΔsigBDE strain) showed a delayed recovery from nitrogen starvation [42] and SigC has been suggested to regulate nitrogen metabolism in stationary phase [59]. Altogether, these results suggest that recruitment of SigC by RNAP core prevent exit of cells from the stress-adapted state.
Likewise, SigB was recruited in larger quantities in ΔssaA at the beginning of recovery. Cells without stress inducible SigB are vulnerable to many stress conditions including heat [60] and high salt [61,62], and SigB helps to keep photosynthesis active under nitrogen deficiency [42]. It remains to be elucidated which role SigB might play in activation of the sigA gene during recovery from nitrogen deficiency induced stationary phase. However, a possible connection can be derived from previous experiments with dark/ light transitions [63].
The amount of SigE increases during nitrogen starvation and early recovery in both strains ( Fig. 7h). However, the recruitment of SigE was less efficient in ΔssaA than in WT in the early recovery phase. SigE has been earlier suggested to function in nitrogen stress responses [38,41], and its role in triggering sugar catabolism is well-characterized [41,43,64]. Since glgX and glgP are positively controlled by SigE [65], the observed effect on SigE recruitment might underlie the decelerated glycogen degradation of the mutant strain. An anti-σ factor, ChlH, has been proposed to inhibit SigE recruitment in a light-dependent manner [66]. Moreover, the sRNA SyR11, which is presumably involved in post-transcriptional activation of sigE expression (unpublished results), showed significantly reduced levels in ΔssaA, particularly during early recovery. Therefore, more detailed research on the involvement of integrated post-translational effects and -certainlyriboregulatory events will further disclose the complexity of the cyanobacterial σ factor network during physiological transitions. While not a conspicuous feature in this comparative study, the asRNA slr0653-as4 appears to be a potent inhibitor of sigA expression during starvation [45], exemplifying the significance of the RNA-regulatory layer in this process.

Conclusion
Summarizing our current knowledge, we suggest a probable mechanism for 6S RNA-dependent regulation within the acclimation process to changing nitrogen supply. In Synechocystis 6803, formation of RNAP-SigA holoenzymes drastically decreases during nitrogen deprivation while recruitment of different group 2 σ factors either increases (SigC and SigE) or remains at the same level (SigB and SigD) as under standard conditions. After prolonged nitrogen deficiency, transcription activity is low in general [45], most probably because expression of household genes is depressed due to low formation of RNAP-SigA holoenzymes. During the recovery phase, group 2 σ factors are first to be efficiently recruited by RNAP core, in order to be replaced later on by the primary σ factor.
6S RNA regulates this process by facilitating the replacement of group 2 σ factors by the primary σ factor. Due to a delayed formation of RNAP-SigA holoenzymes in the ΔssaA strain, activation of many household genes occurs at a slower rate resulting in a decelerated re-synthesis and activation of light harvesting and photosynthetic complexes.
Hence, with our study we extend the established function of 6S RNA towards the recruitment of housekeeping-versus-stress σ factors by RNAP to cyanobacteria. Thereby, 6S RNA is mediating their transcriptional response and supports a rapid adaptation to nutrient -in this case nitrogendepletion and repletion. Further, our findings imply that cyanobacterial 6S RNA might discriminate not only between group 1 and group 2 but also between the different group 2 σ factors.

Construction of 6S RNA mutants and RNAP-HIS mutants
For the construction of a 6S RNA deletion mutant (ΔssaA), the complete Synechocystis 6803 sequence region of the ssaA gene was deleted by replacement with a kanamycin-resistance cassette (Km R ), which was inserted between the chromosomal nucleotide positions 1886182 and 1887765. As distinguished from the previously described ΔssaA mutant [28], the Km R cassette harbored its own promotor and was flanked by ~700 bp regions upstream and downstream of the ssaA gene for homologous recombination. The ΔssaA-mutant was selected on BG11 agar plates containing 50 µg ml -1 kanamycin. PCR analysis was used to verify complete segregation of WT chromosomal copies (Fig. 1B). The primers P1 (ΔssaA_700up-Fw) and P2 (ΔssaA_700down-Rv) are listed below in Table S1.
To create a strain overexpressing 6S RNA (6S(+)), the ssaA gene locus and its promoter region (~150 bp of upstream region) were inserted into the self-replicating vector pVZ321 [69], containing a chloramphenicol resistence gene using XhoI and XbaI. The vector was then transferred to WT cells by conjugation. Exconjugants were selected on BG11 agar plates containing 10 μg ml -1 chloramphenicol.
Complementation of the ΔssaA strain (ΔssaA-c) was achieved by transferring the pVZ321 vector containing the ssaA gene plus ~150 bp of its upstream region and a chloramphenicol resistance gene to ΔssaA mutant cells by conjugation. The ΔssaA-c-mutant was selected on BG11 agar plates containing 50 µg ml -1 kanamycin and 10 μg ml -1 chloramphenicol. For the generation of a WT-RNAP-HIS and ΔssaA-RNAP-HIS mutant strain, Synechocystis 6803 WT and ΔssaA cells were transformed using the plasmid pC-HIS-rpoC1-CAP-I-I (containing a nine amino acids long His-tag attached to the C terminal end of the γ subunit, and a chloramphenicol resistance cassette after the rpoC1 gene) which was described earlier by Koskinen et al. [51]. WT-RNAP-HIS and ΔssaA-RNAP-HIS mutants were probed on BG11 plates containing 10 μg ml -1 chloramphenicol and full segregation of the mutants was tested by PCR analysis using primers rpoC1His5 and rpoC1His6 (Table S1) from Koskinen et al. [51].

Analysis of growth, pigmentation and photosynthetic capacity
Whole cell absorbance spectra were recorded from 550 to 750 nm on a Specord® 200 PLUS spectrophotometer (Analytik Jena, Jena, Germany) and were corrected for a residual scattering at 750 nm. Chlorophyll was extracted in 90% methanol and the content was measured by spectrophotometry [70]. Contents of phycocyanin and allophycocyanin were determined in the soluble protein fraction according to Tandeau de Marsac and Houmard [70]. The pigment contents were normalized to the optical density at 750 nm. Light saturated oxygen evolution rates of the WT and ∆ssaA mutant strain were measured with a Clark-type oxygen electrode (Hansatech) in BG-11 medium in vivo in the presence of 10 mM NaHCO3 at 32 °C, illuminated with white light (photosynthetic photon flux density 3000 µmol m -2 s -1 ) [61]. To determine PSII activity samples were supplemented with 1 mM 2,6-dimethoxybenzoquinone (DMBQ) and 1 mM ferricyanide. In preparation for 77 K fluorescence emission spectroscopy, cells were concentrated to 35 µg Chl ml -1 , and 50 µl samples were immediately frozen in liquid nitrogen. The fluorescence was measured at 77 K with orange light excitation (580 nm) using an Ocean Optics S2000 spectrometer. The spectra were corrected with a moving median, subtraction of the dark level and division by the peak value of PSI at 721 nm and setting this value to 1.

Determination of glycogen content
The content of glycogen per ml cell suspension was determined as described by Gründel et al. [71].
Cell pellets of 0.5 -1.5 ml culture were suspended in 30% (w/v) KOH and incubated for 2 h at 95 °C.
An addition of ice-cold ethanol to a final concentration of 75% (v/v) and a following incubation on ice for at least 1.5 h were required for the precipitation of glycogen. The isolated glycogen was centrifuged for 15 min (16000 g, 4 °C) and the pellet was washed with 70% and 98% (v/v) ice-cold ethanol and dried for 10 min at 60 °C. For the enzymatic hydrolysis of glycogen to glucose, the pellet was suspended in 100 mM sodium acetate (pH 4.5) including 2 mg/ml amyloglycosidase and incubated for 4 h at 95 °C. To determine the content of glucose, a hexokinase reagent was used according to the manufacturer´s protocol and measured spectrophotometrically.

Isolation of RNA polymerase complex and immunoblot analysis
Cells from WT-RNAP-HIS and ΔssaA-RNAP-HIS mutant strain (50 ml, OD750nm = 0.8 -1.1) were collected from growth at standard conditions, after 6d of growth in nitrogen depleted medium or after 1h, 4h and 22h of recovery in nitrogen replete medium. The samples were broken and the soluble protein fraction was obtained as described in Koskinen et al. [51]. For the isolation of RNAP the Dynabeads®HIS-Tag Isolation and Pulldown Kit was applied according to the manufacturer´s instructions. 900 µg of soluble protein fraction was used for the pulldown and RNAP complexes were eluted to 75 µl of elution buffer [51]. 10 µl of His-tag purified RNAP samples were subjected to 10% NEXT GEL SDS-PAGE according to the protocol of Koskinen et al. [51]. RNAP complexes were transferred to Immobilon-P membrane using the semidry blotting method followed by incubation with specific polyclonal antibodies against σ factors SigA, SigB, SigC and SigE [51,72,73]. Proteins were visualized and quantified as described by Koskinen et al. [51]. Membranes were re-probed with antibodies against α or β subunits of the RNAP core [73] and the σ factor content was normalized to α (SigA, SigC, SigE) or β subunit content (SigB). Similarly, 5 g of soluble proteins for SigA, 25 g for SigB, 15 g for SigC and 25 g for SigE were used for separation on SDS-PAGE and were detected as described above.

RNA isolation and Northern blot hybridization
For the isolation of RNA, samples from discrete stages of cultivation were taken at the indicated times and were immediately put on ice and spun down at 4 °C. The preparation of total RNA from Synechocystis 6803 was essentially performed as described previously [74], using the Hot Trizol method, followed by 2-propanol precipitation. To determine the concentration and to check the quality of total RNA, we made use of a NanoDrop N-10000 spectrophotometer and electrophoretic separation on 1.3% agarose-formaldehyde gels. The correct deletion of the ssaA gene was tested by  (Table S1).

Microarray analysis
The microarray design, hybridization procedure and data analysis have been described previously [50,75]. For the microarray analysis total RNA samples from two biological replicates were used. Raw data were processed with the R package limma. Median signal intensity was cyclic loess normalized.
The p values were generated by limma [76] and adjusted for multiple testing with the Benjamini Hochberg method.

Competing interests
The authors declare that they have no competing interests.

Availability of data and materials
The microarray data generated during the current study are available in the database

Consent for publication
Not applicable.

Ethics approval and consent to participate
Not applicable.   and in NaNO3-supplemented medium for 1h, 4h, 22h, and 48h (+N). Three independent biological replicates were measured and the error bars represent the standard deviation.