Identification of the direct regulon of NtcA during early acclimation to nitrogen starvation in the cyanobacterium Synechocystis sp. PCC 6803

Abstract

In cyanobacteria, nitrogen homeostasis is maintained by an intricate regulatory network around transcription factor NtcA. Although mechanisms controlling NtcA activity appear to be well understood, its regulon remains poorly defined. To determine the NtcA regulon during the early stages of nitrogen starvation for the model cyanobacterium Synechocystis sp. PCC 6803, we performed chromatin immunoprecipitation, followed by sequencing (ChIP-seq), in parallel with transcriptome analysis (RNA-seq). Through combining these methods, we determined 51 genes activated and 28 repressed directly by NtcA. In addition to genes associated with nitrogen and carbon metabolism, a considerable number of genes without current functional annotation were among direct targets providing a rich reservoir for further studies. The NtcA regulon also included eight non-coding RNAs, of which Ncr1071, Syr6 and NsiR7 were experimentally validated, and their putative targets were computationally predicted. Surprisingly, we found substantial NtcA binding associated with delayed expression changes indicating that NtcA can reside in a poised state controlled by other factors. Indeed, a role of PipX as modulating factor in nitrogen regulation was confirmed for selected NtcA-targets. We suggest that the indicated poised state of NtcA enables a more differentiated response to nitrogen limitation and can be advantageous in native habitats of Synechocystis.

INTRODUCTION

Cyanobacteria perform oxygenic photosynthesis and play key roles in the global carbon and nitrogen cycles (1,2). They are continuously exposed to environmental fluctuations, such as changes in nutrient availability, light conditions or temperature, and have developed sophisticated mechanisms to sense and respond to these fluctuations to maintain their metabolic homeostasis. This is also the case for nitrogen, an essential element necessary for the synthesis of molecular building blocks, such as amino acids and nucleotides. Its deficiency results in a gradual decrease in transcripts encoding for components of photosynthesis, e.g. the tricarboxylic acid cycle (TCA), the Calvin–Benson cycle, and protein synthesis (3–5). To counteract nitrogen limitation, systems for high-affinity nitrogen uptake and sugar catabolic genes are induced, including those involved in the pentose phosphate pathway (OPP) or glycogen metabolism (3,4). Cyanobacteria prefer ammonium as a nitrogen source, although they can also use nitrate, nitrite, urea and some amino acids. In addition, many cyanobacteria are able to fix N₂ (2). Nitrogen compounds acquired by cyanobacteria are converted to ammonium, which is then incorporated into the carbon skeleton of 2-oxoglutarate (2-OG) through the glutamine synthetase (GS)-glutamine oxoglutarate aminotransferase (GOGAT) cycle. Low intracellular ammonium levels during nitrogen starvation limit the turnover of the GS-GOGAT cycle resulting in increased levels of 2-OG, which serves as an indicator of an unbalanced C/N ratio within the cells (6).

In cyanobacteria, the global regulator for nitrogen assimilation and metabolism is NtcA, a transcription factor belonging to the CRP (cAMP receptor protein) family (7,8). NtcA is highly conserved in cyanobacteria and controls the cellular response to nitrogen availability (including nitrogen fixation in diazotrophic cyanobacteria) by binding as dimer to a consensus sequence GTAN₈TAC within the promoter of its target’s genes (8). In the absence of ammonium, NtcA activates the expression of genes for nitrogen assimilation pathways, including urtA, nirA, ntcB and glnA (8). NtcA also acts as a transcriptional repressor of some genes, such as gifA and gifB, that encode for the GS inactivating factors IF7 and IF17, respectively (8,9). Under nitrogen depletion, the accumulation of the metabolite 2-OG stimulates DNA-binding of NtcA as well as the transcriptional modulation of targets genes by NtcA. Maximal activation of NtcA requires the subsequent binding of the coactivator, P_II interacting protein (PipX), a small monomeric protein conserved among several cyanobacteria (10). This interaction is modulated by both 2-OG levels and the signal transduction protein P_II, an integrator of the nitrogen and carbon balance in bacteria and plants (11). When nitrogen is abundant, P_II binds to PipX to counteract NtcA activity at low 2-OG levels. However, under conditions of low nitrogen abundance (high 2-OG levels), P_II binds 2-OG in a cooperative manner with adenosine triphosphate (ATP) and is phosphorylated (1). This causes the release of the PipX, and its interaction with NtcA, stabilizing the active 2-OG-bound conformation of NtcA (12).

Previous studies based on transcriptomic and bioinformatic predictions have attempted to identify putative binding sites of NtcA in different cyanobacteria (13–15). In Synechocystis sp. PCC 6803 (hereafter Synechocystis), which serves as a model cyanobacterium and promising microbial cell factory, 48 putative binding sites were computationally predicted in an early study; but no experimental validation of these sites was carried out (13). For Anabaena sp. PCC 7120, two genome-wide studies, one based on chromatin immunoprecipitation followed by next-generation sequencing (ChIP-seq) analysis, and another based on the identification of transcriptional start sites (TSS) by RNA sequencing (RNA-seq) under nitrogen-depleted conditions, showed a great discrepancy in the number of possible NtcA-regulated genes. The RNA-seq data obtained by Mitschke et al. (15) suggested 158 TSS as potential NtcA targets, while the ChIP-seq analysis of NtcA performed by Picossi et al. (14) returned 2424 putative NtcA binding DNA regions, 865 of them ascribed to promoter regions. This striking discrepancy in the number of NtcA targets is not surprising, as separate application of RNA-Seq and ChIP-Seq provides only incomplete evidence for the regulatory activity of transcription factors. Although RNA-seq alone can faithfully detect changes in expression, it remains unclear whether these changes are related to the transcription factor of interest or reflect secondary effects caused by downstream events. In contrast, ChIP-seq alone can capture (differential) binding of transcription factors, but does not provide an indication of whether this causes activation or repression of nearby genes. A combination of these complementary high-throughput techniques, however, overcomes the limitations of individual approaches and can determine the NtcA regulon with unprecedented resolution. Through integration of differential expression and chromosomal binding location, we could identify not only in vivo functional NtcA binding sites, but also whether NtcA binding repressed or induced gene transcription during the early phase of nitrogen starvation. We identified 51 genomic regions bound by NtcA in ammonium-replete conditions, and 141 regions after 4 h of nitrogen starvation. Parallel transcriptome profiling revealed 669 genes as differentially expressed between these two conditions. Integration of NtcA binding and RNA-seq data classified 51 genes as being directly activated by NtcA and 28 as being directly repressed, including eight non-coding RNAs (ncRNAs). Direct target genes encoded mainly for proteins known to be involved in nitrogen and carbon metabolism, photosynthesis, respiration and transport, as well as various proteins without functional annotation. Interestingly, we observed significant differential expression for some genes, despite unchanged NtcA binding in their promoter regions as well as NtcA binding with only delayed expression changes of associated genes. Both observations suggest involvement of additional regulatory elements and potential different states of NtcA. A modifying role of PipX was specifically examined in selected NtcA target genes, confirming that PipX assists NtcA with nitrogen control in Synechocystis.

MATERIALS AND METHODS

Cyanobacterial strains and growth conditions

A glucose-tolerant strain of Synechocystis sp. PCC 6803 was grown in flask culture at 30°C under constant illumination (45 μmol photons m⁻² s⁻¹) on a rotatory shaker in liquid BG11₀C medium (16), supplemented with 10 mM NH₄Cl and 20 mM TES (BG11₀C−NH₄). The pipX(ssl0105)-disruptant mutant (ΔpipX) was grown under the same conditions, except that 50 μg ml⁻¹ of kanamycin was added. To induce nitrogen starvation, flask cultures of Synechocystis cells, growing in BG11₀C−NH₄, at linear growth phase (3–6 μg Chl/ml; Supplementary Figure S1) were collected, washed twice with BG11₀C and resuspended in BG11₀C medium under the same growth conditions for 4 h. Samples in the control treatment were also washed twice in BG11₀C supplemented with NH₄ and re-suspended in BG11₀C-NH₄ medium. For all experiments, nitrogen starvation and ammonium-cultured cells were resuspended in the same medium volume at similar cell density.

Construction of ΔpipX mutant Synechocystis strain

A 1639-bp DNA fragment, lacking a 211-bp internal fragment of pipX, was constructed by two-step polymerase chain reaction (PCR), using oligonucleotides pairs pipX-5’-HindIII and pipX-3’-XbaI, as well as pipX-deletion-for and pipX-deletion-rev (Supplementary Table S1). The resulting 1617-pb HindIII-XbaI restriction fragment was cloned into pBS II KS(+). An antibiotic resistance C.K1 cassette, which confers kanamycin resistance (Km^r), was inserted into a BamHI site in the positive transcription orientation to generate pPipX. Transformation of Synechocystis cells with pPipX was carried out. Correct integration of the C.K1 cassette and total segregation of the mutant chromosomes into the Synechocystis ΔpipX mutant strain was confirmed by PCR (Supplementary Figure S2).

Chromatin immunoprecipitation and sequencing procedure

Aliquots of 250 ml of ammonium and nitrogen-starved cultures were used for chromatin immunoprecipitation. To achieve protein–DNA crosslinking, formaldehyde was added to these cultures, yielding a final concentration of 1% and incubated for 15 min at room temperature, with occasional gentle shaking. The crosslinking reaction was terminated by adding 125 mM of glycine followed by a 5-min incubation at room temperature, with occasional gentle shaking. Next, cells were filtered, washed with cold TBS (20 mM Tris–HCl, pH 7.4, 140 mM NaCl) and collected in tubes (50 ml of culture per tube). Cell lysis was carried out, as previously described (14). The lysate was sonicated (15 cycles of 10 s at 10% amplitude, with 40 s on ice between cycles) to fragment chromosomal DNA into sequences of sizes between 100 and 400 bp. Cell debris was removed by centrifugation (15 min at 10 000 × g, 4°C). Chromatin was collected before immunoprecipitation to serve as a control input sample. Immunoprecipitation of NtcA-bound chromatin by an anti-NtcA antibody (generated in our laboratory using purified NtcA protein from Synechocystis, injected in rabbits according to standard immunization procedures) was carried out, as described by Picossi et al. (14). This protocol was repeated four times, using cells from independent inductions. The resulting DNA was pooled on the same DNA purification column (miniElute; QIAGEN, Hilden, Germany) to obtain 40 and 20 ng samples from ammonium and nitrogen-depleted conditions, respectively. Prior to construction of sequencing libraries, quantitative Real Time PCR (qRT-PCR) was performed to assess the enrichment of the promoter region of glnA and glnB in the immunoprecipitated samples, compared with the control input sample. Enrichment of a promoter region in the ChIP sample was determined using the percent input method: 100*2^∧[C_t(IP-sample)−C_t(Adjusted-Input)], with the control input sample adjusted to give 100%, given that 5% of starting chromatin was used. A CFX Connect RT-PCR machine (Bio-Rad Laboratories, Inc., Hercules, CA, USA) and ssoFast EvaGreen Supermix (Bio-Rad Laboratories) were used for qRT-PCR. The sequences of the primers used for the qRT-PCR to validate ChIP are listed in Supplementary Table S2.

Illumina libraries (Illumina, San Diego, CA, USA) were prepared from 32.5 and 12.5 ng of immunoprecipitated DNA from ammonium and nitrogen-depleted samples, respectively, as well as from the two controls DNA (input samples), using the Illumina TruSeq ChIP-seq DNA sample preparation kit. DNA sequencing was performed on the Illumina HiSeq 2500 platform, using single-end 50 bp sequencing. A total of 153 894 213 reads were obtained for four samples (Supplementary Table S3). The Finnish Microarray and Sequencing Centre (FMSC, Turku, Finland) conducted all DNA-sequencing and library preparation.

ChIP-seq peak calling analysis

Raw reads were mapped against the Synechocystis genome (NCBI Reference sequences: NC_000911.1 (chromosome), NC_005229.1 (plasmid pSYSM), NC_005230.1 (plasmid pSYSA), NC_005231.1 (plasmid pSYSG) and NC_005232.1 (plasmid pSYSX)) using Bowtie2 (17). The resulting BAM files were processed using SAMTools and BEDTools (18,19). BAM files were normalized for visual inspection in the Integrative Genomics Viewer (IGV) (20), using BamCoverage from DeepTools2 (21). Peaks for NtcA were identified by two peak calling algorithms: MACS (v1.4.1) (22) and BayesPeaK (v1.22.0) (23). In both algorithms, the background noise of unspecific binding was modeled using the input data (control DNA). Surrounding genes were retrieved using the Bioconductor package ChIPseeker (v1.6.7) (24). Peaks were also visually inspected for artefacts, and false positives were removed.

RNA preparation and differential expression analysis

RNA from Synechocystis cultures was extracted using the PGTX 95 RNA extraction protocol, described in Pinto et al. (25). Samples for RNA-seq were taken simultaneously from the same cultures used for the chromatin immunoprecipitation experiments. The quantity and quality of total RNA were evaluated using RNA electropherograms (Agilent 2100 Bioanalyzer; Agilent Technologies, Santa Clara, CA, USA). The Ribo-Zero Magnetic Kit for Bacteria (Epicentre; Illumina) was applied to remove ribosomal RNA from each sample. Two biological replicates for each condition of RNA samples were analyzed using the Ion PGM Template Hi-Q OT2 kit and Ion Touch 2 Instrument (Thermo Fisher Scientific, Inc., Waltham, MA, USA) and subsequently sequenced on the Personal Genome Machine (Ion PGM; Thermo Fisher Scientific), with reagents from Ion Hi-Q Sequencing kit (STAB VIDA, Lda, Lisbon, Portugal), following the manufacturer’s instructions (Supplementary Table S4). Reads were aligned against the NCBI genome sequence for Synechocystis (NC_000911.1, NC_005229.1, NC_005230.1, NC_005231.1 and NC_005232.1), applying the Torrent Mapping Alignment Program (TMAP; available from https://github.com/iontorrent/TMAP). Raw read counts were calculated using the HTSeq Python script from HTSeq-count (26). The Bioconductor DESeq package from R software (27) was used to detect differentially expressed genes under ammonium repletion and nitrogen starvation. An adjusted P-value of <0.1 was considered to be significant. For Gene Ontology (GO) enrichment analysis, the GSEA tool (28) and Synergy (29) (available at http://synergy.plantgenie.org), were applied. Network construction and visualisation were carried out using Cytoscape 3.2.0 (30).

Hierarchical complete-linkage clustering was performed with Cluster 3.0 (31). Clusters were visualized using Java TreeView software (32). Differential expression was displayed as heat maps, using a color range from yellow to blue.

Potential target genes of NsiR7 were identified with the IntaRNA algorithm v.2.0.2 (33), using default parameters and a window of 275 nt around the start codon (200 upstream and 75 downstream). A threshold P-value of <0.005 was used.

Northern blot analysis

Total RNA was isolated from 30 ml samples of Synechocystis cultures in the mid-exponential growth phase (3 to 4 μg chlorophyll ml⁻¹). Extractions were performed by vortexing cells in presence of phenol–chloroform and acid-washed baked glass beads (0.25–0.3 mm diameter), as described by Garcia-Dominguez and Florencio (34). An aliquot of 5 μg of total RNA was loaded per lane, and electrophoresed in 1.2% agarose denaturing formaldehyde gels (35). For the ncRNA analysis, RNA samples (5–10 μg) were separated on 6% urea-polyacrylamide gels for 3 h at 25 mA. In both cases, migrated gels were later transferred to nylon membranes (Hybond N-Plus; GE Healthcare, Little Chalfont, UK). Prehybridization, hybridization and washes were carried out in accordance with GE Healthcare instruction manuals. Probes for northern blot hybridization were prepared by PCR, using primers shown in Supplementary Table S2. DNA probes were ³²P-labeled, with a random-primer kit (Amersham Biosciences, GE Healthcare), using [α-³²P] dCTP (3000 Ci/mmol). Hybridization signals were quantified with a Cyclone Plus storage phosphor scanner (PerkinElmer, Inc., Waltham, MA, USA). Each experiment was replicated at least twice.

Western blot analysis

To prepare western blot analyses, 5 μg of total proteins from soluble extracts were fractionated on sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS/PAGE) and immunoblotted (35) with antibodies against: thioredoxin A (1:3000) (36); IF7 and IF17 (1:2000) (37); P_II (1:4000) and GS III (1:20000) (34). ECL Prime (GE Healthcare) was used to detect the different antigens, with anti-rabbit secondary antibodies conjugated to horseradish peroxidase (1:25 000).

Glutamine synthetase assay

GS activity was determined in situ by a Mn²⁺-dependent γ-glutamyl-transferase assay in cells permeabilized with mixed alkyltrimethylammonium bromide (MTA) (38). The same assay, but without MTA, was performed in cells from the same samples in parallel. One unit of GS activity corresponds to the amount of enzyme that catalyses the synthesis of 1 μmol min⁻¹ of γ-glutamylhydroxamate.

Gel retardation assays

DNA fragments used in the binding assays were obtained by PCR, with the corresponding oligonucleotides pairs (Supplementary Table S4). DNA probes, including NtcA-binding consensus sequences, were cut with NotI restriction enzyme, generating fragments of ∼200 bp. These fragments were end-labeled with [α-³²P]dCTP, using Sequenase version 2.0 enzyme. The GST-NtcA fusion protein was expressed and purified, as previously described (39). DNA radiolabeled fragments (0.5 nM) were incubated with purified NtcA (0.1–0.4 μM), and with 2-OG (0.6 mM), when indicated. The binding reaction with the corresponding DNA fragment was carried out in a final volume of 15 μl within binding buffer (12 mM HEPES-NaOH pH 8.0, 8 mM Tris–HCl pH 8.0, 10% (w/v) glycerol, 0.5 mM ethylenediaminetetraacetic acid pH 8.0, 100 mM KCl, 2 mM MgCl, 0.05 μg/μl poly (dI–dC), 0.01 μg/μl bovine serum albumin and 1 mM dithiothreitol (DTT)). These mixtures were incubated at 25°C for 20 min, and the DNA–protein complexes were separated on non-denaturing 6% (w/v) polyacrylamide gel. Gels were dried and imaged using a Cyclone Plus storage phosphor scanner (PerkinElmer).

Quantitative real-time polymerase chain reaction analysis

The qRT-PCR was performed in an iQTM5 multicolor RT-PCR detection system (Bio-Rad), in a 10 μl reaction volume using the ssoFast EvaGreen Supermix (Bio-Rad). The sequences of primers used for the qRT-PCR for RNA-seq validation are listed in Supplementary Table S2. The efficiency of the PCR was calculated using the program LinRegPCR (40). Normalized data were calculated by dividing the average of at least three replicates of each sample from the candidate and the reference gene, rnpB.

RESULTS

Nitrogen starvation response of Synechocystis

To characterize the NtcA regulon in response to nitrogen deprivation in Synechocystis sp. PCC 6803, we combined the two powerful genome-wide profiling techniques, RNA-seq and ChIP-seq. Samples were taken from wild-type (WT) cells grown in a medium with ammonium (NH₄) as nitrogen source and after their incubation in a medium depleted of combined nitrogen (−N) for 4 h (Figure 1A). GS activity was measured to verify that nitrogen starvation was induced under the experimental conditions. Results obtained for GS activity were consistent with a transition from a nitrogen-rich medium to a nitrogen-depleted medium, showing higher activity after 4 h (71.92 ± 7.93 U/mg*chl⁻¹) in the depleted medium (Figure 1B). After 24 h in a nitrogen-depleted medium, cell growth arrested and a yellow appearance (bleaching) typical of the chlorotic cultures was observed (Figure 1C), consistent with degradation of the phycobilisomes (PBS) related to prolonged nitrogen starvation.

Figure 1.

Nitrogen starvation conditions for Synechocystis sp. PCC 6803. (A) Schematic representation of the experimental procedure to capture the early acclimation to the nitrogen starvation examined by RNA-seq and ChIP-seq in this study, before prolonged starvation leads to chlorosis. (B) In situ analysis of GS activity from Synechocystis cells growing in BG11C+NH₄ or after incubation for 4 h in nitrogen-free BG11C medium. A sample of 1 ml of Synechocystis culture was used for each assay. Transferase activity was measured after 5 min of incubation. Activity measures represent arithmetic means of three independent experiments, and their standard deviation values. C, control without adding cells. (C) Image of Synechocystis cultures used for RNA-seq and ChIP-seq experiments, after growing in either BG11C+NH₄ or nitrogen-free BG11C medium for 24 h.

Transcriptional profiling by RNA-seq of nitrogen starvation response

To study the transcriptional response after the transition from an NH₄-replete to a N-depleted medium, total RNA obtained under both conditions was sequenced. Samples taken after 4 h of nitrogen starvation were compared with samples taken at the zero time point (with NH₄; Figure 2A). There were 1080 genes significantly regulated (adj.P < 0.1; Supplementary Table S6), of which 669 genes showed more than a 2-fold change in expression. Of these latter ones, 332 and 337 genes were up- and downregulated, respectively (Supplementary Tables S7 and 8). RNA-seq analysis also enabled detection of small open reading frames (ORFs) and ncRNAs, which were not captured by previous DNA microarray analyses (3,4). In total, 21 ncRNAs were detected as differentially expressed, including NsiR4 (Supplementary Tables S7 and 8), which has been recently identified as playing a role in nitrogen control in Synechocystis (41). To identify processes and pathways involved in the transcriptional response to nitrogen starvation, we performed GO analysis of the differentially expressed genes. Our results reveal that downregulated genes tended to be associated with biosynthetic processes, gene expression and translation, indicating a reduction of overall protein synthesis (Figure 2B). Upregulated genes were significantly enriched in genes encoding proteins involved in nitrogen assimilation. Remarkably, genes related to the photosynthetic machinery, especially those coding for photosystem I (PSI) and PBS, were also overrepresented among the upregulated genes.

Figure 2.

Global response to nitrogen starvation in Synechocystis sp. PCC 6803. (A) MA plot: a scatterplot of log₂-fold-change (−N/NH₄) versus average expression in log₂ scale for each gene, produced using the DEseq2 package. Dots shown in red indicate differentially expressed genes with adj. P-value of < 0.1. The nine genes with the highest induction and repression are shown in blue and yellow, respectively. (B) GO enrichment analysis of differentially expressed genes under nitrogen starvation conditions. Only GO of biological processes and cellular components having a false-discovery rate of <0.1 are shown.

Nitrogen assimilation

Adjustment of nitrogen assimilation pathways is crucial for survival under nitrogen-limited conditions. This is reflected in the expression pattern, with 24 genes involved in nitrogen uptake and assimilation being differentially expressed after nitrogen depletion (Table 1). Seven of them (glnA, urtA, amt1, urtB, glnN, bgtB and amt2) are among the 10 most-induced genes (Figure 2A). They include genes encoding the high-affinity nitrate/nitrite transporters (nrtBACD), the glutamine permease (bgtA and bgtB), the urea transport system (urtADBC) and the ammonium permeases (amt1 and amt2). Thus, after 4 h of nitrogen starvation, all scavenging nitrogen systems were activated to compensate for the loss of combined nitrogen from the media.

Table 1. Selected genes involved in nitrogen assimilation which expression is altered after 4 h of nitrogen starvation.

Gene	Symbol	Function	Log2 (ratio)	P-value
Regulators of nitrogen metabolism
ssl0707	glnB	nitrogen regulatory protein PII	3.83	2.6E-101
ssl1911	gifA	GS inactivating factor IF7	−4.20	4.3E-96
sll1515	gifB	GS inactivating factor IF17	−2.32	4.3E-23
Nitrate/Nitrite assimilation
slr0898	nirA	ferredoxin–nitrite reductase	1.86	1.6E-06
sll1450	nrtA	nitrate transport	3.45	2.7E-25
sll1451	nrtB	nitrate transport	2.51	1.5E-05
sll1452	nrtC	nitrate transport	2.39	8.7E-19
sll1453	nrtD	nitrate transport	2.16	2.0E-06
Glutamine/glutamate assimilation
slr1756	glnA	glutamate–ammonia ligase GSI	5.38	9.2E-191
slr0288	glnN	glutamate–ammonia ligase GSIII	4.64	4.5E-117
sll1502	gltB	glutamate synthase large subunit	0.96	9.4E-09
sll1027	gltD	glutamate synthase small subunit	0.50	1.4E-02
slr0710	gdhA	glutamate dehydrogenase (NADP+)	0.93	2.1E-03
slr1735	bgtA	Component of ABC-type Bgt permease	1.27	1.8E-05
sll1270	bgtB	Component of ABC-type Bgt permease	4.19	7.0E-81
Ammonium asimilation
sll0108	amt1	ammonium/methylammonium permease	5.32	2.8E-216
sll1017	amt2	ammonium/methylammonium permease	4.12	2.7E-15
Urea assimilation
slr0447	urtA	Component of ABC-type urea transport system	5.36	7.7E-212
sll0764	urtD	Component of ABC-type urea transport system	1.48	6.5E-09
slr1200	urtB	Urea transport system permease protein	5.19	8.1E-71
slr1201	urtC	Urea transport system permease protein	3.33	4.7E-17

Nitrogen in the form of ammonium is incorporated into amino acids by sequential action of GS and glutamate synthase (GOGAT) (Figure 3). Significant upregulation was observed for glnA and glnN encoding GS type I and III, respectively (Figure 3). The upregulation of GS coincided with a strong downregulation of genes encoding the GS inactivating factors IF7 (gifA) and IF17 (gifB) (Figure 3). Both NADH-GOGAT and Ferredoxin-GOGAT convert glutamine to glutamate, with 2-OG provided by the isocitrate dehydrogenase (icd). This reaction constitutes a crossroad between nitrogen and carbon metabolism (Figure 3). Transcripts of gltB and gltD encoding the large and the small subunits of the NADH-GOGAT were slightly induced (Figure 3 and Table 1). Remarkably, icd together with the genes encoding the phosphoenol pyruvate carboxylase (ppc) and the pyruvate kinase (pyk1) genes were also upregulated, facilitating enhanced synthesis of 2-OG in response to nitrogen starvation (Figure 3; Supplementary Tables S6 and 7). Surprisingly, expression of ntcA and pipX was not significantly affected, despite the observed differential expression of known NtcA targets (i.e. gifB, gifA glnA glnB, amt1 and icd). In contrast, transcripts encoding the regulatory protein P_II (glnB) were accumulated upon nitrogen depletion. P_II is involved in the regulation of central metabolism processes, integrating signals of cellular carbon, nitrogen and energy balances by binding 2-OG and ATP (42).

Figure 3.

Changes in expression of genes involved in nitrogen assimilation and carbon metabolism. This diagram shows metabolic pathways and metabolites of the TCA and GS/glutamine oxoglutarate amidotransferase (GOGAT) cycle, glycolysis, gluconeogenesis, pentose phosphate pathway (OPP) and glycogen metabolism. These pathways were predicted from the KEGG database (www.genome.jp/kegg/pathway.html). NtcA target genes obtained from ChIP-seq analysis are shown for −N (green square) or NH₄ (red square) treatments. Genes encoding metabolic enzymes are shown below; narB, nitrate reductase; nirA, nitrite reductase, gifA, GS-inactivating factor; gifB, GS-inactivating factor; glnA, GS type I; glnN, GS type III; gltD, glutamate synthase small subunit; gltB, GOGAT; glsF, ferredoxin-dependent glutamate synthase, gdhA, glutamate dehydrogenase; icd, isocitrate dehydrogenase; acnB, aconitate hydratase; citH, malate dehydrogenase; gltA, citrate synthase; odhB, pyruvate dehydrogenase; pyk1–2, pyruvate kinase; eno, enolase; ppc, phosphoenolpyruvate (PEP) carboxylase; gpmB, phosphoglycerate mutase; pgk phosphoglycerate kinase; gap1, glyceraldehyde-3-phosphate dehydrogenase (catabolic reaction); gap2, glyceraldehyde-3-phosphate dehydrogenase (anabolic reaction); tpi, glycogen isoamylase; cbbA, fructose-bisphosphate aldolase; fbp, fructose-1,6-bisphosphatase; pgl, 6-phosphogluconolactonase; pgm, phosphoglucomutase; glgC glucose-1-phosphate adenylyltransferase; glgP, glycogen phosphorylase; glgX, glycogen isoamylase; glgA, glycogen synthase; glgB, 1,4-alpha-glucan branching enzyme; zwf, glucose-6-phosphate dehydrogenase; gnd, 6-phosphogluconate dehydrogenase; tktA, transketolase; rpiA or B, ribose-5-phosphate isomerase; rpe, pentose-5-phosphate-3-epimerase; prk, phosphoribulokinase; rbcL, ribulose bisphosphate carboxylase large subunit; rbcS, ribulose bisphosphate carboxylase small subunit.

Photosynthesis, carbon assimilation and central carbon metabolism

We observed a major increase in the expression of genes encoding for almost all subunits of PSI and PBS. In addition, several subunits of the ATP synthase (atpI, atpC, atpH) and photosystem II (PSII), namely psbO, psbU, psbK, psbT and psbB, were upregulated (Supplementary Table S7). In contrast, CO₂ fixation appeared to be diminished after nitrogen starvation, as genes encoding the bicarbonate transporter (Na⁺/HCO₃⁻ symport), (bicA), as well as the carbon-concentrating mechanism proteins, ccmN and ccmM, were repressed (Supplementary Table S8). Additionally, ndhF3-ndhD3-cupA encoding subunits of the NDH–1MS complex associated with high-affinity CO₂ uptake were downregulated (43,44) (Supplementary Table S6). Although the expression of the genes encoding the two subunits of the ribulose-1,5,-bisphosphate carboxylase/oxygenase (RuBisCo) remained unaltered, expression of other genes encoding enzymes of the Calvin-Benson cycle was downregulated, especially for prk, pgk, tktA and rpiA (Figure 3 and Supplementary Table S8).

Genes involved in sugar and glycogen metabolism displayed divergent patterns. Genes encoding sugar catabolic enzymes were induced under nitrogen starvation, while genes for sugar anabolism were downregulated. In this case, increased expression of the oxidative pentose phosphate pathway (OPP) genes zwf, talB and gnd was observed, whereas expression of fbp (slr0952) encoding for the fructose-1,6-bisphosphatase, which acts exclusively in the direction of the gluconeogenesis, was diminished (Figure 3; Supplementary Tables S7 and 8). Similar divergence was recorded for glycogen metabolism. Both glgX (slr1857) and glgP (slr1356) involved in the catabolism of glycogen were upregulated; while glgB and glgC, which participate in the synthesis of this molecule, were down- and upregulated, respectively (Figure 3; Supplementary Tables S7 and 8).

To confirm and validate differential expression detected by RNA-seq, qRT-PCR was carried out for genes related to nitrogen and carbon metabolism (glnA, glnB, glnN, amt1, gifB, icd and rre37). For these genes, RNA-seq and qRT-PCR data provided very consistent results (Supplementary Figure S3).

ChIP-seq analysis of NtcA binding

ChIP with anti-NtcA antibodies enabled in vivo genome-wide detection of NtcA binding sites. For quality control, enrichment of the promoter regions of two bona fide NtcA targets, glnA and gifB, within the immunoprecipitated fraction was confirmed by qRT-PCR. Primers against the promoter region of the ORF, slr1875, which is not regulated by NtcA, served as a negative control. Our qRT-PCR results confirmed the specificity of the immunoprecipitation, as strong enrichment of the two known binding sites for NtcA was measured, especially after nitrogen depletion, in contrast to the negative control (Supplementary Figure S4).

For comprehensive detection of binding sites, two different peak-calling methods were applied, MACS (22) and BayesPeak (23). NtcA binding was identified, if peaks in the sequence coverage were detected by at least one of the two peak-calling algorithms and passed visual inspection. We identified 51 NtcA binding peaks in NH₄ and 141 peaks after 4 h of nitrogen depletion. Notably, 27 peaks were detected under both conditions (Figure 4A and B; Supplementary Tables S9 and 10). Three-quarters of the binding regions (151/192) were shorter than 450 bp, consistent with DNA fragment sizes obtained after DNA sonication. Thirty-two regions were slightly larger in size (500–850 bp), while only one region extended over more than 1 kbp (1200 bp). Visualization of this latter peak, revealed two peaks that overlapped each other within the bidirectional-promoter region for both sll0783 and slr0821 genes. Interestingly, sll0783 gene is reported to be involved in polyhydroxybutyrate accumulation under nitrogen starvation conditions in Synechocystis (45).

Figure 4.

Genome-wide NtcA DNA binding analysis. (A) NtcA binding across the genome was compared for NH₄ (blue track) or −N (red track) conditions. The x-axis indicates the genomic position of the ChIP-seq peaks, while the y-axis indicates the read count after each dataset was normalized using BamCoverage (Ramirez, F. 2016). Names of genes assigned to peaks with high read count are also shown. B. Venn diagram showing overlap of genes with significant binding by NtcA in +NH₄ versus −N conditions. Overlap of genes differentially expressed (P-value < 0.1 and fold change > 2) after nitrogen depletion versus genes with significant binding by NtcA are also shown. (C) Distribution of NtcA binding peaks for NH₄ and −N conditions. NtcA peaks were classified into four categories: intragenic region (green), gene promoter (blue), ncRNA promoter (yellow) and antisense promoter (purple). (D) Genes with NtcA binding peaks assigned to NH₄ or −N conditions, were grouped into functional categories according to the CyanoBase classification. FPS: Fatty acid, phospholipid and sterol metabolism; PPNN: Purines, pyrimidines, nucleosides and nucleotides; DNA_RRMRR: DNA replication, restriction, modification, recombination and repair.

The majority of chromosomal NtcA-IP peaks (73%) in the −N medium were located immediately upstream of gene coding sequences, ncRNAs and asRNAs, consistent with preferential NtcA binding to promoter regions under nitrogen-depleted conditions (Figure 4C). In contrast, only 46% of chromosomal peaks correspond to putative promoter regions in the NH₄ condition, and of those 92% were also present in the −N condition (Figure 4C). Interestingly, we detected 24 peaks that were present exclusively under NH₄ conditions. The vast majority of these peaks occurred within intragenic locations, with only four in promoter regions.

NtcA binding in promoter regions was assigned to transcribed genes, based on experimentally defined TSS in Synechocystis (5,46) and on gene expression data from our RNA-seq analysis. An unambiguous assignment to a single TSS was not possible for six of the NtcA binding peaks, because of the presence of nearby flanking genes. These peaks were assigned to both genes. In nine cases, we observed NtcA binding close to small RNAs, which were co-expressed with down-stream genes of the same transcriptional unit; a genetic feature that has recently been termed ‘actuaton’ (47). For these transcriptional structures, both coding genes and small RNAs were considered as NtcA-targets. Collectively, 18 ncRNAs and 5 asRNAs had an NtcA binding region within their promoter regions, with the exception of ncRNA ncl0530 and the downstream gene sigD, with its peak located within the ncl0530 sequence (Figure 5).

Figure 5.

Read coverage of potential NtcA targets detected by parallel differential profiling. The ChIP-seq and RNA-seq density profiles are shown in blue for NH₄ and in red for −N treatments. Examples of genes included in the NtcA regulon are displayed: NtcA activated (glnA) or repressed (tktA) genes, long transcripts containing ncRNA and downstream protein-coding genes that are transcribed by read-through over the ncRNA terminator (ncl0530-sigD); and genes with internal peaks (metX, glgB and pilA4). Values on the x-axis are the genomic coordinates. Arrows indicate transcriptional start sites (TSS) obtained from Mitschke et al. (46).

Genes identified as NtcA targets were classified into 16 functional categories (Figure 4D). About a third of the assigned genes (48) encode unknown or hypothetical proteins. The most represented functional categories were ‘amino acids biosynthesis’ and ‘transport and binding’. Both functional categories include most of the well-known NtcA-regulated genes, such as glnA, glnB, gifA, as well as several genes coding for the nitrogen uptake system (amt1, amt2, urtABCD, bgtA, bgtB, nrtABCD). Interestingly, NtcA binding sites were also found upstream of ten genes coding regulatory proteins. These included PipX (49), response regulator Rre37 (50), P_II (34) and NtcA, but also newly identified ones, i.e. the response regulators Rre8, Rre12 and SphR, and the predicted transcription factor Sll0782.

Electrophoretic mobility shift assays (EMSA) were performed to validate the capability of NtcA to bind in vitro to DNA fragments of binding regions determined by ChIP-seq (Supplementary Figure S5). As a positive control, the promoter region of glnA, which contains a verified NtcA binding site, was used (51). In contrast, a DNA fragment containing the promoter region of nrsR, which has not been described as an NtcA target, was used as the negative control. We selected seven binding regions, including those with high (sll0327, urtA, pilA4, ncl0350) and low (gltB, metX, glgP) fold enrichment, as identified by ChIP-seq analysis (for both NH₄ or −N conditions; Supplementary Tables S9 and 10). A strong positive correlation between ChIP-seq fold enrichment and NtcA EMSA affinities was found, indicating that our ChIP-seq analysis produced reliable results (Supplementary Figure S5).

Parallel differential profiling of transcriptome and NtcA binding sites

Occupancy of transcription factor binding sites alone gives only a weak indication of their regulatory potential. Therefore, we used parallel profiling of changes in both NtcA binding and transcription in response to nitrogen depletion. To define the NtcA regulon and to determine its dynamics, we intersected potential direct targets of NtcA, with their observed differential expression. This approach identified genes for which changes in NtcA binding were indicative of changes in transcriptional activity. We detected 79 genes (including ncRNAs and asRNAs) that fulfilled this criterion; they define the direct NtcA regulon in early phase of nitrogen starvation. Fifty-one of these genes were upregulated and 28 were downregulated, after 4 h of nitrogen depletion (Figure 4B and Table 2). Read coverage from RNA-Seq and ChIP-seq is shown for representative genes in Figure 5. The direct NtcA regulon (Figure 6) included eight enzymes involved in amino acid biosynthesis (glnA, glnN, nirA, argG, gltB, hisC, icd, aspA), six nitrogen compound transporter-related genes (urtA, urtB, amt1, amt2, bgtA, bgtB) and four regulators of nitrogen metabolism (gifA, gifB, glnB, rr37). Genes coding for proteins involved in the synthesis of the two main nitrogen storage pools in cyanobacteria, cphA and cpcB for cyanophycin and phycocyanin, respectively, were also targeted and regulated by NtcA, as previously described in Anabaena sp. PCC 7120 (14,52). Additionally, several genes that encode proteins related to carbon metabolism displayed NtcA-dependent regulation: transketolase (tktA; Figures 3 and 5), glycolate oxidase subunit (glcF), carbon dioxide concentrating mechanism protein (ccmk2), Sll0783 protein required for accumulation of polyhydroxybutyrate in Synechocystis (45), and glycogen phosphorylase (glgP; Figures 3 and 6). In particular, glgP contains an intragenic NtcA-binding peak under NH₄-replete condition suggesting a repressive role of NtcA, triggered by binding to an intragenic region and blocking transcription in nitrogen-rich media. This correlates well with the observed diminished expression of glgP in NH₄, compared with the −N conditions (Figure 5). Similar potential NtcA-regulation by intragenic binding was observed for genes, such as pilA4 in NH₄, or metX (Figure 5) and secA under −N conditions. We were particularly interested in identifying new NtcA-regulated genes. Our study reveals that NtcA coordinates a wide range of functions to cope with the early acclimation of nitrogen starvation, given that NtcA targets included genes coding for porins (apqZ and slr1841), the periplasmic protein involved in phosphate uptake (pstS), and the photomixotrophic growth protein A (pmgA). Furthermore, the NtcA regulon comprises 26 genes encoding hypothetical or unknown proteins, as well as asRNA sll1802-as and eight ncRNAs (Figure 6).

Table 2. The NtcA regulon.

Treatment	Regulated gene	Symbol	Peak start	Peak end	Region	log2 ratio (-N/NH4)	Gene function	TSS located inside peak
−N	slr1756	glnA	531291	531451	promoter	5.380	glutamate–ammonium ligase	531399
−N, NH4	slr0447	urtA	2098581	2098991	promoter	5.363	periplasmic protein, ABC-type urea transport system substrate-binding protein	2098760
−N	sll0108	amt1	2971052	2971551	promoter	5.318	ammonium/methylammonium permease	2971398
−N, NH4	slr1200	urtB	881702	882201	promoter	5.185	urea transport system permease protein	881923
−N, NH4	sll0327		2389552	2390401	promoter	5.141	unknown protein	2390100c
−N	slr0288	glnN	2128602	2128851	promoter	4.643	glutamate–ammonium ligase	2128736
−N	sll1270	bgtB	1115933	1116332	promoter	4.195	periplasmic substrate-binding and integral membrane protein of the ABC-type Bgt permease	1116162c
−N	sll1017	amt2	401952	402201	promoter	4.123	ammonium/methylammonium permease	402151
−N	sll0944		2267152	2267301	promoter	4.094	hypothetical protein	2267292
−N	ssl0707	glnB	2152602	2152751	promoter	3.826	nitrogen regulatory protein PII	2152741
−N	sll1119		874502	874651	promoter	3.823	hypothetical protein	874644
−N, NH4	slr2002	cphA	1447927	1448046	promoter	3.121	cyanophycin synthetase	1448016
−N	slr1289	icd	282874	283225	promoter	3.043	isocitrate dehydrogenase (NADP+)	283033
−N	slr0909		2785363	2785603	intragenic	2.874	unknown protein
−N	slr1912		615686	616010	promoter	2.874	anti-sigma F factor antagonist	615878
−N	ncl0540/sll1698	NsiR4/	1289152	1289451	promoter	2.861	hypothetical protein	1255935
−N	sll0733		3419302	3419651	promoter	2.783	unknown protein	3419527
−N, NH4	sll1330	Rre37	3295152	3295451	promoter	2.704	two-component system response regulator OmpR subfamily	3295334
−N	slr1142		801102	801851	promoter	2.657	hypothetical protein	801482
−N	ncl0350/sll1077	/speB	801102	801851	promoter	2.657	Agmatinase	801420
−N	ncr0210	NsiR7	450202	450501	promoter	2.633	ncRNA	450322
−N, NH4	sll1577	cpcB	727802	727951	promoter	2.614	phycocyanin beta subunit	727919
−N, NH4	sll0783		3071502	3072751	promoter	2.509	unknown protein	3072049
−N	slr1164	nrdA	1927974	1928364	promoter	2.438	ribonucleotide reductase subunit alpha	1928191
NH4	slr1852		1190528	1191867	intragenic	2.225	unknown protein
−N, NH4	slr1456	pilA4	3302952	3303101	intragenic	2.151	type 4 pilin-like protein, or general secretion pathway protein G
−N	sll1831	glcF	625122	625461	Intragenic TSS	2.146	glycolate oxidase subunit (Fe-S) protein	625336
−N	slr0898	nirA	2768902	2769151	promoter	1.861	ferredoxin–nitrite reductase	2769102
−N, NH4	slr1841		958152	958451	promoter	1.364	probable porin	958082
NH4	sll1356	glgP	1073751	1074000	intragenic	1.347	glycogen phosphorylase
−N	sll1762		1233657	1233801	promoter	1.331	putative polar amino acid transport system substrate-binding protein	1233755c
−N	sll1973		1576602	1576801	promoter	1.304	hypothetical protein	1576662
−N	slr0585	argG	3533652	3534001	promoter	1.276	argininosuccinate synthetase	3533815
−N	slr1735	bgtA	1317495	1317738	promoter	1.275	ATP-binding subunit of the ABC-type Bgt permease	1317670
−N	sll1968	pmgA	909452	909601	promoter	1.214	photomixotrophic growth related protein	909579c
−N	slr1770		549805	550138	intragenic	1.006	hypothetical protein
−N	sll1502	gltB	482302	482501	promoter	0.955	NADH-dependent glutamate synthase large subunit	482370c
−N	slr1624		2034892	2035528	promoter	0.909	hypothetical protein	2035343c
−N	sll0680	pstS	2641802	2641951	promoter	0.877	phosphate-binding periplasmic protein precursor (PBP)	2641853
−N	slr2057	apqZ	1412702	1412951	promoter	0.850	water channel protein	1412833
−N	slr1028		647502	647651	intragenic	0.705	unknown protein
−N	sll0254	crtL	1507002	1507101	intragenic	0.705	probable phytoene dehydrogenase Rieske iron-sulfur component
NH4	sll1665		268551	268700	promotor	0.472	unknown protein	3304862
−N, NH4	slr0442		2081302	2081501	intragenic	0.471	unknown protein
−N	sll1070	tktA	819816	820281	promoter	−0.399	transketolase	820067
−N	sll1049		90302	90801	promoter	−0.399	hypothetical protein	90594c
−N	sll0616	secA	2648652	2648801	intragenic	−0.671	preprotein translocase SecA subunit
−N	sll1536	moeB	2041452	2041601	promoter	−0.707	molybdopterin biosynthesis MoeB protein	2041560
NH4	sll0413		2544264	2545469	intragenic	−0.802	hypothetical protein	2545087
−N	sll1958	hisC	1411852	1412351	promoter	−0.831	histidinol phosphate aminotransferase	1412183
−N	sll1273		1109887	1110178	intragenic	−0.891	unknown protein
−N, NH4	slr0616	mvrA	2948302	2948751	promoter	−1.035	methyl viologen stress protein	2948503
−N	ncr1071		2215852	2216151	promoter	−1.127	ncRNA	2215968
−N	slr1146		811052	811551	promoter	−1.175	hypothetical protein	811318
NH4	sll0142		2197197	2198566	intragenic	−1.338	probable cation efflux system protein
−N	sll1563		1964452	1964601	intragenic	−1.358	unknown protein
−N	sll0927	metX	340352	340501	intragenic	−1.474	S-adenosylmethionine synthetase
−N	ssr1562		3235202	3235651	promoter	−1.477	hypothetical protein	3235428
−N	slr1705	aspA	732302	732451	promoter	−1.538	aspartoacylase	732376
−N	slr0426	folE	2717952	2718051	promoter	−1.632	GTP cyclohydrolase I	2718008
NH4	slr0320		2267622	2269004	intragenic	−1.777	hypothetical protein
−N	ncl0880	Syr6	1816295	1816865	promoter	−1.859	ncRNA	1816624c
−N	ncl0530/sll2012	/sigD	1255752	1255940	intragenic	−2.248	group2 RNA polymerase sigma factor SigD	1204108
−N, NH4	slr1254	pds	1397352	1397951	promoter	−2.311	phytoene dehydrogenase	1397780
−N	sll1515	gifB	458552	459201	promoter	−2.324	GS inactivating factor IF17	458961
−N, NH4	ssr1038		2949752	2950201	promoter	−2.597	unknown protein	2949923
−N	ssr0692		2695446	2695626	promoter	−2.738	hypothetical protein	2695575
−N, NH4	slr0082	rimO	2885852	2886301	promoter	−2.975	hypothetical protein	2886040
−N, NH4	ssl1911	gifA	631902	632351	promoter	−4.204	GS inactivating factor IF7	632192
−N, NH4	ncl0250/sll1291	/rre12	692659	693007	promoter	1.462	two-component response regulator PatA subfamily	692816
−N	Ncl0930/slr1681		1969237	1969498	promoter	2.618	unknown protein	1969319
−N	sll0405		2552752	2552901	intragenic	1.100	unknown protein
−N	sll1802-as		840302	840791	promoter	−1.405	Antisense RNA	840574
−N	sll1028	ccmK2	219070	219290	intragenic	−0.626	carbon dioxide concentrating mechanism protein

List of genes differentially expressed after nitrogen starvation that contain an NtcA binding site. The condition in which NtcA peaks were higher is highlighted in bold.

Figure 6.

NtcA regulatory network and its transcriptional dynamics for the early stages of nitrogen depletion. Target genes are represented by nodes that are colored based on gene expression after 4 h of nitrogen depletion: yellow for upregulated genes, and blue for downregulated genes. Node sizes correspond to the magnitudes of expression changes as a log₂ expression ratio (−N/NH₄). Genes are grouped according to the functional categories obtained from CyanoBase. PPNN: purines, pyrimidines, nucleosides and nucleotides.

Sequence analysis of NtcA-bound promoters

The NtcA consensus binding sequence GTAN₈TAC was first described in the NtcA-regulated promoters of Synechococcus (53). Promoters activated by NtcA tend to contain a consensus sequence, centered close to position −41 with respect to the TSS, although actuating NtcA binding in positions further upstream has been reported in some cases (53). Conversely, NtcA-mediated repression is thought to be caused by NtcA binding at positions incompatible with correct assembly and positioning of the RNA polymerase. This model is supported by a promoter analysis of the NtcA regulon for Synechocystis. Promoters of the 10 most-induced genes of the NtcA regulon showed a highly conserved GTAN₈TAC motif, centered close to position −41.5. Notably, GTA at positions 1−3 was strictly conserved, and only little variation was observed for TAC at positions 12−14 (Figure 7A). In the case of the top ten repressed genes, the potential NtcA binding sites displayed greater derivation from the consensus sequence, and were widely distributed along the promoter (Figure 7B). These patterns were also found, when we examined NtcA binding positions for the entire regulon. While 71% of binding sites of induced genes were concentrated at positions −41 and −44, binding for repressed genes ranged from −69 to +83, with respect to the TSS, with slight accumulation around −30 (Figure 7C). This finding confirms previous observations that the distance between the NtcA binding site and the TSS determines the regulatory mode of NtcA (53).

Figure 7.

Sequence analysis of NtcA binding sites. (A and B) Promoter regions having NtcA binding sites for ten most strongly induced and repressed genes, respectively. Nucleotides confirming the putative NtcA consensus-binding sequence are indicated in blue, potential −10 promotor elements in red and transcriptional start sites (TSS) in orange. An alternative NtcA binding site for the gifA promoter at position −45 is underlined in red. (C) Distribution of positions of NtcA binding sites identified for genes of the NtcA regulon. The location of NtcA binding was determined by matching the NtcA consensus sequence. The relative position was defined as the distance of the seventh nucleotide of a putative NtcA binding site to the TSS. Blue and red bars indicate the frequency of these positions for induced and repressed genes, respectively. (D) NtcA binding motif defined by target genes of the NtcA regulon. (E) Binding motif found using all the putative NtcA binding sites identified by ChIP-seq analysis. (Representation by WebLogo 3.0) (54).

In addition to examining the positioning of the established consensus binding sequence, studying the NtcA regulon provides the opportunity to refine the NtcA binding motif for Synechocystis. Using the WebLogo application (54), we obtained an approximately palindromic motif of length 14 nt, with a high prevalence of G, T and C at positions 1, 2 and 14; and medium prevalence of A, T and A at positions 3, 12 and 13 (Figure 7D). In most of the remaining positions (4–11), a weak tendency towards A or T was noticed. Subsequently, we extended this analysis to include all NtcA binding regions, determined in our ChIP-seq analysis. For this enlarged set, the information content of the motif deteriorated, although it still resembled the consensus sequence GTAN₈TAC (Figure 7E). Interestingly, the prevalence of the flanking nucleotides (G at 1, C at 14) increased, demonstrating their importance for NtcA dimer binding, even when other nucleotides were less conserved.

Noncoding RNAs regulated by NtcA

Several NtcA binding peaks were assigned to promoter regions of ncRNAs and asRNAs. In total, 18 ncRNAs and 5 asRNAs with assigned NtcA binding were identified in our ChIP-seq analysis (Supplementary Tables S9 and 10). Eight of these ncRNAs were also differentially regulated; thus, we added these to the defined NtcA regulon. These included NsiR4, which has been previously reported to be involved in nitrogen assimilation control by targeting IF7 (41). Notably, three of these NtcA targeted ncRNAs were transcribed from their own exclusive transcriptional units (ncr1071, ncl0880, ncr0210; Figure 8A). Specifically, ncr1071 codes for a predicted 1057-nt long ncRNA, while ncl0880 (also called Syr6) and ncr0210 (also called NsiR7) have been reported to belong to the 33 most abundant ncRNAs in Synechocystis (5,47). Since these ncRNAs were transcribed from their own transcriptional units, they are unlikely to be by-products of the expression of other genes; thus, we decided to analyze them further.

Figure 8.

Regulation of non-coding RNAs by NtcA. (A) Visualization of NtcA-binding peaks assigned to ncr1071, Syr6 and NsiR7. Their ChIP-seq and RNA-seq density profiles are shown in blue for the NH₄ condition and in red for the −N conditions. (B) Electrophoretic mobility shift assay verification of NtcA binding sites. Promoter sequences of ncr1071, syr6 and nsiR7 were PCR amplified (primers listed in Supplementary Table S5) before being mixed with 0.1, 0.4, 0.6 μM of NtcA. An aliquot of 0.6 mM of 2-oxoglutarate (2-OG) was added, when indicated. (C) Northern blot analysis of expression of Ncr1071, Syr6 and NsiR7. Total RNA was isolated from WT cells transferred from BG11C supplemented with ammonium to nitrogen-free BG11C medium for 24 h. The filters were hybridized with Ncr1071, Syr6 and NsiR7 probes and subsequently stripped and rehybridized with rnpB probe as a control.

The RNA-seq data showed the accumulation of NsiR7 transcripts after 4 h of nitrogen depletion, having an NtcA binding site upstream at position −43 from the TSS, consistent with activation by NtcA (Figure 8A and Supplementary Table S10). In contrast, both ncr1071 and syr6 were downregulated and their NtcA binding sites were found at positions +3 and +16, respectively (Figure 8A and Supplementary Table S9). To experimentally validate the NtcA binding peaks detected by ChIP-Seq, we performed EMSA experiments, which clearly showed NtcA binding to promoter sequences of these selected ncRNAs (Figure 8B). We also carried out northern blot experiments to examine the expression levels of these ncRNAs over a prolonged period of nitrogen starvation. For this purpose, Synechocystis cultures grown in NH₄ were shifted to nitrogen-depleted media, and samples were collected after 4, 12 and 24 h. Both ncr1071 and syr6 were strongly repressed after 4 h of nitrogen starvation; although in the case of syr6, repression was not complete and residual transcription was still observed (Figure 8C). nsiR7 showed a weak induction that was detected by northern blot, after only 12 h.

To predict potential targets of these three ncRNAs, we used two RNA target predictions programs: CopraRNA (55) and IntaRNA (33). CopraRNA has better accuracy than IntaRNA since it uses phylogenetic conservation in its scoring system (56), but this implies that presence of homologous sRNA sequence from distinct organisms is necessary. For both ncr1071 and syr6 homologous where found in other cyanobacteria but not for nsiR7 that is only present in Synechocystis. Thus CopraRNA was applied for Ncr1071 and Syr6, while IntaRNA was only used for NsiR7. Assuming repression as the dominant mode of regulation by ncRNAs, we expected that genes under the control of repressed ncRNAs after nitrogen depletion (Ncr1071 and Syr6) would show increased expression in our RNA-seq data; while those under the control of induced NsiR7 would show decreased expression. In the case of Ncr1071, four predicted targets were significantly induced: sll1451 encoding the nitrate transport protein NrtB, slr2136 encoding the GcpE protein homolog involved in the terpenoid biosynthesis and slr0151 and sll1219 encoding the hypothetical proteins Slr0151 and Sll1219, respectively (Supplementary Table S11). Alternatively, it may act as an asRNA to sll1864 coding for a chloride channel protein. In the case of Syr6, also four of the predicted targets displayed increased expression: ssl2598 (psbH) encoding the photosystem II reaction center protein H, slr0079 encoding the general secretion pathway protein E (GspE), sll1533 encoding the twitching motility protein PilT and sll0376 encoding the hypothetical protein Sll0376 (Supplementary Table S11). In the case of NsiR7, the only predicted target that was repressed after nitrogen step-down was the transcription factor nusA (slr0743).

Role of PipX in nitrogen regulation in Synechocystis

For a number of NtcA binding sites, the normalized peak height was preserved during nitrogen starvation, suggesting a similar NtcA binding affinity under both conditions. Surprisingly, the clearest example was observed in the promoter region of IF7 coding gene gifA. Despite significant expression change in gifA (−4.2 log FC; Supplementary Tables S8 and 12), following transition to a nitrogen-free medium, its NtcA peak height remained almost identical, with less 1% differential fold enrichment (DFE) between both conditions (Figure 9A). This contrasts with gifB encoding IF17 (Figure 9A), for which its downregulation (−2.3 log fold change, FC; Supplementary Table S8) was correlated with an NtcA binding peak present only under nitrogen starvation. Given that promoters of both GS inactivation factors contain NtcA binding sites at positions considered repressive for transcription (9), the persistence of NtcA binding under nitrogen-replete conditions, when gifA is highly induced, is remarkable. Intriguingly, we found several other NtcA targets (some of them well established) with unexpectedly small changes in their NtcA binding, considering their large expression changes. These included rre37 (23% DFE, 2.7 log FC), urtA (24% DFE, 5.3 log FC; Figure 9A) and urtB (<38% DFE, 5.1 log FC). Collectively, eight genes having more than a two log FC in expression showed less than 40% DFE (Supplementary Table S12).

Figure 9.

The role of PipX in the NtcA-regulated genes. (A) Read coverage of NtcA-binding peaks assigned to gifA, gifB and urtA. Their ChIP-seq and RNA-seq density profiles are shown in blue for NH₄ and in red for −N conditions. (B) Growth of Synechocystis WT and ΔpipX strains after nitrogen step-down. Cells of WT and ΔpipX strains cultivated to their midlog growth phase in BG11C medium supplemented with NH₄ were transferred to nitrogen-free BG11C medium.Their growth was measured at OD₇₅₀ for 48 h. (C) Images of cyanobacterial cell suspensions of Synechocystis WT and ΔpipX obtained during growth curve analysis (panel B) at indicated times. (D) Northern blot analysis of expression of gifA, gifB, glnB, glnN, urtA and ntcA in response to nitrogen starvation. Total RNA was isolated from WT cells transferred from BG11C supplemented with ammonium to nitrogen-free BG11C medium for 24 h. The filters were hybridized with gifA, gifB, glnB, glnN, urtA and ntcA probes and subsequently stripped and rehybridized with rnpB probe as a control. (E) Western blot analysis of IF7, IF17, P_II, GSIII and NtcA in response to nitrogen starvation. Cells were grown in BG11C supplemented with ammonium to midlog growth phase and then transferred to nitrogen-free BG11C medium and cultivated for 24 h. Samples of 10 μg of total proteins from soluble extracts were separated by 15% SDS/PAGE and analyzed by western blots to detect IF7, IF17, P_II, GSIII, NtcA and TrxA.

It has previously been demonstrated that high cytoplasmic concentrations of NtcA alone are not sufficient to promote activation of selected target genes (57). In fact, activation of several NtcA-dependent genes, including the amt1 or nir operon, has been directly stimulated by 2-OG in Synechococcus elongatus PCC 7942 (58). Later studies revealed that 2-OG enhances complex formation between NtcA and PipX, which is crucial for activation of NtcA target genes in S. elongatus PCC 7942 and Anabaena sp. PCC 7120 under nitrogen-depleted conditions (49,59). Whether this is the case for Synechocystis remains unclear, as the role of PipX in nitrogen control in this cyanobacterial model has not been fully established.

To investigate the role of PipX in the regulation of the NtcA-controlled genes in Synechocystis, especially those having promoters with similar affinity to NtcA under both conditions (NH₄ and −N), a pipX-deficient mutant (ΔpipX) was generated (see ‘Materials and Methods’ section and Supplementary Figure S2). Cells of Synechocystis WT and ΔpipX cultivated in BG11C medium supplemented with ammonium were shifted to a nitrogen-free BG11C medium and cultivated for an additional 48 h. Although the growth kinetics for both strains did not show significant differences during this nitrogen-starvation treatment, chlorosis of the cells in the ΔpipX strain was slightly delayed, compared with WT (Figure 9B and C). A similar observation was made for a pipX-deficient mutant of S. elongatus PCC 7942 (10). Subsequently, we analyzed the transcript and protein accumulation of several genes involved in nitrogen assimilation, including IF7 and IF17, GSIII, P_II and the urea transport protein UrtA. Consistent with a role of PipX as an NtcA enhancer, both transcriptional activation and repression of the selected NtcA targets were delayed in the ΔpipX strain. This delay was also reflected in their corresponding protein levels (Figure 9D and E). Furthermore, transcriptional induction of glnN, glnB and urtA was weaker in the ΔpipX strain, suggesting that PipX enables faster and more efficient adaptation to nitrogen starvation. Such a function might be especially important, given that transcription of ntcA remained unaltered in both strains during our treatment with only a small increase in protein levels after 24 h, in agreement with previous report (3). Remarkably, we did not find any differences in the transcriptional expression of gifA between both strains under nitrogen-replete condition, indicating that PipX is not acting as a coactivator of NtcA for gifA transcription. A possible explanation for the persistent NtcA binding peak within the promoter of gifA in NH₄-replete media might be the presence of a degenerate NtcA consensus sequence GTAAATTTTTGTTT at position −45. At this position, NtcA could bind and enhance transcription of gifA (Figure 7). Stimulated by 2-OG or other factors, NtcA could slide to the binding site at position −33, acting as a repressor, when nitrogen is limited.

DISCUSSION

To understand how cyanobacteria respond to varying conditions in the environment, it is important to have comprehensive models of their underlying transcriptional regulation. Here, we present the first genome-wide analysis of a transcription factor regulon for Synechocystis, based on both ChIP-seq and RNA-seq techniques. Combining these two high-throughput approaches permitted us not only to identify the regulon of NtcA, but also to capture the activity of this key transcriptional regulator of nitrogen. We were able to identify 48 and 121 NtcA-bound regions in the Synechocystis genome under nitrogen-eplete conditions and after 4 h of nitrogen step-down, respectively. In parallel, we obtained transcriptional profiles of Synechocystis under the same experimental conditions. This enabled us to classify target genes into three categories: activated, repressed or unaffected by NtcA. Activated and repressed genes were used to define the NtcA regulon during the early phase of nitrogen starvation. They included various genes involved in nitrogen and carbon metabolism or photosynthesis, as well as several ncRNAs. Our ChIP-seq detected 25 previously known or predicted NtcA targets in Synechocystis (Supplementary Table S13), demonstrating the sensitivity of our approach. Importantly, we identified 67 new NtcA targets within the defined NtcA regulon, expanding the scope of the NtcA regulon during the early phase of nitrogen acclimation in Synechocystis. The higher proportion of intragenic NtcA binding sites under nitrogen-replete conditions indicates a distinct role for NtcA under this condition. Finally, expression analyses of functionally relevant NtcA targets in a pipX mutant strain revealed the importance of PipX for efficient transcriptional regulation by NtcA in Synechocystis during nitrogen starvation.

Transcriptional adaptation to early stages of nitrogen starvation

The RNA-seq analysis revealed that transcription of genes involved in transcription and translation, biosynthetic process and protein synthesis were downregulated, indicating a reduction of overall protein synthesis, as previously reported for nitrogen deprivation (3–4,60). Major changes in transcript levels were observed for genes involved in nitrogen uptake and metabolism (Table 1 and Figure 2), including many NtcA targets: glnA, glnB, amt1, urtA, gifA and gifB. However, genes coding for both NtcA and the nitrogen coactivator PipX were not affected (Figure 9). Absence of induction of NtcA under nitrogen starvation condition has been previously observed (3); suggesting that post-transcriptional regulation of NtcA plays a more important role than the amount of NtcA per se. Surprisingly, expression of the PSI- and PBS-related genes was induced. This finding is consistent with a previously reported transcriptome profiling after 6 h of nitrogen starvation by Krasikov et al. (3), in which PSI- and PBS-related genes were transiently activated, while cells proliferated at a normal growth rate. In contrast, Osanai et al. observed in their microarray study of nitrogen starvation that photosynthetic genes were repressed (4). The discrepancies between these studies might be related to differences in culture conditions prior or during nitrogen starvation or in a strain-specific genetic background. For example, illumination intensity used by Osanai (70 μmol photons⁻² s-¹) was 55 and 40% greater than used in our experiment (45 μmol photons⁻² s⁻¹) or by Krasikov et al. (50 μmol photons⁻² s⁻¹), respectively. In any case, the observed transient upregulation of PBS- and PSI-related genes may indicate increased cyclic electron transport around PSI as early stress response. This hypothesis is reinforced by the induction of ndhD1 (slr0331), which codes for a component of the NADH dehydrogenase complex involved in PSI cyclic electron flow, and some subunits of ATPase synthase (atpI, atpC and atpH; Supplementary Table S6). Increased cyclic electron flow could enhance ATP synthesis to provide the necessary energy required for nitrogen assimilation pathways. However, this phase is only transitional to a more severe response to nitrogen starvation, in which growth decreases and degradation of the PBS leads to a yellow appearance of cyanobacterial cultures. This acclimation process, known as chlorosis or bleaching, requires the expression of the cotranscribed nblA genes (nblA1 and nblA2) (5). In fact, expression of nblA2 (ssl0453) after 4 h of nitrogen starvation was observed, suggesting that a transcriptional response towards PBS degradation and long-term acclimation had already started.

Nitrogen depletion in cyanobacteria leads to downregulation of genes related to carbon fixation and induction of sugar catabolic genes (3–4,60,48). Here, we observed the induction of genes involved in the oxidative pentose phosphate (zwf, talB and gnd), as well as catabolism (glgX and glgP) and anabolism (glgC) of glycogen (Figure 3). Expression of genes for glycogen degradation, which are likely not to be active during nitrogen starvation, is in agreement with an anticipatory state of chlorotic Synechocystis cells described recently by Klotz et al. (61). Cells anticipate future recovery from nitrogen starvation by preparing proteins necessary for fast resuscitation without having to synthesize the corresponding enzymes de novo.

Strikingly, 33 genes involved in regulatory functions responded to nitrogen step-down, suggesting that acclimation to nitrogen depletion is a process that requires extensive transcriptional reprogramming. This is corroborated by our observation that several genes coding for two-component systems were affected by nitrogen deprivation. Two-component systems enable cells to respond to both environmental and intracellular changes. Six histidine kinases and six response regulators were upregulated, including rre37, which is involved in the activation of sugar catabolism under nitrogen starvation (62,63). The complexity of the nitrogen starvation response is highlighted by the large number of genes involved in heavy metal and oxidative stress, which altered their expression in response to nitrogen depletion. These included genes encoding the copper sensing system CopRS (sll0789, sll0790 and their respective copies in the plasmid pSYSX: slr6040 and slr6041), the ferric uptake regulator Fur (sll0567), the peroxide stress response transcriptional regulator PerR (slr1738), and the LexA repressor (sll1636). Furthermore, many genes coding for metal importer systems were downregulated, such as the ATPases for copper (ctaA and pacS), cobalt (coaT) and zinc (ziaA), as well as iron import proteins FutA1 (futA1) and FutA2 (futA2). Such widespread adjustment of metal uptake could be the consequence of diminished protein synthesis following nitrogen deprivation. Given that up to a third of the total microbe proteome contains metal cofactors (64), reduction in metal uptake could be an important response to avoid build-up of excess free metals in the cytosol that would lead to detrimental reactions.

The asymmetric distribution of the NtcA binding peaks in NH₄ and −N media points to condition-dependent modus operandi of NtcA

Comparative ChIP-seq analysis enabled us to characterize changes in the binding behavior of NtcA. It revealed that NtcA binds to 141 DNA regions (121 in the chromosome and 20 in the plasmids) after 4 h of nitrogen step-down, but also to a large number of regions (51 with 48 in the chromosome and 3 in the plasmids) in nitrogen-replete medium. Its unexpected binding in vivo under nitrogen-replete conditions supports in vitro measurements showing that 2-OG is not absolutely required for NtcA binding to DNA (12). The 169 binding sites identified and assigned to 157 genes, included the vast majority of currently known NtcA target genes for Synechocystis, indicating the high sensitivity of our ChIP-Seq experiment. Further support of its reliability was provided by validation of NtcA binding to target regions with different features (intragenic or promoter binding region, different peak enrichment, etc.; Supplementary Figure S5).

Under nitrogen starvation conditions, most of the binding sites were located upstream of gene coding regions (73% of peaks; Supplementary Tables S9 and 10), consistent with preferential binding of NtcA to promoter regions. Unexpectedly, this preference changed in ammonium-rich medium, with 54% of binding peaks located at intragenic positions. This highly significant change in the distribution of binding loci (P = 0.00116, Fisher’s exact test) could indicate a condition-dependent functional mode for NtcA in Synechocystis. Under nitrogen depletion, many genes involved in nitrogen metabolism and other cellular functions were induced or repressed by NtcA, supporting its established role as the master regulator of nitrogen control in cyanobacteria. In contrast, the role of NtcA in the presence of ammonium has remained elusive. The lower number of peaks and the high proportion of intragenic binding sites for NH₄ compared to −N treatment could simply indicate that NtcA rests in an inactive state under nitrogen-replete conditions, as previously suggested (57). However, other potential scenarios could exist, given that extensive binding of transcription factors to intragenic regions has been reported in recent years, based on genome-wide profiling. For example, substantial intragenic binding has been detected for OmpR in Salmonella enterica, GlxR in Corynebacterium glutamicum and RutR in Escherichia coli (65–67). Remarkably, a high proportion of intragenic binding sites have also been observed in a previous ChIP-seq analysis for NtcA in Anabaena sp. PCC 7120 (14). Hence, an examination of the functional relevance of intergenic binding is clearly vindicated. Closer inspection of intragenic binding sites detected by our ChIP-seq experiment revealed the existence of the canonical NtcA consensus sequence, located upstream of internal TSS in some cases (i.e. ssll0377, sll0735, slr1065, slr1864, glcF; Supplementary Tables S9 and 10). In other cases, our transcriptomic data suggest that NtcA could act as a repressor, blocking transcription elongation (Figure 5) in a similar manner to that reported for CodY in Bacillus subtilis (68). Candidates for this type of transcriptional repression by NtcA are: slr1852, glgP and pilA4 under nitrogen-replete conditions; or slr1028, slr0442, secA, sll1273, mrvA and metX under nitrogen starvation (Figure 5 and Table 2). Such a regulatory mode, however, cannot be generalized, as we found other cases, such as sll0142 in NH₄ or slr0909 in −N treatments, which showed increased expression and stronger intragenic NtcA binding—an observation that eludes an obvious explanation. In summary, examination of intragenic binding sites supports a model in which NtcA is not simply inactive under nitrogen-replete conditions, but in which it remains functionally important, albeit with a different modus operandi. Such a model is consistent with previous ChIP–chip and ChIP-seq studies of E. coli and Anabaena suggesting that the CRP transcriptional regulator family could work not only as a canonical transcriptional regulator, but also as a chromosome-shaping protein by binding to multiple low affinity sites (14,69).

NtcA targets in nitrogen control and cellular metabolic processes

Parallel profiling with ChIP-seq and RNA-seq techniques led to the definition of a direct target regulon for NtcA, having unprecedented resolution. In total, 79 genes were identified to have NtcA binding peaks with significantly altered expression after 4 h of nitrogen step-down (Figure 5). Analysis of their respective gene functions revealed that NtcA plays a role in the coordination of cellular processes beyond nitrogen metabolism. Indeed, various other metabolic processes (biosynthesis of cofactors, cellular processes, carbon metabolism, energy metabolism, central intermediary metabolism, photosynthesis and respiration) were dominant among these NtcA targets (Figure 6). Additionally, a small proportion of the NtcA regulon was associated with other biological functions, such as cell envelope, translation and transcription, or regulatory functions (Figure 6).

A considerable number of NtcA targets lack current annotation. In particular, 27 genes correspond to hypothetical and unknown proteins. They are prime candidates for further experimental investigation to enhance our understanding of nitrogen control and metabolism, and to extend the functionality of the NtcA regulon. For instance, genes sll0327 and ssr0692 showed levels of induction and repression, similar to those involved in nitrogen metabolism, suggesting related roles (Figure 6). Likewise, sll0327 codes for a small protein containing 139 amino acids, and is a closed homolog to the amino acid transport system permease component LivM of Microcystis aeruginosa (58.1% identity; 72.1% similarity). Meanwhile, ssr0692 was reported to be highly induced under conditions of low CO₂ i.e. conditions which resemble a relative high abundance of nitrogen (70). It may be involved in the accumulation of the NDH-1 complex under low carbon/high nitrogen, given that Ssr0692 interacts with NdhH, according to yeast-two-hybrid data (71). Furthermore, the expression of both sll0327 and ssr0692 is strongly correlated with known NtcA target genes across a large set of conditions compiled in CyanoEXpress—a platform for a gene expression meta-analysis in Synechocystis (72). This is a remarkable finding, since CyanoEXpress does not include expression data for nitrogen starvation. Thus, this correlation is solely based on differential expression of NtcA target genes under other conditions. Strikingly, we found that sll0327 together with bgtB, glnA and urtB (all members of the NtcA regulon) are highly upregulated under iron limitation and downregulated under carbon limitation or oxidative stress (Supplementary Figure S6A). Conversely, ssr0692 together with gifA and gifB are downregulated under iron limitation and upregulated under carbon limitation or oxidative stress (Supplementary Figure S6B). This suggests that sll0327 and ssr0692 have a common regulation, along with other bona fide members of the NtcA regulon. In addition, the strong correlation of NtcA targets observed in gene expression meta-analysis points to potential regulatory activity of NtcA under a wider range of environmental perturbations.

Finally, asRNAs and ncRNAs were the third most abundant class in the NtcA expression network. Although differential expression of asRNAs and ncRNAs during acclimation to different nitrogen stresses in cyanobacteria has been documented in several recent studies (15,41,61), only one ncRNA, NsiR4, has been experimentally validated to be under the control of NtcA to date (41). Our results greatly expand the repertoire of NtcA regulated ncRNAs. Eight ncRNAs were included in the NtcA regulon (Figure 6) and we have experimentally validated three of them: Ncr1071, NsiR7, Syr6 (Figure 8). Interestingly, NsiR7 is also induced under conditions of carbon limitation and was recently suggested as an NdhR-regulated sRNA (73,74). Thus, NsiR7 might play an important role in the integration of signals integration related to the intracellular carbon and nitrogen status.

NtcA binding defines poised states of transcriptional regulation in Synechocystis, which are modulated by PipX and additional factors

Surprisingly, almost half of the genes having assigned NtcA binding were not transcriptionally affected after 4 h of nitrogen step-down. However, comparison with a longer time series study of nitrogen starvation (3) revealed that many of the initially unaffected genes were activated or repressed at later time points (Supplementary Figure S7). Examples are the induction of genes coding for the urease gamma subunit (ureA), the serine/threonine kinase (spkD) or repression of the PSII oxygen evolving complex protein (psbP2) after 12 h of nitrogen depletion. Even longer delays in transcriptional response were detected for the repression of pipX (24 h) and the induction of the ferrichrome-iron receptor coding gene fhuA (96 h). Importantly, late transcriptional responses are not simply the consequence of delayed NtcA accumulation, given that both NtcA transcript and protein levels remained unaltered after nitrogen step-down. Rather, these findings indicate that NtcA binding in vivo may not be sufficient in certain cases to evoke immediate full transcriptional control, but defines a poised regulatory state, which correlates with later changes in expression. Such poised regulation suggests the need for additional factors, as previously proposed (14,57). Further support for this model was gained through comparison of the ChIP-seq data obtained under NH₄ and −N treatments. We identified 14 NtcA targets having assigned NtcA binding peaks that did not show significant changes, but were significantly differentially expressed after nitrogen depletion. Our examination of a pipX mutant supports the role of PipX as an NtcA modulating factor in Synechocystis. Transcriptional regulation of NtcA-controlled genes and their corresponding proteins levels were delayed in the pipX mutant strain compared to the WT strain, consistent with a role of PipX as an NtcA enhancer, as reported in other cyanobacteria (10,49,59). This mechanism could explain why several genes showed strong transcriptional change after 4 h of nitrogen starvation, without an equivalent variation in the effective affinity of NtcA to their promoters (Supplementary Table S12). This indicates that interaction of NtcA with other factors is essential to activate/repress target genes.

It is tempting to speculate on the physiological function of a poised regulatory state for NtcA. One advantage could be that it prepares the transcriptional machinery for a more extensive response, but without a full commitment. It would provide a temporal buffer, which is especially advantageous under conditions of fluctuating nitrogen availability, where premature commitment to unmitigated chlorosis might prove to be adverse. In this way, some NtcA targets, for which no differential expression was recorded after 4 h of nitrogen depletion, might constitute a ‘latent NtcA regulon’; in contrast to the NtcA regulon defined herein, which is activated during early acclimation.

In conclusion, this study provides the first genome-wide captures of the dynamics of in vivo NtcA binding events and the regulatory network that they define. Application of parallel differential profiling enabled us to distinguish direct from indirect regulation, extending the scope of the NtcA regulon. Finally, indications of a poised state for NtcA and of regulatory activity in a wide range of conditions beyond nitrogen limitation offer exciting avenues for further investigations.

DATA AVAILABILITY

RNA-seq and ChIP-seq datasets are available at GEO accession (GSE97291).

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

National Portuguese Funding through FCT (Fundaçao para a Ciência e a Tecnologia) projects [PTDC/BIA-MIC/4418/2012, IF/00881/2013, UID/BIM/04773/2013–CBMR, UID/Multi/04326/2013–CCMAR]; Ministerio de Economia y Competitividad (MINECO) Grants [BFU2013–41712, BIO2016–75634 to F.J.F.]; Junta de Andalucía-European Regional Funds (FEDER) [Group BIO-284, P12-BIO-1119]. Funding for open access charge: School of Biomedical & Healthcare Sciences, Plymouth University Peninsula Schools of Medicine and Dentistry.

Conflict of interest statement. None declared.