Antisense RNA controls the degradation of phycobilisomes in heterocyst-forming cyanobacteria by a transcriptional interference mechanism

Isidro Álvarez-Escribano; Manuel Brenes-Álvarez; Jens Georg; Agustín Vioque; Alicia M Muro-Pastor

doi:10.1093/nar/gkag575

. 2026 Jun 22;54(11):gkag575. doi: 10.1093/nar/gkag575

Antisense RNA controls the degradation of phycobilisomes in heterocyst-forming cyanobacteria by a transcriptional interference mechanism

Isidro Álvarez-Escribano ^1,^b, Manuel Brenes-Álvarez ^2,^b, Jens Georg ³, Agustín Vioque ⁴, Alicia M Muro-Pastor ^5,^✉

PMCID: PMC13283749 PMID: 42324081

Abstract

Transcriptional interference is rarely documented in bacteria. In cyanobacteria, phycobilisomes are the major light-harvesting antennae, and their degradation under non-optimal conditions follows a tightly regulated genetic program leading to the production of NblA, a widely conserved proteolytic adapter. NblA production is regulated at both the transcriptional and post-transcriptional levels. Here, we uncover an additional regulatory layer mediated by an antisense RNA conserved in heterocyst-forming cyanobacteria. In Nostoc sp. PCC 7120, transcription of the abundant antisense RNA (as_nblA) limits nblA mRNA accumulation. A strain unable to transcribe as_nblA produces an excess of NblA, ultimately so harmful that suppressor mutations of nblA expression accumulate rapidly, underscoring the essential role of as_nblA. Rifampicin time-series experiments and the inability of as_nblA to regulate nblA expression in trans support transcriptional interference as the primary regulatory mechanism of as_nblA. Mathematical modeling of nblA expression, supported by biological data, shows that as_nblA plays a pivotal role in preventing leaky nblA expression under non-inducing conditions. Our work highlights the importance of antisense RNA-mediated regulation, particularly transcriptional interference, in establishing thresholds that prevent spurious expression of genes encoding critical cellular functions.

Graphical Abstract

For image description, please refer to the figure legend and surrounding text.

Introduction

The NblA protein plays a crucial role in the environmentally regulated degradation of the phycobilisomes (PBS), which are essential light-harvesting complexes that transfer light energy to the photosystems in cyanobacteria. Degradation of PBS (also referred to as bleaching) is particularly significant under nutrient-limiting conditions, such as nitrogen starvation. NblA was first described in the unicellular cyanobacterium Synechococcus elongatus PCC 7942 as a small polypeptide that triggered the degradation of phycobiliproteins (PBP) under nutritional limitation, with the expression of the nblA gene being induced upon nitrogen, sulfur, or phosphate removal [1]. Since PBS account for a large amount of the total protein content in the cells, and therefore a large amount of the total nitrogen content, bleaching under those circumstances is an adaptive strategy that allows the cyanobacteria to recycle nutrients under stress conditions. Induction of nblA transcription and PBS degradation have also been observed upon exposure to high light in S. elongatus PCC 7942 [2], preventing excessive energy input to the photosynthetic apparatus. NblA functions as a proteolysis adapter protein that interacts with PBP, facilitating their degradation by the Clp protease complex [3–6]. This interaction is critical, because it designates PBP for degradation, thereby allowing the cyanobacteria to recycle nutrients and adjust light absorbance to their metabolic capacity. Recently, nblA has emerged as a gene with a broad impact at the ecological level because the presence of active NblA proteins encoded in cyanophages negatively affects oceanic cyanobacterial photosynthesis [7].

In contrast to unicellular non-diazotrophic cyanobacteria, which usually nearly completely degrade PBP within 1–2 days of nitrogen deprivation, the PBS in vegetative cells of heterocyst-forming cyanobacteria, such as Nostoc sp. PCC 7120 (also known as Anabaena sp. PCC 7120, from now on, Nostoc), are only transiently degraded until the nitrogenase located in specialized cells, called heterocysts, becomes active and combined nitrogen forms are again available to support the growth of the filament. In these strains, which are ultimately able to fix atmospheric nitrogen, the transient response to combined nitrogen deficiency involves the induction of the nblA gene and partial degradation of PBS [4, 8]. However, PBS degradation proceeds near completion only in heterocysts, terminally differentiated cells where the photosynthetic apparatus is dismantled, and therefore the consequences of the nblA mutation in Nostoc are more obvious in this specialized cell type, as heterocysts of an nblA null mutant retain PBP [8]. The expression of nblA is induced upon nitrogen removal [9], and the accumulation of nblA transcripts is very low in a mutant of the major nitrogen homeostasis transcription factor gene ntcA [10]. In addition, post-transcriptional regulation by a small, nitrogen-regulated non-coding RNA (NsrR1, nitrogen stress-repressed RNA 1) has been described to inhibit the translation of the nblA transcript in Nostoc [10]. Overexpression of NsrR1 leads to degradation of the nblA transcript, suggesting reduced stability of the untranslated mRNA. Under nitrogen-replete conditions, the interaction of the nblA mRNA with NsrR1 was proposed to reduce translation of the small amounts of NblA protein resulting from leaky nblA transcription [10]. In the case of unicellular non-nitrogen fixing strains, the redox-responsive transcription factor RpaB has been described to regulate nblA expression in response to changes in light availability [2, 11].

Similar to the observations made for other groups of bacteria, transcriptomic analyses carried out in both unicellular and filamentous cyanobacteria have shown abundant transcription of antisense RNAs (asRNAs) [12]. In the case of Nostoc, early transcriptomic approaches already pointed to the relevance of antisense transcription [9, 13], and the recent definition of the transcriptomic map for this organism shows that in this heterocyst-forming cyanobacterium ~65% of the transcriptional units contain regions in antisense orientation to another transcriptional unit [14]. Physiological consequences of such antisense transcription are largely unexplored, but transcription of asRNAs may have regulatory consequences on the genes they overlap. In Nostoc post-transcriptional regulation operated by nitrogen-regulated asRNAs has been described for glpX, encoding sedoheptulose-1,7-bisphosphatase/fructose-1,6-bisphosphatase [15], gltA, encoding citrate synthase [14], and glnA, encoding glutamine synthetase [16]. In the case of glpX and gltA, the expression of their asRNAs specifically in heterocysts would contribute to the metabolic remodeling that takes place in this specific cell type, in particular affecting flows through the Calvin cycle and the tricarboxylic acid cycle, respectively.

The main regulatory mechanisms described for bacterial asRNAs are the co-degradation of the asRNA–mRNA duplexes and different types of transcriptional interference (TI) [17]. While several reports have described the co-degradation of asRNA–mRNA duplexes by RNase III in bacteria [18–20], including cyanobacterial examples [15, 16], reports about TI events are scarce [21, 22]. Recently, the mathematical analysis of rifampicin time-series data has been developed as a new method to directly detect TI events in vivo [23], pointing to a frequent occurrence of this mechanism in bacteria, including the unicellular cyanobacterium Synechocystis sp. PCC 6803 [23].

Antisense transcription of the nblA region was already detected in early transcriptomic analysis of Nostoc, and a transcriptional unit antisense to nblA was defined in its transcriptomic map [14]. Because antisense transcription in the nblA region is conserved in other heterocyst-forming cyanobacteria, we considered the possibility (see below) that such asRNA could regulate the expression of nblA. Using promoter fusions to the gfp gene, encoding green fluorescent protein (GFP), we show that transcription from the as_nblA promoter is stronger than that from the nblA promoter. We have prepared a strain that lacks as_nblA and demonstrate that the amount of nblA transcripts and the accumulation of the NblA protein is much higher in the absence of as_nblA, both in response to nitrogen limitation and exposure to high light. Strains lacking as_nblA, accumulating high levels of NblA, are unhealthy and prone to accumulate mutations that revert their phenotype. Rifampicin time-series analysis and mathematical modeling strongly support that TI is the main regulatory mechanism of as_nblA. Taken together, these observations indicate a critical role of as_nblA in the maintenance of appropriate amounts of NblA protein, which must be kept low when the degradation of PBS is not required.

Materials and methods

Strains and growth conditions

The strains used in this work are described in Supplementary Table S1. Cultures used in the northern blots and primer extension assays included in Fig. 1, RACE experiment (Supplementary Fig. S3B) and rifampicin addition time-series used for Fig. 4 were bubbled with an air/CO₂ mixture [1% CO₂ (v/v)] and grown photoautotrophically at 30°C in BG11 medium [24] containing ferric citrate, instead of ammonium ferric citrate, and 10 mM NaHCO₃. As nitrogen sources 17.6 mM NaNO₃ or 6 mM NH₄Cl plus 12 mM N-tris (hydroxymethyl) methyl-2-aminoethanesulfonic acid-NaOH buffer (pH 7.5) were used, as indicated in each case. For all other experiments, cells were grown in flasks with shaking. Nitrogen deficiency was induced by filtering, washing, and resuspending the cells in nitrogen-free BG11 medium. Standard 4000K light intensity (referred to as low light) was 50 μmol photons·m⁻²·s⁻¹. Exposure to high light was achieved by shifting the culture flasks to 500 μmol photons·m⁻²·s⁻¹. Darkness was achieved by wrapping the culture flasks with aluminum foil and a black plastic bag.

The Nostoc strains containing SmSp^R genes were grown in the presence of streptomycin (Sm) and spectinomycin (Sp), 1.5 µg/ml each (liquid medium) or 3 µg/ml each (solid medium). The medium was solidified by the addition of 10 g/l of Bacto Agar (BD). Escherichia coli strains were grown in LB medium, supplemented with appropriate antibiotics [25].

Construction of Nostoc derivative strains

The plasmids and oligonucleotides used in this work are described in Supplementary Tables S2 and S3, respectively. All PCR (polymerase chain reaction) fragments to be cloned were amplified using high fidelity iProof™ DNA polymerase. The sequences of all PCR-amplified fragments were verified entirely by sequencing.

To generate vectors for the expression of a fusion between the promoters of nblA or as_nblA and a promoterless gfpmut2 gene in Nostoc, the promoter regions were amplified from genomic DNA using oligonucleotides 1008 + 1115 or 1107 + 1108, respectively. The products were digested with AccI–XhoI (nblA promoter) or ClaI–XhoI (as_nblA promoter) and cloned into ClaI–XhoI-digested pSAM270 [26], rendering pIAE92 and pIAE93, respectively. pIAE92 and pIAE93 were introduced in Nostoc wild type by conjugation [27], generating strains that contained the corresponding plasmid inserted in the alpha plasmid.

To generate a strain lacking the nblA region, two overlapping fragments were amplified by PCR using genomic DNA as the template and oligonucleotides 1109 + 1110 and 1111 + 1112, respectively. The resulting products were then used as templates for a third PCR with oligonucleotides 1109 + 1112 resulting in a DNA fragment of the nblA genomic region that lacks the nblA promoter, the complete nblA 5′ UTR, coding sequence and 3′ UTR, and the as_nblA promoter (coordinates 5407030–5408106). This PCR product was digested with BamHI and cloned into the BamHI site of pCSRO [28], rendering pIAE94. pIAE94 was introduced in Nostoc by conjugation [27] and double recombinants were selected rendering strain ΔnblA. Genomic DNA was extracted [29] from clones exhibiting sucrose resistance (double recombinants). The segregation of the introduced deletion was analyzed by PCR using oligonucleotides 563 + 564 (Supplementary Fig. S4B).

Plasmids for the expression of a 3xFLAG-tagged version of NblA (NblA-3xFLAG) in Nostoc were constructed using pMBA20 [15] as a backbone. A segment corresponding to the nblA region (coordinates 5406683–5407871) was amplified using genomic DNA and oligonucleotides 514 + 1117 and cloned into pSpark™ II rendering pIAE116, used as a template for further PCR amplification of different segments. A DNA segment including the promoter of nblA and the nblA coding sequence without sequences downstream of nblA (coordinates 5406949–5407697) was amplified using oligonucleotides 1115 + 1191 (the last one including the sequence encoding 3xFLAG). The resulting fragment was digested with AccI and XhoI and ligated to ClaI–XhoI-digested pMBA20, rendering pIAE119. A segment containing sequences downstream of nblA but not the as_nblA promoter (coordinates 5407701–5408028) was amplified using oligonucleotides 1188 + 1189, digested with XhoI, and ligated into XhoI-digested pIAE119 rendering pIAE130. The same segment plus the sequences corresponding to the as_nblA promoter (coordinates 5407701–5408157) was amplified using oligonucleotides 1188 + 1190, digested with XhoI, and ligated to XhoI digested pIAE119, rendering pIAE123 (see the scheme of constructs in Fig. 3 and Supplementary Fig. S6). pIAE123 or pIAE130 were introduced in the ΔnblA strain by conjugation [27], generating strains that contained the corresponding plasmid inserted in the alpha plasmid.

Plasmids for the constitutive expression of as_nblA in Nostoc were constructed based on vector pMBA37, which was previously designed for expression from the trc promoter. pMBA37 bears unique NsiI and XhoI sites for the cloning of DNA segments between the trc promoter and the T1 transcriptional terminator from the E. coli rrnB gene [15]. Two different fragments were amplified with oligonucleotides 1116 + 1120 or 1117 + 1119, digested with NsiI and XhoI, and cloned into NsiI–XhoI-digested pMBA37, rendering pIAE97 and pIAE99, respectively. Plasmid pMBA51, containing only the trc promoter and the T1 terminator [15], was used as control. Each plasmid was introduced into Nostoc wild-type or Δalr0280 RNase III-defective strains [15] by conjugation [27], generating strains that contained the corresponding plasmid inserted in the alpha plasmid.

RNA isolation and analysis

Total RNA was isolated using hot phenol as described [30] with modifications [31]. When appropriate, the samples were treated with RNase-free DNase, according to the instructions of the TURBO DNA-free™ kit (Invitrogen). Northern blot hybridization was carried out according to standard procedures after RNA separation in urea-polyacrylamide gels [32, 33], using ³²P-labeled probes. The ³²P-labeled nblA strand-specific probe was prepared with Taq DNA polymerase using a PCR fragment generated with oligonucleotides 253 + 254 as the template in a reaction with α-³²P-dCTP and a single oligonucleotide (254) as primer. The ³²P-labeled as_nblA strand-specific probe was prepared in the same way using a PCR fragment generated with oligonucleotides 1116 + 1120 as template in a reaction with α-³²P-dCTP and a single oligonucleotide (1120) as primer. Hybridization with rnpB [34] probes was used as a loading and transfer control. Primer extension analysis was performed as previously described [32] using the oligonucleotides 26 (for nblA) or 990 (for as_nblA) labeled with γ-³²P-ATP. Images of the radioactive membranes were obtained and analyzed using a Cyclone Storage Phosphor System and OptiQuant image analysis software (Packard).

For semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR), synthesis of first strand complementary DNA (cDNA) was performed using Superscript III™ following the manufacturer’s instructions, with 1 µg of total RNA as template and using the oligonucleotides 1303 for nblA mRNA, 493, 1115, 1118, or 1188 for as_nblA RNA, and 1209 for rnpB in a total volume of 20 µl. After dilution to 30 µl and desalting, 1 µl of the reverse transcription (RT) product was used as a template in the PCR reactions. PCR was performed using oligonucleotide pairs 1304 + 1305 for nblA mRNA (27 cycles), 499 + 1188 (Fig. 3, 22 cycles) or 519 + 1118 (Supplementary Fig. S3, 30 cycles) for as_nblA RNA, and 1226 + 1227 for rnpB (22 cycles).

3´-RACE assays were performed essentially as described [35], with 2.5 µg of dephosphorylated total RNA (rAPid Alkaline phosphatase, Roche) extracted from cells incubated for 12 h in the absence of combined nitrogen, incubated for 4 h in the dark, or incubated for 30 min under high light. The 3′RNA-adapter (15 pmol) was ligated to the dephosphorylated RNA by incubation at 25°C for 2 h with T4 RNA ligase (New England Biolabs). After phenol/chloroform extraction and isopropanol precipitation, the RNA was reverse-transcribed with Superscript III™ (Invitrogen) using 2 pmol of oligonucleotide 281, which is complementary to the 3′RNA-adapter. The products of reverse transcription were amplified by PCR using MyTaq DNA Polymerase (Bioline) with oligonucleotides 282 (nested to 281) and gene-specific oligonucleotide 990, using 1 µl of the reverse transcription reaction as a template. PCR products were purified on an agarose gel and cloned into the pSpark™ TA cloning vector. Nine clones from the nitrogen stress sample and 10 clones from the high-light stress sample were sequenced.

For RNA stability analysis shown in Fig. 4, 650 ml culture of Nostoc cells growing in the presence of ammonium or 9 h after nitrogen removal were used. The cells in a 50-ml sample were collected in a Millipore nitrocellulose filter (0.45 μm pore diameter) and the filter was immediately frozen in liquid N₂ (sample labeled time 0). Rifampicin was added to the remaining 600 ml to a final concentration of 500 μg/ml (300 mg in 3 ml of dimethyl sulfoxide) and samples were taken as described above at 0.5, 1, 2, 4, 6, 8, 10, 15, and 30 min following the addition of rifampicin. All samples were filtered and frozen as described for time 0 and stored at −80°C until RNA extraction. A 197-bp-long barcoded spike-in RNA mix was created by in vitro transcription (MEGAshortscript, Thermo Fisher) of PCR-amplified fragments from the PhiX174 bacteriophage genome as described previously [36]. Twenty-three fmol of the spike-in RNA mix was added per 1 μg of total RNA. RNA samples were treated with TURBO DNA-free™ kit (Invitrogen), and strand-specific libraries were prepared using the Illumina Stranded TOTAL RNA preparation RIBO-ZERO PLUS kit and sequenced on the Illumina platform NovaSeq 6000 SP at the Genomics Core Facility of Cabimer (Seville, Spain). Raw RNA-seq data can be accessed in the GEO database under accession number GSE299220.

For RNA samples used in calculation of half-life shown in Fig. 4C, 180 ml culture of Nostoc cells bearing pIAE123 or pIAE130 growing in flasks in the presence of nitrate and subjected to 1 h of 700 μmol photons·m⁻²·s⁻¹ were used. The cells in a 20-ml sample were collected in a Millipore nitrocellulose filter (0.45 μm pore diameter) and the filter was immediately frozen in liquid N₂ (sample labeled time 0). Rifampicin was added to the remaining 160 ml to a final concentration of 500 μg/ml (80 mg in 0.8 ml of dimethyl sulfoxide) and samples were taken as described above at 1, 2, 4, 6, 8, 10, 15, and 30 min following the addition of rifampicin. RNA was isolated, treated with DNase, and synthesis of first strand cDNA was performed using Superscript III™ following the manufacturer’s instructions, with 0.9 µg of total RNA as template and using oligonucleotides 1303 for nblA and 1209 for rnpB in the same reaction. Quantitative PCR (qPCR) was performed using SsoAdvanced™ Universal Inhibitor-Tolerant SYBR® Green Supermix (Bio-Rad #152-5017) following the manufacturer’s instructions for the Bio-Rad CFX96 Cycler. The amplicons used were obtained using oligonucleotides 1501 + 1502 (nblA) and 1226 + 1227 (rnpB). For each data timepoint nblA Cq values were substracted by the respective rnpB control Cq values. Final expression values were calculated by

The resulting expression values were fitted to the simple delayed decay or the rifampicin sensitive termination (RST) model from Wanney et al. [23] using the R nls function with the inverse square root of the expression values as weights to prevent overestimation of the early values. We evaluated which model represents the data better using the Bayesian information criterion BIC, which supported the RST model for the plAE123 data (BIC; simple: −174.4772; RST: −196.1046) and the simple model for plAE130 (BIC; simple: −199.8851; RST: −194.1043).

RNA bioinformatic analysis

Raw RNA-seq data from Nostoc punctiforme PCC 73102 [37, 38], Nodularia spumigena UHCC 0039 [39], and N. spumigena CCY9414 [40] were downloaded from the NCBI Sequence Read Archive under the accession numbers SRR8435111, SRR8435116, SR8435121, SRR15276499, SRR5053804, SRR5053806, SRR5053809, SRR696127, and SRR1572178. Reads were mapped to the N. punctiforme PCC 73102 (ASM2002v1), N. spumigena UHCC 0039 (ASM305447v1), and N. spumigena CCY9414 (ASM34056v3) genomes using HISAT2 [41] as described for the treatment of Nostoc sp. PCC 7120 data [14].

For RNA stability analysis, reads were mapped to the Nostoc genome (ASM970v1) using BWA-MEM [42]. The sequencing depth at each position was calculated using samtools-depth [43] and averaged into segments of 25 nucleotides. Reads associated with the spike-in RNAs were counted using featureCounts [44] and this data was later used by DESeq2 to calculate the size factors needed for the normalization of the samples [45]. The transcription rates, half-life of the transcripts, and TI events were calculated using the Rifi R package [23]. We used the different bins covering the nblA gene to calculate a mean half-life and to assess whether mRNA stability differs significantly between NH₄⁺ and N₂ conditions [t½ = ln(2)/λ, where λ is the decay constant estimated by the NLS fit]. A paired Wilcoxon signed-rank test was applied directly to the per-bin half-lives (P = .0078, n = 8 bins), treating bins as approximate replicates of transcript stability. Bin-level estimates are approximately independent because the transcriptional delay parameter absorbs the position-dependent rifampicin effect, reads from non-overlapping bins partly derive from physically distinct molecules with independent Poisson counting noise, and the paired design mitigates residual correlation shared between conditions at the same position.

Bayesian ODE model of RNA dynamics

Model overview: We developed a mechanistic ordinary differential equation (ODE) model to describe the dynamics of nblA mRNA (RNA, R), its asRNA (asRNA, asR), and NsrR1 (sRNA, S) under nitrogen starvation conditions. The model integrates three regulatory mechanisms: TI between convergent sense and antisense transcription, post-transcriptional regulation via sRNA-mediated RNA decay, and time-varying transcription rates. Parameters were estimated using Bayesian inference by fitting simultaneously to multiple data types while accounting for measurement-specific uncertainties.

Modeled reactions: The model captures the following biochemical processes:

Synthesis:

RNA is transcribed at effective rate syn_R_eff, reduced by TI from asRNA
asRNA is transcribed at effective rate syn_asR_eff, reduced by TI from RNA
sRNA is transcribed at rate syn_S_eff

Degradation:

Free RNA degrades at rate λ_Rfree
Free sRNA degrades at rate λ_Sfree
asRNA degrades at rate λ_asR
RNA–sRNA complexes degrade at rate λ_complex

Complex formation (RNA–sRNA interaction):

RNA + sRNA ⇌ RNA·sRNA complex with association and dissociation rate constants that define the dissociation constant K_d = k_off / k_on (see QSSA below).

State variables: the model tracks three molecular species over time:

R_tot: total RNA (free RNA + RNA·sRNA complex)
S_tot: total sRNA (free sRNA + RNA·sRNA complex)
asR_tot: total antisense RNA (based on experimental findings asRNA does not form a stable complex in this model)

Quasi-steady-state approximation (QSSA):

RNA–sRNA complex formation and dissociation are assumed to be fast relative to synthesis and degradation, justifying a QSSA [46]. The dissociation constant is defined as:

K_d = k_off / k_on,

where k_on and k_off are the association and dissociation rate constants, respectively. Under the QSSA, the complex concentration C at any instant is obtained analytically by solving the binding equilibrium:

C = {(R_tot + S_tot + K_d) − √[(R_tot + S_tot + K_d)² − 4·R_tot·S_tot]}/ 2

Free species concentrations are then recovered as

R_free = R_tot – C
S_free = S_tot – C
K_d is a free parameter estimated by the model.

Time-varying transcription rates:

Transcription rates change over the 24-h nitrogen starvation time course. We model this using piecewise-linear interpolation between anchor points at t = 0, t = 9 h, and t = 24 h:

f_R(t): dimensionless multiplier for RNA transcription, normalized to 1.0 at t = 0
f_asR(t): dimensionless multiplier for asRNA transcription, set to 102.2 at t = 0 [relation to f_R(t = 0) based on reporter assays]
f_S(t): dimensionless multiplier for sRNA transcription, normalized to 1.0 at t = 0.

Explicitly: f(t) = v₀ for t ≤ 0; v₀ + (v₉ − v₀)·t/9 for 0 < t ≤ 9; v₉ + (v₂₄ − v₉)·(t − 9)/15 for 9 < t ≤ 24; and v₂₄ for t > 24, where v₀, v₉, and v₂₄ are the values at the three anchor points. The raw synthesis rates entering the ODE system are therefore:

syn_R_raw(t) = base_syn · f_R(t)
syn_asR_raw(t) = base_syn · f_asR(t)
syn_S_raw(t) = base_syn_S · f_S(t).

The anchor values f_RNA_0, f_RNA_9, f_asRNA_0, f_asRNA_9, and f_asRNA_24 are informed directly by GFP reporter assay data (Data Type 4 below); f_RNA_24, f_sRNA_9, and f_sRNA_24 are free parameters estimated by the model.

Transcriptional interference:

Convergent transcription of sense RNA and asRNA results in mutual TI, whereby elongating RNA polymerases (RNAPs) from one promoter cause premature termination of transcription from the opposing promoter [47, 48]. The effective synthesis rates after TI are:

syn_R_main = max(0, syn_R_raw − θ_asR · syn_asR_raw)
syn_R_eff = syn_R_main + synthesis_leak
syn_asR_eff = max(0, syn_asR_raw − θ_R · syn_R_raw),

where

θ_asR is the efficiency with which asRNA transcription causes premature termination of RNA transcription (estimated parameter).
θ_R is the efficiency with which RNA transcription causes premature termination of asRNA transcription (estimated parameter).
synthesis_leak represents basal RNA production that escapes TI, accounting for the presence of prematurely terminated RNA.

Because asRNA synthesis is ∼102-fold higher than RNA synthesis at t = 0, a θ_asR ∼0.01 results in nearly complete suppression of sense transcription. The synthesis_leak parameter permits low-level RNA accumulation under strong asRNA-mediated blocking. We assume that the asRNA likely blocks all full-length RNA synthesis at t = 0, but that premature terminated RNA exists in the cell, which lead to detectability at low levels in the RNAseq data. In the sRNA knockout condition, syn_S_raw is set to 0, eliminating synthesis of the sRNA and hence interaction with the RNA. In the asRNA knockout condition, syn_asR_raw is set to 0, eliminating TI.

Differential equations

Combining the above, the system evolves according to

dR_tot/dt = syn_R_eff − λ_Rfree · R_free − λ_complex · C
dS_tot/dt = syn_S_eff − λ_Sfree · S_free − λ_complex · C
dasR_tot/dt = syn_asR_eff − λ_asR · asR_tot,

where C, R_free, and S_free are obtained from the QSSA at each time point as described above.

Condition-specific initial conditions

Initial conditions are computed as analytical steady-state solutions, ensuring the system begins at equilibrium for each genetic background:

Wild type (WT): full TI active; sRNA-mediated decay active
sRNA knockout: syn_S_raw = 0 (no sRNA); TI still active
asRNA knockout: syn_asR_raw = 0 (no TI); sRNA-mediated decay active.

This approach avoids arbitrary initial condition choices and ensures biological consistency across conditions.

Estimated parameters

The following parameters are estimated from the data via Bayesian inference:

K_d (RNA–sRNA dissociation constant), λ_Rfree (decay rate of free RNA), λ_Sfree (decay rate of free sRNA), λ_complex (decay rate of RNA–sRNA complex), λ_asR (decay rate of asRNA), base_syn [basal RNA/asRNA transcription rate, i.e syn_R_raw(t = 0)], base_syn_S (basal sRNA transcription rate at t = 0), θ_asR (TI efficiency of asRNA on RNA transcription), θ_R (TI efficiency of RNA on asRNA transcription), synthesis_leak (basal RNA escaping TI), f_RNA_9 & f_RNA_24 (RNA transcription multipliers at t = 9 h, t = 24 h), f_asRNA_9 & f_asRNA_24 (asRNA transcription multipliers at t = 9 h, t = 24 h), f_sRNA_9 & f_sRNA_24 (sRNA transcription multipliers at t = 9 h, t = 24 h), and measurement noise per data type (σ_rnaseq, σ_microarray, σ_qpcr, σ_nb1, σ_decay, σ_reporter).

Fitting data sources and connection to model variables

We fit the model to fold-changes and ratios rather than absolute expression values, because different measurement technologies (RNA-seq, microarray, northern blot, RT-PCR) have incompatible absolute scales due to differences in sensitivity, dynamic range, and normalization. Relative changes are more reproducible and biologically meaningful, and allow the model to be constrained without requiring knowledge of technology-specific scaling factors. All observations are modeled with lognormal likelihoods:

log(observed ratio) ∼ Normal(log(predicted ratio), σ)

The following observed data contribute to the joint log-likelihood, each constraining the model by comparing experimental measurements to predicted values of state variables, their ratios, or derived quantities such as effective decay rates:

RNA-seq ratios at t = 0 and t = 9 h (this work and Mitschke et al. [9]): At t = 0, the observed asRNA/RNA ratio was 1152/19.9 = 57.9 and the sRNA/RNA ratio was 3098/19.9 = 155.7 (sRNA from dRNA-seq). At t = 9 h, the asRNA/RNA ratio was 1017/923 = 1.1. These ratios constrain the relative abundances of the three species and anchor the initial condition balance. The dramatic shift in asRNA/RNA ratio from 57.9 at t = 0 to 1.1 at t = 9 h reflects the transition from TI-dominated (low RNA) to sRNA-decay-dominated (higher RNA) regulation. Noise parameter: σ_rnaseq.
RNA-seq RNA fold-change, WT t = 0 to t = 9 h (this work): RNA_t0 = 19.9, RNA_t9 = 923, giving an observed fold-change of 46.4. This constrains R_tot(t = 9)/R_tot(t = 0) for the WT trajectory and anchors the dynamic range of RNA induction. Noise parameter: σ_rnaseq.
Northern blot time series for WT and sRNA knockout (Fig. 5C and Álvarez-Escribano et al. [10]): For the WT condition, raw signal values at t = 0, 3, 6, 9, 12, and 24 h were 1, 9, 14, 19, 18, and 7 (arbitrary units), normalized to t = 3 h, giving fold-changes of 14/9, 19/9, 18/9, and 7/9 at t = 6, 9, 12, and 24 h, respectively. For the sRNA knockout, raw values at t = 0, 2, 4, 6, and 8 h were 1, 5.76, 9.77, 38.1, and 46.9, normalized to t = 2 h, giving fold-changes of 9.77/5.76, 38.1/5.76, and 46.9/5.76 at t = 4, 6, and 8 h, respectively. Reference timepoints of t = 3 h (WT) and t = 2 h (sRNA KO) were chosen because the t = 0 signal was near the detection limit. These data constrain the shape of the WT and sRNA KO RNA trajectories. Noise parameter: σ_nb.
The GFP reporter assay values (Fig. 2B): f_asRNA_9 = 120.7, f_asRNA_24 = 96.7, and f_RNA_9 = 10.89 (all relative to t = 0 RNA transcription) directly inform the transcription multiplier parameters at t = 9 h and t = 24 h. Noise parameter: σ_reporter (GFP reporter values).
RT-PCR fold-changes for the asRNA knockout (Fig. 3C): In the asRNA KO condition, raw signal values were 1.82 at t = 0 and 4.47 at t = 9 h, giving an observed fold-change of 4.47/1.82 = 2.46. The ratio of asRNA KO to WT at t = 9 h was 4.47/0.805 = 5.55. These constrain the magnitude of TI (θ_asR) by quantifying how much RNA increases when asRNA-mediated interference is absent. Noise parameter: σ_qpcr (RT-PCR ratios).
Cross-strain ratios (northern blot and RT-PCR): WT/sRNA KO ratio at t = 6 h was 14/38.1 = 0.37, reflecting that sRNA KO RNA exceeds WT at this timepoint due to loss of sRNA-mediated decay. The asRNA KO/WT ratio was 4.47/0.805 = 5.55 at t = 8 h and 3.35/1.31 = 2.56 at t = 24 h, reflecting the progressive reduction of TI impact as RNA synthesis increases during nitrogen starvation. These constrain the regulatory effect sizes of sRNA-mediated decay (WT versus sRNA KO) and TI (asRNA KO versus WT). Noise parameters: σ_nb (northern blot) and σ_qpcr (RT-PCR).
Microarray relative changes [26]: RNA, sRNA, and asRNA signals at t = 6, 8, 12, 24 h were 11.24, 12.55, 14.32, 6.45 (RNA); 0.272, 0.332, 0.321, 0.779 (sRNA); and 1.181, 0.590, 0.727, 0.946 (asRNA),respectively, all expressed relative to t = 6 h. These constrain the temporal dynamics of all three species simultaneously across the time series. Ratios relative to t = 0 were not used due to low nblA expression at t = 0 being at or near microarray background. Noise parameter: σ_microarray.
Effective decay rate measurements (Fig. 4A): The apparent decay rate of RNA was measured at t = 0 (13.73 h⁻¹) and t = 9 h (4.4 h⁻¹). The model predicts the effective decay rate as:λ_eff(t) = (λ_Rfree · R_free(t) + λ_complex · C(t))/R_tot(t) The decay rate decrease reflects the transition from sRNA-excess conditions at t = 0 (high complex fraction, fast apparent decay) to sRNA-depleted conditions at t = 9 h (more free RNA, slower apparent decay). Noise parameter: σ_decay.
Informed t = 0 ratio for asRNA KO versus WT: WT RNA is below the northern blot detection limit at t = 0, while the asRNA KO signal is detectable (1.82 au). Since the observed asRNA KO/WT ratio decreases from 5.55 at t = 8 h to 2.56 at t = 24 h, the ratio at t = 0 must be substantially larger than 5.55. We model this as a lognormal likelihood centered on 20 with σ = 0.7 (log scale), representing a conservative lower bound on the ratio R_tot_asRNA-KO(t = 0)/R_tot_WT(t = 0).

Inference and convergence

The model was implemented in Stan [49] using the cmdstanr interface [50] in R version 4.3.2 and cmdstan 2.34.1. Sampling used the No-U-Turn Sampler with 8 chains, each with 2000 warmup and 2000 sampling iterations (adapt_delta = 0.95; max_treedepth = 15; seed = 42; 16 000 total post-warmup samples). Convergence was assessed using the split R̂ statistic and effective sample size (ESS). All parameters showed R̂ ≤ 1.002 and both bulk and tail ESS exceeding 4255. A minor fraction of iterations (0.12%, 16/16 000) resulted in divergent transitions; visual inspection confirmed these were randomly distributed throughout the posterior and did not cluster near boundaries, suggesting they reflect stochastic numerical limitations in the ODE solver rather than systematic posterior geometry problems. The fitted model R object and the scripts are available at GitHub (https://github.com/JensGeorg/Bayesian-modeling-of-nblA-regulation.git) and Zenodo (https://doi.org/10.5281/zenodo.19221631).

Fluorescence microscopy

The fluorescence of filaments of Nostoc (wild type, ΔnblA, or strains bearing plasmids pIAE92, pIAE93, or pIAE123) growing for 4–5 days on solidified nitrogen-free medium supplemented with appropriate antibiotics was analyzed as described [51] using a Leica HCX PLAN-APO 63_ 1.4 NA oil immersion objective attached to a Leica TCS SP2 confocal laser-scanning microscope (Fig. 2) or an Olympus FLUOVIEW FV3000 confocal laser-scanning microscope equipped with an UPlanApo 60x/1.5 NA oil immersion objective (Supplementary Fig. S4C). Samples were excited at 488 nm, and the fluorescent emission was monitored by collection across windows of 500–538 nm (GFP) and 630–700 nm (cyanobacterial autofluorescence) (Fig. 2 and Supplementary Fig. S4C). Lambda scan analysis of Nostoc filaments carrying plasmids pIAE123 or pIAE130 growing for 4–5 days on top of solidified nitrogen-free medium supplemented with appropriate antibiotics was carried out using an Olympus FLUOVIEW FV3000 confocal laser-scanning microscope equipped with an UPlanApo 60x/1.5 NA oil immersion objective (Supplementary Fig. S5). Samples were excited at 488 nm and the fluorescent emission was monitored by collection in 5-nm-wide windows in the range 600–720 nm. Quantification of GFP accumulation (Fig. 2D) or single cell emission spectra (Supplementary Fig. S5B) was determined from the images using Fiji [52].

Cell fractionation and western blotting

Crude extracts were prepared using glass beads. Cells from 30 ml cultures grown to a chlorophyll content of 4–5 μg/ml were harvested by filtration, washed with 50 mM Tris–HCl buffer (pH 8), and resuspended in 500 µl of resuspension buffer (50 mM Tris–HCl, pH 8.0, 2 mM of 2-mercaptoethanol) containing a protease inhibitor cocktail (cOmplete, EDTA-free, Roche). The cellular suspension was mixed with ~75 µl of glass beads (Sigma, 212–300 μm) and subjected to seven cycles of vortexing for 1 min followed by cooling on ice for 1 min. The cell extract was separated from cell debris and unbroken cells by centrifugation (3 min at 3000 × g at 4°C). The soluble fraction was obtained by centrifugation of the crude extract at 16 000 × g for 30 min at 4°C. The protein concentration was determined as described [53].

For western blot analysis of Nostoc proteins, 20–30 µg of proteins from the soluble fraction were fractionated on 10% (Fig. 2) or 15% (Fig. 3) SDS/PAGE (sodium dodecyl sulfate–polyacrylamide gel electrophoresis), transferred to nitrocellulose membranes (Transblot Turbo RTA Transfer Kit, Bio-Rad), and immunoblotted with monoclonal anti-GFP 1:5000 (Roche, #11814460001) or monoclonal anti-FLAG M2-Peroxidase (HRP) 1:5000 (Sigma #A8592) antibodies. The ECL Plus immunoblotting system (Immobilon Western Chemiluminescent HRP Substrate, Merck) was used to detect the different antibodies, using anti-mouse 1:3000 (Bio-Rad #1706516) horseradish peroxidase-conjugated secondary antibodies in the case of anti-GFP. Ponceau Red staining was used as the loading and transfer control.

Growth curve and whole cell spectra

Cells were grown under standard conditions in the presence of NO₃⁻ supplemented with appropriate antibiotics. Growth was monitored at different time points by the absorbance at 750 nm of samples from the different cultures mixed with an equal volume of 50% glycerol. The absorbance spectra of the different cultures were taken between 400 and 750 nm on a microplate reader (Varioskan) and normalized for the chlorophyll content by the absorbance at 680 nm after subtraction of the absorbance at 750 nm.

Isolation and analysis of suppressors of the yellow phenotype of NblA overexpressing strains

Cells containing pIAE130 inserted in the alpha plasmid were plated and grown for 15 days on solidified nitrogen-free medium supplemented with appropriate antibiotics. Genomic DNA was isolated as described [29] from individual green colonies, and the nblA region was amplified by PCR using oligonucleotides 59 + 68 and 5 ng of genomic DNA as template. PCR products were purified (Isolate II PCR and Gel kit, Bioline) and sequenced using the same oligonucleotides.

Results

An antisense transcript to the nblA mRNA is produced in heterocyst-forming cyanobacteria

The transcriptome of Nostoc cells growing in the presence of ammonium or subjected to nitrogen deprivation for 9 or 24 h has been previously analyzed [14] (Supplementary Fig. S1A). The transcriptional unit TU03050 corresponding to the nblA gene contains, in addition to the 195-nt nblA coding sequence, long 5′ and 3′ UTRs (Fig. 1A). Five different TSSs (TSS1 to TSS5) were previously identified for the nblA mRNA [9] and three of them are induced upon nitrogen deprivation [9, 10] (Fig. 1B, left panel). According to our previous data [14], a transcriptional unit (TU03051) antisense to nblA would also be transcribed from the region depicted in Fig. 1A. We analyzed available transcriptomic datasets for heterocyst-forming cyanobacteria [37–40] and found that long 5′ and 3′ UTRs and the presence of antisense transcription in the nblA region are conserved features in the heterocyst-forming cyanobacteria N. punctiforme PCC 73102 (Supplementary Fig. S1B), N. spumigena UHCC 0039 (Supplementary Fig. S1C), and N. spumigena CCY9414 (Supplementary Fig. S1D).

Here, we analyzed the regulation of nblA expression in Nostoc by exposure to high light intensity or darkness. Exposure to high light induced the transient expression of nblA, already detected after 15 min, from the three nitrogen-regulated TSSs (TSS1, TSS2, and TSS5), identified either by primer extension (Fig. 1B, middle panel) or by northern blot hybridization (Fig. 1C). The three major transcripts detected were consistent with transcription from TSS1, TSS2, and TSS5. In response to high light, expression from TSS5 was more highly induced than expression from TSS1 or TSS2. A shift to darkness produced a decrease in the low level of nblA expression observed under standard light conditions (Fig. 1B, right panel).

We confirmed transcription of TU03051 (renamed as_nblA) by primer extension (Fig. 1D) and northern blot (Fig. 1E). Primer extension with an oligonucleotide corresponding to the end of the nblA sequence (#990 in Fig. 1A) resulted in the detection of a main band whose size is in agreement with a transcript that starts at the TSS located previously by dRNA-seq at position 5408038r [9]. Northern blot hybridization showed a major band with a size also consistent with the length annotated for TU03051 (840 nt) [14] and several additional smaller bands that could result from processing of the main transcript or premature termination. The expression of as_nblA was only transiently repressed upon nitrogen removal or high-light conditions, but induced upon exposure to darkness (Fig. 1D and E). The size of the largest band detected in the northern blots indicates that as_nblA completely overlaps the nblA gene.

Sequence conservation in the promoter regions upstream nblA across different cyanobacterial strains is shown in Supplementary Fig. S2. No conserved sequence corresponding to binding of NtcA, known to regulate nblA induction upon nitrogen removal [9], is observed. However, other conserved sequences are found upstream of TSS1, 2, and 5. The position of conserved sequences is compatible with binding of transcriptional activators in the case of TSS1, 2, and 5, but in the case of TSS5 additional conserved sequences overlapping the −10 region suggest a transcriptional repressor. In fact, the conserved sequences overlapping −10 region in TSS5 are similar to the described binding site of the RpaB regulator [11], which has been identified as regulator of nblA in unicellular strains [2].

The 3′ end of nblA transcripts produced either upon nitrogen deficiency or exposure to high light was analyzed by RACE, and the results indicate that termination is most frequent around 91–94 nt after the stop codon of nblA (Supplementary Fig. S3, white asterisk inside the gray arrow depicting TU03050 in Fig. 1A). As expected, and in agreement with the primer extension shown in Fig. 1B, a very weak band was detected in RACE using RNA isolated from cells incubated in the dark (Supplementary Fig. S3B). The extension of as_nblA was analyzed by RT-PCR using oligonucleotides at three different positions to generate cDNA. The results (Supplementary Fig. S3D) indicate that the 3′ end of the as_nblA transcripts reaches the position of TSS1, and, to a lesser extent, even further upstream, since a faint band was observed even using cDNA generated with oligonucleotide #1115 (see scheme in Supplementary Fig. S3A). Therefore as_nblA entirely overlaps the nblA transcripts in the antisense disposition.

Expression of nblA and as_nblA in nitrogen-fixing filaments of Nostoc

To analyze expression of nblA and its antisense transcript, as_nblA, we prepared plasmids bearing the promoter regions of TU03050 (nblA) (pIAE92) or TU03051 (as_nblA) (pIAE93) fused to the gene encoding GFP (see scheme in Fig. 2A). The plasmids were introduced in Nostoc by conjugation and expression of GFP from each promoter region was analyzed by western blot using anti-GFP antibodies. GFP was barely detectable in extracts from ammonium-grown cells bearing the P_nblA–gfp fusion (pIAE92) and ammonium removal led to accumulation of GFP (Fig. 2B). In contrast, a much higher amount of GFP was detected in the case of cells containing the P_{as_nblA}−gfp fusion (pIAE93) both in the presence or absence of ammonium (Fig. 2B), indicating unregulated expression of as_nblA. GFP expression in cells bearing pIAE92 or pIAE93 was also analyzed along filaments growing on nitrogen-free plates by confocal fluorescence microscopy (Fig. 2C). Expression of GFP in filaments containing the P_nblA−gfp fusion (pIAE92) was stronger in heterocysts than in vegetative cells (Fig. 2C and D), whereas expression of GFP in filaments containing the P_{as_nblA}−gfp fusion (pIAE93) was comparatively very strong but unpatterned along the filaments (Fig. 2C). Figure 2D shows the quantification of the green (GFP) and red (chlorophyll autofluorescence, shown in magenta) signals in filaments of cells bearing pIAE92. Three heterocysts are indicated in the image; two of them (1 and 3) show reduced autofluorescence, an indication of maturity, whereas a heterocyst located at an intermediate position with respect to the two mature heterocysts (2) would be at an earlier stage of differentiation according to the presence of red autofluorescence.

Transcription of an antisense to nblA affects NblA accumulation and degradation of phycobiliproteins

We hypothesized that the transcription of as_nblA might affect the expression of nblA and contribute to the regulation of NblA accumulation in Nostoc. To test this hypothesis, we first prepared a strain in which the entire region encoding both TU03050 (nblA) and TU03051 (as_nblA) was deleted, so that neither nblA nor as_nblA would be transcribed in this strain (see the scheme and verification of deletion in Supplementary Fig. S4A and B). It has been previously described that the lack of NblA leads to a phenotype characterized by heterocysts that maintain red autofluorescence, similar to vegetative cells, in contrast to the wild-type strain in which heterocysts lack red autofluorescence as a consequence of PBS degradation [8]. The mutant phenotype of the ΔnblA strain we constructed was therefore verified by fluorescence confocal microscopy (Supplementary Fig. S4C, middle panels). Complementation of the mutant phenotype of the ΔnblA strain was achieved by introducing a construct encoding a 3xFLAG-tagged version of NblA (see pIAE123 in Fig. 3A) in this strain. The lack of red autofluorescence in the heterocysts of the strain bearing pIAE123 (Supplementary Fig. S4C, right panels) indicates that the 3xFLAG-tagged version of NblA is functional and restores the reduction of the red autofluorescence phenotype observed in the heterocysts of the wild-type strain.

Two versions of the nblA region encoding 3xFLAG-tagged NblA were compared (see scheme in Fig. 3A). pIAE123 included both the entire nblA promoter and the as_nblA promoter, so that both the nblA mRNA (with its 5′ and 3′ UTRs) and as_nblA would be transcribed. In contrast, pIAE130 contained the nblA gene and its upstream and downstream regions but not the as_nblA promoter region. It should be noted that the most common 3′ UTR of nblA mRNA extends only ~100 nt downstream the end of the coding sequence according to 3′ RACE (Supplementary Fig. S3C), therefore, deletion of the as_nblA promoter in pIAE130 does not alter the 3′ UTR of the nblA mRNA produced. In both cases, the T1 terminator of the rrnB gene of E. coli was included after the nblA region. Transcription of as_nblA was determined by RT-PCR and, as expected, took place in the strain bearing pIAE123 but not in the strain bearing pIAE130 (Fig. 3B). This result confirms that the deletion of the promoter corresponding to the TSS at position 5408038r leads to the absence of as_nblA transcription.

We then analyzed nblA expression and the accumulation of the 3xFLAG-tagged NblA protein in strains containing pIAE123 or pIAE130 in response to either combined nitrogen removal (Fig. 3C, E) or exposure to high light intensity (Fig. 3D and F). When the entire region encoding both nblA and as_nblA was present (strain bearing pIAE123), the nblA transcript and the NblA protein were barely detectable in the presence of combined nitrogen (lanes labeled 0) and could only be observed upon removal of combined nitrogen (lanes labeled 8 and 24) (Fig. 3C and E). This is consistent with the previously observed induction of the nblA promoter upon removal of combined nitrogen (Figs 1B and C, and 2B). In contrast, in the strains bearing pIAE130 (therefore lacking as_nblA), both the nblA transcript and the 3xFLAG-tagged NblA protein were observed even in the presence of combined nitrogen (lanes labeled 0). Similar observations were made in the case of nblA expression and NblA accumulation in response to exposure to high light (Fig. 3D and F). In this case, in the strain bearing pIAE130, both the nblA transcript and the 3xFLAG-tagged NblA protein were observed even in the presence of regular low light before exposure to high light intensity (lanes labeled 0). Primer extension analysis (Fig. 3C and D, top panels) indicates that transcription from TSS5, which is induced either by nitrogen removal or exposure to high light in cells bearing pIAE123, takes place in the strain bearing pIAE130 even in the presence of combined nitrogen under standard illumination (lanes labeled 0). Therefore, as_nblA seems to be required to suppress the background transcription of nblA under non-inducing conditions.

The integrity of phycobilisomes is altered in strains lacking as_nblA

A decrease in the amount of PBP in cell extracts of strains bearing pIAE130 could be observed in the membranes stained with Ponceau red (Fig. 3E and F, lower panels). Since PBS degradation is an indication of NblA function, this observation is consistent with the increased accumulation of NblA in the strain containing pIAE130 (Fig. 3E and F, upper panels). Liquid cultures of the strains bearing pIAE123 and pIAE130 showed a difference in color, which was yellowish in the case of cells bearing pIAE130 (that do not transcribe as_nblA) but green in the case of cells bearing pIAE123 (that includes the entire region encoding nblA and as_nblA) (Fig. 3G). Whole cell absorption spectra taken of cells growing in the presence of nitrate showed that the PBS content was lower in the strain bearing pIAE130 than in the strain bearing pIAE123 (Fig. 3G). We also analyzed filaments of strains bearing pIAE123 or pIAE130 by performing single-cell in vivo fluorescence spectra at room temperature. Cells were excited at 488 nm (a wavelength that excites phycocyanin and chlorophyll), and images were taken at 5-nm wide windows for wavelengths ranging from 600 to 720 nm. Images corresponding to 645 nm (emission of phycocyanin) and 680 nm (emission of photosystem II chlorophyll) are shown (Supplementary Fig. S5A), together with the fluorescence emission recorded for three heterocysts of each strain in the range 600–720 nm (Supplementary Fig. S5B). Comparison of the graphs shows that the peak corresponding to the fluorescence at 645 nm is missing in heterocysts of the strain containing pIAE130, an indication of a more extensive degradation of PBP in those heterocysts.

Accumulation of NblA in strains lacking as_nblA is deleterious

The growth rate of the strain bearing pIAE130 was reduced in comparison to that of the strain bearing pIAE123, indicating that the reduction in the PBS content under standard autotrophic growth conditions is detrimental for growth (Supplementary Fig. S6A and B). In plates containing the strain bearing pIAE130, green colonies appeared with time over the yellowish background (Supplementary Fig. S6C). Green colonies were isolated from plates lacking combined nitrogen of cells containing pIAE130, and the region encoding NblA was PCR amplified and sequenced. In all cases the amplified fragments were longer than expected. Sequencing of PCR fragments obtained from 14 different colonies revealed that they carried insertions of a transposase, identical to those present in the chromosome of Nostoc or in plasmids alpha, beta, or delta, in the promoter or coding sequence of nblA (Supplementary Fig. S6D), indicating that those transposases had been mobilized and disrupted the expression of nblA.

as_nblA regulates nblA expression through transcriptional interference

To decipher the regulatory mechanism(s) of as_nblA, we first explored the possibility of a co-degradation mechanism. For this purpose we generated strains expressing in trans two segments of as_nblA under the control of a strong and constitutive promoter (Supplementary Fig. S7A), an experimental approach that we have used previously to analyze antisense regulation by RNase III-dependent duplex degradation [15, 16]. The different as_nblA versions were introduced both in a wild-type and a Δalr0280 strain, an RNase III deletion mutant [15]. The overexpression in trans of either version of as_nblA produced no clear reduction in nblA mRNA accumulation in any of the two genetic backgrounds (Supplementary Fig. S7B). This is in contrast with the clear effect observed for as_nblA when expressed in cis (Fig. 3).

The observation that as_nblA affects nblA mRNA accumulation when expressed in cis but not when it is transcribed in trans from a strong promoter suggests that a TI mechanism is involved in the regulation by as_nblA. The analysis of naturally occurring TI events entails genetic and technical challenges. Recently, the mathematical analysis of RNA-seq data from rifampicin time-series has been developed as a new method to detect in vivo TI events in bacteria [23]. Since rifampicin prevents transcription initiation but does not affect elongation, transcript levels at positions downstream of a TSS remain constant until the last elongating RNAP passes that position [54]. This leads to a delayed onset of exponential decay (the “delay”) that depends on the distance to the TSS and the elongation speed. In addition to this behavior, Wanney et al. detected in phylogenetically diverse organisms that the abundance of some transcript fragments actually increased after rifampicin addition, which can be the consequence of rifampicin relieving termination caused by TI due to collision of RNAPs transcribing in opposite directions, generating an increase in the transcripts abundance after rifampicin addition [23].

To take advantage of this new methodology, we analyzed transcriptomic data from wild-type cells subjected to treatment with rifampicin, both in the presence of ammonium, where only basal transcription of nblA would be observed, or 9 h after ammonium removal, when induction of nblA has already taken place. Bioinformatic analysis using the “Rifi” R package [23] revealed several findings in the genomic region around nblA (Fig. 4).

First, in the presence of ammonium, the synthesis rate of as_nblA was much higher than that of nblA, and we detected an increase in nblA transcription from TSS1 9 h after nitrogen removal. These observations are consistent with results shown in Figs 1B and C, and 2. Second, the observed difference in mRNA stability was highly significant (paired Wilcoxon signed-rank test, P = .0078), with the mean half-life of nblA mRNA increasing approximately three-fold, rising from 3.03 min under NH₄⁺ conditions to 9.46 min after combined nitrogen removal (Fig. 4A). Third, in the presence of ammonium, we detected a particularly long delay (∼4 min) for the start of the exponential decay of the nblA transcript after the addition of rifampicin (Fig. 4B, gray shaded area in the bottom panels). This long delay suggests a significant distance of the positions analyzed from their TSS, indicating that the detected small amount of nblA transcripts in the presence of ammonium primarily originates from readthrough transcription of the upstream operon. Fourth, we observed a characteristic pattern of increased transcript abundance after rifampicin addition, consistent with TI, around position 5406950 in cells grown in the presence of ammonium (Fig. 4B, bottom panels, middle). This position corresponds to the end of the transcript resulting from readthrough transcription from the upstream gene alr4516. This characteristic pattern is detectable because this position is sufficiently far from the upstream transcript’s TSS to be observed within the timeframe of our rifampicin time-series. Conversely, we detected no increased transcript abundance after rifampicin addition resulting from TI within the 5′UTR of nblA mRNA during nitrogen depletion nor delay in the start of exponential decay (Fig. 4B, position 5407250 upper panel). Under these conditions, nblA is strongly transcribed from its own promoters. The increased sense to antisense synthesis ratio leads to a much lower total fraction of termination by interference at this condition, which is not visible in the data anymore. In order to further test the possibility of a TI mechanism for as_nblA, we analyzed nblA accumulation after rifampicin addition in strains bearing pIAE123 and pIAE130 subjected to 1 h of high-light stress (Fig. 4C). In this case, we used RT-qPCR and an amplicon internal to nblA but as far as possible from the nblA TSSs (positions 5407566–5407673) to try to detect the characteristic increase after rifampicin addition within our timeframe resolution. No differences in nblA mRNA half-life between the two strains were detected, whereas a characteristic increase after rifampicin addition was only observable in the strain bearing pIAE123 (that produces as_nblA) but not in the strain bearing pIAE130 (that does not produce as_nblA).

A mathematical model of nblA regulation

To better understand the physiological implications of the TI mechanism exerted by as_nblA, we modeled the expression of nblA based on the data presented here and previously available experimental data. Our model for nblA regulation considers two key observations. First, as_nblA exerts its regulatory function mainly through TI as lacking the asRNA did not change the mRNA stability (Fig. 4C). Second, a significant, approximately three-fold difference was observed in nblA RNA stability between standard and nitrogen depletion conditions (Fig. 4A). We propose that the difference in nblA mRNA half-life is primarily due to the interaction and co-degradation of nblA mRNA with the small RNA (sRNA) NsrR1, as supported by literature [10]. A decreased translation rate for the nblA:NsrR1 complex was not incorporated in the model as our focus was on RNA levels.

This leads to a model (Fig. 5A) where nblA mRNA levels are regulated by (i) its synthesis rate, mainly driven by the three nitrogen-regulated inducible TSSs. The relative synthesis rate of as_nblA and nblA was estimated from the transcriptional fusions shown in Fig. 2. (ii) Interaction with the sRNA NsrR1 that increases the nblA decay. There are no quantitative data for the actual dynamics of the NtcA-dependent promoter activity of NsrR1, thus these data are fitted by the model priored by microarray expression data [26]. (iii) TI by transcription of the as_nblA asRNA, which directly reduces the synthesis of full-length mature nblA mRNA. Informed priors for RNA stability were taken from our rifampicin time-series data (Fig. 4).

Using Bayesian fitting, we demonstrate that the proposed model successfully captures the observed biological effects and describes a comprehensive set of available experimental time-series data during nitrogen depletion. These include microarray and RNA-seq expression data for nblA, as_nblA, and nsrR1 in a WT background [14, 26], northern blot data for nblA in WT and NsrR1 knockout strains [10], RT-PCR data for nblA in strains bearing pIAE123 (wild type) or pIAE130 (lacking as_nblA) (Fig. 3), and observed decay constants of nblA at standard and nitrogen depletion conditions (Fig. 4).

Using the fitted parameters (Supplementary Table S4), we modeled the observed (free + complexed nblA:NsrR1) concentrations for nblA mRNA in the wild type, as_nblA knockout (Δas_nblA), and NsrR1 knockout (ΔNsrR1) scenarios (Fig. 5C and Supplementary Fig. S8). In the wild type, nblA, as_nblA, and NsrR1 are all present, and TI by the asRNA effectively suppresses nblA expression from readthrough transcription of the upstream operon during the early phase of nitrogen depletion. A low mRNA collision termination probability of ca. 1% is sufficient to achieve this effect given the high asRNA synthesis rate (Fig. 5C). Finally, although disentangling gene expression in vegetative versus heterocyst cells from bulk transcriptomics is challenging, our single-cell transcriptional GFP fusion measurements (Fig. 2) allow us to estimate that the nblA synthesis rate is approximately three-fold higher in heterocysts compared to vegetative cells under nitrogen depletion (Fig. 2D). Incorporating this difference into our simulations reveals that the system permits rapid nblA induction in heterocysts (rising after ∼0.1 h), whereas vegetative cells remain at baseline levels (∼15 a.u.) for ∼0.5 h. While defining a strict critical threshold for nblA concentration is difficult, the steady-state levels of mutant strains with known growth phenotypes serve as anchors. The baseline nblA level of the ΔNsrR1 strain (∼104 a.u.), which exhibits a minor phenotype under standard conditions [10], is reached after only 0.6 h in heterocysts, compared to 2.3 h in vegetative cells. Furthermore, the baseline level of the Δas_nblA strain (∼534 a.u.), which displays a severe growth phenotype, is reached after 2.2 h in heterocysts but requires 7.5 h in vegetative cells. Notably, the maximal induction in vegetative cells (∼861 a.u.) only marginally exceeds the baseline level in strain Δas_nblA (Supplementary Fig. S9). Ultimately, the model predicts that the regulatory system amplifies a 3-fold difference in synthesis rate (heterocysts versus vegetative cells) into an ∼11.6-fold difference in maximal nblA concentration (Fig. 5C and Supplementary Fig. S9). The actual concentration difference is likely even greater, as NsrR1 expression is presumably lower in heterocysts due to NtcA repression, an effect not included in this model due to a lack of quantitative data.

Discussion

The proteolysis adapter NblA plays a central role in PBS degradation and must therefore be tightly controlled to balance stress adaptation with cellular homeostasis. At the transcriptional level, nblA induction in response to nitrogen deprivation is regulated by NtcA in Nostoc [9], although no consensus NtcA-binding sequence is identified in the conserved sequences upstream of the regulated TSS (Supplementary Fig. S2). This observation suggests that NtcA indirectly regulates nblA expression through a yet unknown transcription factor. We have shown here induction of expression of nblA in response to high light (Fig. 1B and C), and conserved sequences corresponding to an RpaB-binding site are located upstream of TSS5 in a position compatible with repression by RpaB (Supplementary Fig. S2), similar to the case of unicellular strains [2, 11]. At the post-transcriptional level, NblA accumulation is negatively affected by binding of an sRNA, NsrR1, preventing translation of NblA and promoting nblA mRNA degradation in the presence of combined nitrogen [10].

Accumulating evidence points to the physiological consequences of antisense transcription, which is widespread in cyanobacteria. In fact, the transcriptional landscape of Nostoc contains ~65% of transcripts with a segment in antisense orientation to other transcripts [14]. In this work, we show that as_nblA entirely overlaps the nblA mRNA in antisense disposition. Transcription from the promoter of as_nblA is much stronger than transcription from the promoter of nblA (Fig. 2) and takes place also in the presence of combined nitrogen and under standard illumination, a condition in which transcription of nblA is barely detectable. According to fluorescence microscopy of strains bearing a fusion to gfp, the expression of nblA is stronger in heterocysts than in vegetative cells (Fig. 2C and D).This observation is in agreement with the immunolocalization of NblA carried out in Tolypothrix sp. showing that, upon nitrogen step-down, NblA is preferentially located in the differentiated heterocysts [55], and its presence correlates with the observed reduction of autofluorescence that is characteristic of mature heterocysts but does not take place in nblA mutants of Nostoc [8] (Supplementary Fig. S4).

Detection of the native NblA protein had remained elusive in Nostoc, because antibodies raised against NblA from Nostoc were unable to detect native NblA in protein extracts of this strain [4, 10]. Therefore, we have used a functional 3xFLAG-tagged version of NblA for detection in protein extracts. A strain that contains sequences encoding both the nblA mRNA and the as_nblA (Fig. 3) showed regulated expression of nblA and accumulated NblA in response to either nitrogen removal or exposure to high light. In contrast, a strain in which as_nblA is not transcribed (Fig. 3A and B) accumulated NblA even in the presence of combined nitrogen and under regular standard light intensity (Fig. 3). These observations demonstrate the role of as_nblA in maintaining regulated low levels of NblA under non-stress conditions.

Unregulated high accumulation of NblA in the strain lacking as_nblA resulted in cultures with a yellowish color, reduced amounts of PBS, and slightly decreased growth rate even in the presence of nitrate. When plated in the absence of combined nitrogen, the colonies were yellow and showed extremely poor growth. After a few days of incubation, green colonies arose in the plates that seemed to have overcome the deleterious consequences of NblA accumulation (Supplementary Fig. S6A–C). Although the reasons for such deleterious effects are currently unknown, one possibility is that non-physiological amounts of NblA may operate the proteolytic processing of additional proteins besides PBP, since NblA is reported to interact with other proteins not related to PBP degradation [4]. In fact, a recent report describes that the ectopical overexpression of NblA encoded in cyanophages leads to proteolytic processing of additional photosynthesis-related proteins beyond PBP, as well as proteins not involved in the photosynthetic process [7]. Additionally, excess NblA might titrate the Clp protease and inhibit proper processing of other proteins. Green colonies appeared with high frequency in the strain that lacks as_nblA, suggesting a selective pressure to revert the deleterious effects of increased accumulation of NblA in this strain. In fact, sequencing of the nblA region from those green colonies revealed the insertion of different transposases, preventing the production of a functional NblA protein (Supplementary Fig. S6D). Perhaps the stress induced by the overexpression of nblA could activate the mobilization of some of the many transposable elements present in the genome of Nostoc [56]. Taken together, these observations underscore the strong selective pressure to limit NblA levels and highlight the physiological relevance of antisense-mediated control.

Antisense regulation mediated by RNA duplex degradation by RNase III has been previously described in Nostoc [15, 16]. When as_glnA or as_gltA were overexpressed in trans, almost complete degradation of their sense mRNA partners took place in the wild-type background but not in the RNase III mutant background [15, 16]. However, using the same experimental approach here, we show that the overexpression of as_nblA in trans did not reduce the amount of nblA mRNA (Supplementary Fig. S7). Furthermore, the absence of as_nblA did not significantly affect the half-life of nblA mRNA (Fig. 4C), suggesting that duplex degradation is not the major regulatory mechanism operated by as_nblA. Therefore, we explored alternative mechanisms. A recent study highlighted the role of TI in several bacteria, including the unicellular cyanobacterium Synechocystis sp. PCC 6803, and the authors developed an informatic pipeline for the quantitative modeling using RNA-seq of rifampicin time-series [23]. We have therefore generated a dataset from cells treated with rifampicin and applied this same procedure to its analysis. Our data support TI as the main mode of action of as_nblA, by limiting readthrough transcription from upstream regions (Fig. 4).

TI occurs when transcription from opposing or tandem promoters disrupts gene expression through mechanisms such as promoter occlusion, displacement of paused RNAPs or transcription factors, or physical collisions between elongating complexes [47, 48]. The exact molecular mechanisms, e.g. transcription–translation coupling and impact on DNA supercoiling, are still under investigation [57]. The regulatory potential of TI was first implicated by high-throughput transcriptomic studies revealing pervasive antisense transcription in both prokaryotes [17, 58, 59] and eukaryotes [60]. TI modulates, for instance, rifampicin sensitivity in Bacillus subtilis [61]. A main obstacle to large-scale TI investigation has been the absence of methods capable of distinguishing TI from other phenomena. The method used here [23] exploits a diagnostic kinetic signature unique to TI. Under steady-state conditions, antisense RNAPs continuously terminate a fraction of sense RNAPs within the sense/antisense overlap zone, depressing sense RNA levels downstream of the antisense TSS. Upon rifampicin addition, new antisense initiation is blocked, relieving antisense transcription-based termination and allowing the remaining sense RNAPs to read through the former interference zone. This results in a paradoxical post-rifampicin-treatment increase in sense RNA at positions downstream of the antisense TSS. Crucially, this method detects this signal in bulk transcriptomic data from living cells without genetic manipulation of the locus, making it applicable at genome-wide scale. Additionally, our data reveal a strongly asymmetric termination probability between sense and antisense RNAPs, which was proposed also in another study [62], consistent with a role for transcription–translation coupling in determining collision outcomes.

We have modeled the accumulation of nblA in response to nitrogen limitation (Fig. 5) and the model suggests that there is a delay in the accumulation of full-length nblA mRNA upon nitrogen removal, and this delay is dependent on the presence of as_nblA. Via the TI mechanism, the asRNA establishes a nearly perfectly tight threshold linear response [63], which is difficult to achieve by mechanisms that require RNA–RNA interaction. This mechanism establishes a threshold response, such that nblA transcripts accumulate only when transcriptional activation exceeds a critical level. Furthermore, by incorporating previous data on NsrR1 in the model [10, 26], we conclude that NsrR1 also plays a crucial role in controlling nblA mRNA amount (Fig. 5C). Together, as_nblA and NsrR1 (to a lower extent) fully repress nblA synthesis in the presence of combined nitrogen (Fig. 5C), which explains the strong phenotype of Δas_nblA observed in this work and the weaker phenotype of ΔNsrR1 strains [10] in this condition. In addition, this regulatory architecture also supports cell-type specificity within filaments. The model shows that the system allows that a roughly 3-fold synthesis rate difference between vegetative cells and heterocysts is amplified to a >11-fold change in the maximal nblA RNA accumulation (Fig. 5C). By introducing nonlinearity into the response, antisense transcription amplifies relatively small differences in transcriptional activity, contributing to the pronounced accumulation of NblA in heterocysts compared to vegetative cells. That allows a binary situation with high concentrations of NblA in heterocysts and low concentrations in vegetative cells as observed in Tolypothrix sp. [55].

In summary, the regulatory system we describe here operates by the transcription of an asRNA along with a trans-acting sRNA (Fig. 6). The double regulation of nblA expression by as_nblA and NsrR1 highlights the importance of additional regulatory elements that contribute to the suppression of background expression from transcriptionally regulated genes that, under non-inducing conditions, may be expressed due to leaky regulation of the promoter or, as in the case of nblA, because of readthrough upstream transcription. The regulatory mechanism operated by as_nblA seems conserved in heterocyst-forming cyanobacteria since the analysis of available transcriptomic data for N. punctiforme or Nodularia shows that also in these cyanobacteria there is a non-coding antisense transcript to nblA like as_nblA analyzed in this work (Supplementary Fig. S1). Interestingly, available data indicate that the accumulation of nblA and as_nblA exhibits opposite dynamics in response to light availability (Fig. 1 and Supplementary Fig. S1). Such inverse dynamics in the expression of two overlapping transcripts produced in opposite orientation (the so called excludon arrangement) has been observed in several bacteria [58]. Furthermore, in the unicellular strain Synechocystis sp. PCC 6803 (Supplementary Fig. S10), there is also antisense transcription to nblA, produced by overlapping transcription of an mRNA corresponding to a gene located downstream of nblA and transcribed in the opposite orientation [64, 65].

In conclusion, nblA expression is controlled by a multilayered regulatory system in which antisense transcription and a trans-acting sRNA cooperate to suppress basal expression and establish a threshold for induction. This combination enables precise control of a protein whose activity is beneficial under stress but deleterious when misregulated, illustrating the broader importance of antisense-mediated mechanisms in fine-tuning bacterial gene expression.

Supplementary Material

gkag575_Supplemental_File

gkag575_supplemental_file.pdf^{(1MB, pdf)}

Acknowledgements

We are grateful to Professor Wolfgang R. Hess (University of Freiburg) for sharing the raw data used to elaborate Supplementary Fig. S10 and Alicia Orea (IBVF, CSIC-Universidad de Sevilla) for help with microscopy.

Author contributions: Isidro Álvarez-Escribano (Investigation, Visualization, Writing – review & editing), Manuel Brenes-Álvarez (Investigation, Visualization, Writing – review & editing), Jens Georg (Formal analysis, Methodology, Software, Visualization, Funding acquisition, Writing – review & editing), Agustín Vioque (Conceptualization, Investigation, Visualization, Writing – review & editing), and Alicia M. Muro-Pastor (Conceptualization, Funding acquisition, Investigation, Project administration, Visualization, Writing – original draft, Writing – review & editing)

Contributor Information

Isidro Álvarez-Escribano, Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, E-41092 Sevilla, Spain.

Manuel Brenes-Álvarez, Genetics and Experimental Bioinformatics, Faculty of Biology, University of Freiburg, D-79104 Freiburg, Germany.

Jens Georg, Genetics and Experimental Bioinformatics, Faculty of Biology, University of Freiburg, D-79104 Freiburg, Germany.

Agustín Vioque, Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, E-41092 Sevilla, Spain.

Alicia M Muro-Pastor, Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas and Universidad de Sevilla, E-41092 Sevilla, Spain.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was supported by Agencia Estatal de Investigación, Spain (MCIN/AEI/10.13039/501100011033, FEDER, UE) (PID2019-105526GB-I00, PID2022-138128NB-I00 to A.M.M.-P.), and Deutsche Forschungsgemeinschaft (530000184 to J.G.). Funding to pay the Open Access publication charges for this article was provided by Consejo Superior de Investigaciones Científicas (CSIC), Spain.

Data availability

Raw RNA-seq data can be accessed in the GEO database under accession number GSE299220. The fitted model R object and the scripts are available at GitHub (https://github.com/JensGeorg/Bayesian-modeling-of-nblA-regulation.git) and Zenodo (https://doi.org/10.5281/zenodo.19221631).

References

1. Collier JL, Grossman AR. A small polypeptide triggers complete degradation of light-harvesting PBP in nutrient-deprived cyanobacteria. Embo J. 1994;13:1039–47. 10.1002/j.1460-2075.1994.tb06352.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
2. Kato H, Kubo T, Hayashi M et al. Interactions between histidine kinase NblS and the response regulators RpaB and SrrA are involved in the bleaching process of the cyanobacterium Synechococcus elongatus PCC 7942. Plant Cell Physiol. 2011;52:2115–22. 10.1093/pcp/pcr140. [DOI] [PubMed] [Google Scholar]
3. Bienert R, Baier K, Volkmer R et al. Crystal structure of NblA from Anabaena sp. PCC 7120, a small protein playing a key role in phycobilisome degradation. J Biol Chem. 2006;281:5216–23. 10.1074/jbc.M507243200. [DOI] [PubMed] [Google Scholar]
4. Karradt A, Sobanski J, Mattow J et al. NblA, a key protein of phycobilisome degradation, interacts with ClpC, a HSP100 chaperone partner of a cyanobacterial Clp protease. J Biol Chem. 2008;283:32394–403. 10.1074/jbc.M805823200. [DOI] [PubMed] [Google Scholar]
5. Nguyen AY, Bricker WP, Zhang H et al. The proteolysis adaptor, NblA, binds to the N-terminus of beta-phycocyanin: implications for the mechanism of phycobilisome degradation. Photosynth Res. 2017;132:95–106. 10.1007/s11120-016-0334-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Sendersky E, Kozer N, Levi M et al. The proteolysis adaptor, NblA, initiates protein pigment degradation by interacting with the cyanobacterial light-harvesting complexes. Plant J. 2014;79:118–26. 10.1111/tpj.12543. [DOI] [PubMed] [Google Scholar]
7. Nadel O, Hanna R, Rozenberg A et al. Viral NblA proteins negatively affect oceanic cyanobacterial photosynthesis. Nature. 2025;648:434–42. 10.1038/s41586-025-09656-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Baier K, Lehmann H, Stephan DP et al. NblA is essential for phycobilisome degradation in Anabaena sp. strain PCC 7120 but not for development of functional heterocysts. Microbiology. 2004;150:2739–49. 10.1099/mic.0.27153-0. [DOI] [PubMed] [Google Scholar]
9. Mitschke J, Vioque A, Haas F et al. Dynamics of transcriptional start site selection during nitrogen stress-induced cell differentiation in Anabaena sp. PCC7120. Proc Natl Acad Sci USA. 2011;108:20130–5. 10.1073/pnas.1112724108. [DOI] [PMC free article] [PubMed] [Google Scholar]
10. Álvarez-Escribano I, Vioque A, Muro-Pastor AM. NsrR1, a nitrogen stress-repressed sRNA, contributes to the regulation of nblA in Nostoc sp. PCC 7120. Front Microbiol. 2018;9:2267. 10.3389/fmicb.2018.02267. [DOI] [PMC free article] [PubMed] [Google Scholar]
11. Riediger M, Kadowaki T, Nagayama R et al. Biocomputational analyses and experimental validation identify the regulon controlled by the redox-responsive transcription factor RpaB. iScience. 2019;15:316–31. 10.1016/j.isci.2019.04.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
12. Georg J, Hess WR. Widespread antisense transcription in prokaryotes. Microbiol Spectr. 2018;6. 10.1128/microbiolspec.RWR-0029-2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Flaherty BL, Van Nieuwerburgh F, Head SR et al. Directional RNA deep sequencing sheds new light on the transcriptional response of Anabaena sp. strain PCC 7120 to combined-nitrogen deprivation. BMC Genomics. 2011;12:332. 10.1186/1471-2164-12-332. [DOI] [PMC free article] [PubMed] [Google Scholar]
14. Brenes-Álvarez M, Vioque A, Muro-Pastor AM. Nitrogen-regulated antisense transcription in the adaptation to nitrogen deficiency in Nostoc sp. PCC 7120. PNAS Nexus. 2023;2:pgad187. 10.1093/pnasnexus/pgad187. [DOI] [PMC free article] [PubMed] [Google Scholar]
15. Olmedo-Verd E, Brenes-Álvarez M, Vioque A et al. A heterocyst-specific antisense RNA contributes to metabolic reprogramming in Nostoc sp. PCC 7120. Plant Cell Physiol. 2019;60:1646–55. 10.1093/pcp/pcz087. [DOI] [PubMed] [Google Scholar]
16. Álvarez-Escribano I, Suárez-Murillo B, Brenes-Álvarez M et al. Antisense RNA regulates glutamine synthetase in a heterocyst-forming cyanobacterium. Plant Physiol. 2024;195:2911–20. 10.1093/plphys/kiae263. [DOI] [PMC free article] [PubMed] [Google Scholar]
17. Georg J, Hess WR. cis-Antisense RNA, another level of gene regulation in bacteria. Microbiol Mol Biol Rev. 2011;75:286–300. 10.1128/MMBR.00032-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
18. Gatewood ML, Bralley P, Weil MR et al. RNA-seq and RNA immunoprecipitation analyses of the transcriptome of Streptomyces coelicolor identify substrates for RNase III. J Bacteriol. 2012;194:2228–37. 10.1128/JB.06541-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Lasa I, Toledo-Arana A, Dobin A et al. Genome-wide antisense transcription drives mRNA processing in bacteria. Proc Natl Acad Sci USA. 2011;108:20172–7. 10.1073/pnas.1113521108. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Lybecker M, Zimmermann B, Bilusic I et al. The double-stranded transcriptome of Escherichia coli. Proc Natl Acad Sci USA. 2014;111:3134–9. 10.1073/pnas.1315974111. [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Callen BP, Shearwin KE, Egan JB. Transcriptional interference between convergent promoters caused by elongation over the promoter. Mol Cell. 2004;14:647–56. 10.1016/j.molcel.2004.05.010. [DOI] [PubMed] [Google Scholar]
22. Chatterjee A, Johnson CM, Shu C-C et al. Convergent transcription confers a bistable switch in Enterococcus faecalis conjugation. Proc Natl Acad Sci USA. 2011;108:9721–6. 10.1073/pnas.1101569108. [DOI] [PMC free article] [PubMed] [Google Scholar]
23. Wanney WC, Youssar L, Kostova G et al. Improved RNA stability estimation indicates that transcriptional interference is frequent in diverse bacteria. Commun Biol. 2023;6:732. 10.1038/s42003-023-05097-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Rippka R, Deruelles J, Waterbury JB et al. Generic assignments, strain histories and properties of pure cultures of cyanobacteria. Microbiology. 1979;111:1–61. 10.1099/00221287-111-1-1. [DOI] [Google Scholar]
25. Sambrook JF, Russell DW. Molecular cloning: a laboratory manual. Cold Spring Harbor, N Y: Cold Spring Harbor Laboratory, 2001. [Google Scholar]
26. Brenes-Álvarez M, Mitschke J, Olmedo-Verd E et al. Elements of the heterocyst-specific transcriptome unravelled by co-expression analysis in Nostoc sp. PCC 7120. Environ Microbiol. 2019;21:2544–58. 10.1111/1462-2920.14647. [DOI] [PubMed] [Google Scholar]
27. Elhai J, Wolk CP. Conjugal transfer of DNA to cyanobacteria. Methods Enzymol. 1988;167:747–54. 10.1016/0076-6879(88)67086-8. [DOI] [PubMed] [Google Scholar]
28. Merino-Puerto V, Mariscal V, Mullineaux CW et al. Fra proteins influencing filament integrity, diazotrophy and localization of septal protein SepJ in the heterocyst-forming cyanobacterium Anabaena sp. Mol Microbiol. 2010;75:1159–70. 10.1111/j.1365-2958.2009.07031.x. [DOI] [PubMed] [Google Scholar]
29. Cai Y, Wolk CP. Use of a conditionally lethal gene in Anabaena sp. strain PCC 7120 to select for double recombinants and to entrap insertion sequences. J Bacteriol. 1990;172:3138–45. 10.1128/jb.172.6.3138-3145.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Mohamed A, Jansson C. Influence of light on accumulation of photosynthesis-specific transcripts in the cyanobacterium Synechocystis 6803. Plant Mol Biol. 1989;13:693–700. 10.1007/BF00016024. [DOI] [PubMed] [Google Scholar]
31. Brenes-Álvarez M, Olmedo-Verd E, Vioque A et al. Identification of conserved and potentially regulatory small RNAs in heterocystous cyanobacteria. Front Microbiol. 2016;7:48. 10.3389/fmicb.2016.00048. [DOI] [PMC free article] [PubMed] [Google Scholar]
32. Muro-Pastor AM, Valladares A, Flores E et al. The hetC gene is a direct target of the NtcA transcriptional regulator in cyanobacterial heterocyst development. J Bacteriol. 1999;181:6664–9. 10.1128/JB.181.21.6664-6669.1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
33. Steglich C, Futschik ME, Lindell D et al. The challenge of regulation in a minimal photoautotroph: non-coding RNAs in Prochlorococcus. PLoS Genet. 2008;4:e1000173. 10.1371/journal.pgen.1000173. [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Vioque A. Analysis of the gene encoding the RNA subunit of ribonuclease P from cyanobacteria. Nucleic Acids Res. 1992;20:6331–7. 10.1093/nar/20.23.6331. [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Ionescu D, Voβ B, Oren A et al. Heterocyst-specific transcription of NsiR1, a non-coding RNA encoded in a tandem array of direct repeats in cyanobacteria. J Mol Biol. 2010;398:177–88. 10.1016/j.jmb.2010.03.010. [DOI] [PubMed] [Google Scholar]
36. Riediger M, Spät P, Bilger R et al. Analysis of a photosynthetic cyanobacterium rich in internal membrane systems via gradient profiling by sequencing (Grad-seq). Plant Cell. 2021;33:248–69. 10.1093/plcell/koaa017. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Gonzalez A, Riley KW, Harwood TV. et al. A tripartite, hierarchical sigma factor cascade promotes hormogonium development in the filamentous cyanobacterium Nostoc punctiforme. mSphere. 2019;4:e00231–19. 10.1128/mSphere.00231-19. [DOI] [PMC free article] [PubMed] [Google Scholar]
38. Harwood TV, Risser DD. The primary transcriptome of hormogonia from a filamentous cyanobacterium defined by cappable-seq. Microbiology. 2021;167:001111. 10.1099/mic.0.001111. [DOI] [PubMed] [Google Scholar]
39. Teikari JE, Hou S, Wahlsten M et al. Comparative genomics of the Baltic Sea toxic cyanobacteria Nodularia spumigena UHCC 0039 and its response to varying salinity. Front Microbiol. 2018;9:356. 10.3389/fmicb.2018.00356. [DOI] [PMC free article] [PubMed] [Google Scholar]
40. Kopf M, Möke F, Bauwe H et al. Expression profiling of the bloom-forming cyanobacterium Nodularia CCY9414 under light and oxidative stress conditions. ISME J. 2015;9:2139–52. 10.1038/ismej.2015.16. [DOI] [PMC free article] [PubMed] [Google Scholar]
41. Kim D, Paggi JM, Park C et al. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat Biotechnol. 2019;37:907–15. 10.1038/s41587-019-0201-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
42. Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv, 10.48550/arXiv.1303.3997, 16 Mar 2013, preprint; not peer reviewed. [DOI]
43. Li H, Handsaker B, Wysoker A et al. The sequence alignment/map format and SAMtools. Bioinformatics. 2009;25:2078–9. 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]
44. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30:923–30. 10.1093/bioinformatics/btt656. [DOI] [PubMed] [Google Scholar]
45. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550. 10.1186/s13059-014-0550-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
46. Segel LA, Slemrod M. The quasi-steady-state assumption: a case study in perturbation. SIAM Rev. 1989;31:446–77. 10.1137/1031091. [DOI] [Google Scholar]
47. Shearwin KE, Callen BP, Egan JB. Transcriptional interference—a crash course. Trends Genet. 2005;21:339–45. 10.1016/j.tig.2005.04.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
48. Sneppen K, Dodd IB, Shearwin KE et al. A mathematical model for transcriptional interference by RNA polymerase traffic in Escherichia coli. J Mol Biol. 2005;346:399–409. 10.1016/j.jmb.2004.11.075. [DOI] [PubMed] [Google Scholar]
49. Carpenter B, Gelman A, Hoffman MD et al. Stan: a probabilistic programming language. J Stat Soft. 2017;76:1–32. 10.18637/jss.v076.i01. [DOI] [PMC free article] [PubMed] [Google Scholar]
50. Gabry J, Češnovar R, Johnson A. cmdstanr: r interface to ’CmdStan’_. 10.1101/2022.05.06.490971. 2024. [DOI]
51. Muro-Pastor AM. The heterocyst-specific NsiR1 small RNA is an early marker of cell differentiation in cyanobacterial filaments. mBio. 2014;5:e01079–14. 10.1128/mBio.01079-14. [DOI] [PMC free article] [PubMed] [Google Scholar]
52. Schindelin J, Arganda-Carreras I, Frise E et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012;9:676–82. 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
53. Lowry OH, Rosebrough NJ, Farr AL et al. Protein measurement with the Folin phenol reagent. J Biol Chem. 1951;193:265–75. 10.1016/S0021-9258(19)52451-6. [DOI] [PubMed] [Google Scholar]
54. Chen H, Shiroguchi K, Ge H et al. Genome-wide study of mRNA degradation and transcript elongation in Escherichia col i. Mol Syst Biol. 2015;11:781. 10.15252/msb.20145794. [DOI] [PMC free article] [PubMed] [Google Scholar]
55. Ochoa de Alda JAG, Lichtlé C, Thomas JC et al. Immunolocalization of NblA, a protein involved in phycobilisome turnover, during heterocyst differentiation in cyanobacteria. Microbiology. 2004;150:1377–84. 10.1099/mic.0.26992-0. [DOI] [PubMed] [Google Scholar]
56. Kaneko T, Nakamura Y, Wolk CP et al. Complete genomic sequence of the filamentous nitrogen-fixing cyanobacterium Anabaena sp. strain PCC 7120. DNA Res. 2001;8:205–13.; 10.1093/dnares/8.5.205. [DOI] [PubMed] [Google Scholar]
57. O’Connor NJ, Bordoy AE, Chatterjee A. Engineering transcriptional interference through RNA polymerase processivity control. ACS Synth Biol. 2021;10:737–48. 10.1021/acssynbio.0c00534. [DOI] [PubMed] [Google Scholar]
58. Sanmartín Á, Iturbe P, Rodríguez-Beltrán J et al. ExcludonFinder: mapping transcriptional overlaps between neighboring genes. Nucleic Acids Res. 2025;53:gkaf686. 10.1093/nar/gkaf686. [DOI] [PMC free article] [PubMed] [Google Scholar]
59. Sesto N, Wurtzel O, Archambaud C et al. The excludon: a new concept in bacterial antisense RNA-mediated gene regulation. Nat Rev Microbiol. 2013;11:75–82. 10.1038/nrmicro2934. [DOI] [PubMed] [Google Scholar]
60. Fischer M, Hoffmann S. Transcriptional interference revisited. Nat Genet. 2026;58:974–83. 10.1038/s41588-026-02536-8. [DOI] [PubMed] [Google Scholar]
61. Sudzinová P, Knežová Balgová T, Schwarz M et al. Bacteria sense the antibiotic rifampicin through a widespread dual-promoter based alarm system. Nucleic Acids Res. 2026;54:gkaf1407. 10.1093/nar/gkaf1407. [DOI] [PMC free article] [PubMed] [Google Scholar]
62. Hoffmann SA, Hao N, Shearwin KE et al. Characterizing transcriptional interference between converging genes in bacteria. ACS Synth Biol. 2019;8:466–73. 10.1021/acssynbio.8b00477. [DOI] [PubMed] [Google Scholar]
63. Levine E, Hwa T. Small RNAs establish gene expression thresholds. Curr Opin Microbiol. 2008;11:574–9. 10.1016/j.mib.2008.09.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
64. Kopf M, Klähn S, Pade N et al. Comparative genome analysis of the closely related Synechocystis strains PCC 6714 and PCC 6803. DNA Res. 2014;21:255–66. 10.1093/dnares/dst055. [DOI] [PMC free article] [PubMed] [Google Scholar]
65. Kopf M, Klähn S, Scholz I et al. Comparative analysis of the primary transcriptome of Synechocystis sp. PCC 6803. DNA Res. 2014;21:527–39. 10.1093/dnares/dsu018. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gkag575_Supplemental_File

gkag575_supplemental_file.pdf^{(1MB, pdf)}

Data Availability Statement

[B1] 1. Collier JL, Grossman AR. A small polypeptide triggers complete degradation of light-harvesting PBP in nutrient-deprived cyanobacteria. Embo J. 1994;13:1039–47. 10.1002/j.1460-2075.1994.tb06352.x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B2] 2. Kato H, Kubo T, Hayashi M et al. Interactions between histidine kinase NblS and the response regulators RpaB and SrrA are involved in the bleaching process of the cyanobacterium Synechococcus elongatus PCC 7942. Plant Cell Physiol. 2011;52:2115–22. 10.1093/pcp/pcr140. [DOI] [PubMed] [Google Scholar]

[B3] 3. Bienert R, Baier K, Volkmer R et al. Crystal structure of NblA from Anabaena sp. PCC 7120, a small protein playing a key role in phycobilisome degradation. J Biol Chem. 2006;281:5216–23. 10.1074/jbc.M507243200. [DOI] [PubMed] [Google Scholar]

[B4] 4. Karradt A, Sobanski J, Mattow J et al. NblA, a key protein of phycobilisome degradation, interacts with ClpC, a HSP100 chaperone partner of a cyanobacterial Clp protease. J Biol Chem. 2008;283:32394–403. 10.1074/jbc.M805823200. [DOI] [PubMed] [Google Scholar]

[B5] 5. Nguyen AY, Bricker WP, Zhang H et al. The proteolysis adaptor, NblA, binds to the N-terminus of beta-phycocyanin: implications for the mechanism of phycobilisome degradation. Photosynth Res. 2017;132:95–106. 10.1007/s11120-016-0334-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B6] 6. Sendersky E, Kozer N, Levi M et al. The proteolysis adaptor, NblA, initiates protein pigment degradation by interacting with the cyanobacterial light-harvesting complexes. Plant J. 2014;79:118–26. 10.1111/tpj.12543. [DOI] [PubMed] [Google Scholar]

[B7] 7. Nadel O, Hanna R, Rozenberg A et al. Viral NblA proteins negatively affect oceanic cyanobacterial photosynthesis. Nature. 2025;648:434–42. 10.1038/s41586-025-09656-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B8] 8. Baier K, Lehmann H, Stephan DP et al. NblA is essential for phycobilisome degradation in Anabaena sp. strain PCC 7120 but not for development of functional heterocysts. Microbiology. 2004;150:2739–49. 10.1099/mic.0.27153-0. [DOI] [PubMed] [Google Scholar]

[B9] 9. Mitschke J, Vioque A, Haas F et al. Dynamics of transcriptional start site selection during nitrogen stress-induced cell differentiation in Anabaena sp. PCC7120. Proc Natl Acad Sci USA. 2011;108:20130–5. 10.1073/pnas.1112724108. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B10] 10. Álvarez-Escribano I, Vioque A, Muro-Pastor AM. NsrR1, a nitrogen stress-repressed sRNA, contributes to the regulation of nblA in Nostoc sp. PCC 7120. Front Microbiol. 2018;9:2267. 10.3389/fmicb.2018.02267. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B11] 11. Riediger M, Kadowaki T, Nagayama R et al. Biocomputational analyses and experimental validation identify the regulon controlled by the redox-responsive transcription factor RpaB. iScience. 2019;15:316–31. 10.1016/j.isci.2019.04.033. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B12] 12. Georg J, Hess WR. Widespread antisense transcription in prokaryotes. Microbiol Spectr. 2018;6. 10.1128/microbiolspec.RWR-0029-2018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B13] 13. Flaherty BL, Van Nieuwerburgh F, Head SR et al. Directional RNA deep sequencing sheds new light on the transcriptional response of Anabaena sp. strain PCC 7120 to combined-nitrogen deprivation. BMC Genomics. 2011;12:332. 10.1186/1471-2164-12-332. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B14] 14. Brenes-Álvarez M, Vioque A, Muro-Pastor AM. Nitrogen-regulated antisense transcription in the adaptation to nitrogen deficiency in Nostoc sp. PCC 7120. PNAS Nexus. 2023;2:pgad187. 10.1093/pnasnexus/pgad187. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B15] 15. Olmedo-Verd E, Brenes-Álvarez M, Vioque A et al. A heterocyst-specific antisense RNA contributes to metabolic reprogramming in Nostoc sp. PCC 7120. Plant Cell Physiol. 2019;60:1646–55. 10.1093/pcp/pcz087. [DOI] [PubMed] [Google Scholar]

[B16] 16. Álvarez-Escribano I, Suárez-Murillo B, Brenes-Álvarez M et al. Antisense RNA regulates glutamine synthetase in a heterocyst-forming cyanobacterium. Plant Physiol. 2024;195:2911–20. 10.1093/plphys/kiae263. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B17] 17. Georg J, Hess WR. cis-Antisense RNA, another level of gene regulation in bacteria. Microbiol Mol Biol Rev. 2011;75:286–300. 10.1128/MMBR.00032-10. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B18] 18. Gatewood ML, Bralley P, Weil MR et al. RNA-seq and RNA immunoprecipitation analyses of the transcriptome of Streptomyces coelicolor identify substrates for RNase III. J Bacteriol. 2012;194:2228–37. 10.1128/JB.06541-11. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B19] 19. Lasa I, Toledo-Arana A, Dobin A et al. Genome-wide antisense transcription drives mRNA processing in bacteria. Proc Natl Acad Sci USA. 2011;108:20172–7. 10.1073/pnas.1113521108. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B20] 20. Lybecker M, Zimmermann B, Bilusic I et al. The double-stranded transcriptome of Escherichia coli. Proc Natl Acad Sci USA. 2014;111:3134–9. 10.1073/pnas.1315974111. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B21] 21. Callen BP, Shearwin KE, Egan JB. Transcriptional interference between convergent promoters caused by elongation over the promoter. Mol Cell. 2004;14:647–56. 10.1016/j.molcel.2004.05.010. [DOI] [PubMed] [Google Scholar]

[B22] 22. Chatterjee A, Johnson CM, Shu C-C et al. Convergent transcription confers a bistable switch in Enterococcus faecalis conjugation. Proc Natl Acad Sci USA. 2011;108:9721–6. 10.1073/pnas.1101569108. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B23] 23. Wanney WC, Youssar L, Kostova G et al. Improved RNA stability estimation indicates that transcriptional interference is frequent in diverse bacteria. Commun Biol. 2023;6:732. 10.1038/s42003-023-05097-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B24] 24. Rippka R, Deruelles J, Waterbury JB et al. Generic assignments, strain histories and properties of pure cultures of cyanobacteria. Microbiology. 1979;111:1–61. 10.1099/00221287-111-1-1. [DOI] [Google Scholar]

[B25] 25. Sambrook JF, Russell DW. Molecular cloning: a laboratory manual. Cold Spring Harbor, N Y: Cold Spring Harbor Laboratory, 2001. [Google Scholar]

[B26] 26. Brenes-Álvarez M, Mitschke J, Olmedo-Verd E et al. Elements of the heterocyst-specific transcriptome unravelled by co-expression analysis in Nostoc sp. PCC 7120. Environ Microbiol. 2019;21:2544–58. 10.1111/1462-2920.14647. [DOI] [PubMed] [Google Scholar]

[B27] 27. Elhai J, Wolk CP. Conjugal transfer of DNA to cyanobacteria. Methods Enzymol. 1988;167:747–54. 10.1016/0076-6879(88)67086-8. [DOI] [PubMed] [Google Scholar]

[B28] 28. Merino-Puerto V, Mariscal V, Mullineaux CW et al. Fra proteins influencing filament integrity, diazotrophy and localization of septal protein SepJ in the heterocyst-forming cyanobacterium Anabaena sp. Mol Microbiol. 2010;75:1159–70. 10.1111/j.1365-2958.2009.07031.x. [DOI] [PubMed] [Google Scholar]

[B29] 29. Cai Y, Wolk CP. Use of a conditionally lethal gene in Anabaena sp. strain PCC 7120 to select for double recombinants and to entrap insertion sequences. J Bacteriol. 1990;172:3138–45. 10.1128/jb.172.6.3138-3145.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B30] 30. Mohamed A, Jansson C. Influence of light on accumulation of photosynthesis-specific transcripts in the cyanobacterium Synechocystis 6803. Plant Mol Biol. 1989;13:693–700. 10.1007/BF00016024. [DOI] [PubMed] [Google Scholar]

[B31] 31. Brenes-Álvarez M, Olmedo-Verd E, Vioque A et al. Identification of conserved and potentially regulatory small RNAs in heterocystous cyanobacteria. Front Microbiol. 2016;7:48. 10.3389/fmicb.2016.00048. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B32] 32. Muro-Pastor AM, Valladares A, Flores E et al. The hetC gene is a direct target of the NtcA transcriptional regulator in cyanobacterial heterocyst development. J Bacteriol. 1999;181:6664–9. 10.1128/JB.181.21.6664-6669.1999. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B33] 33. Steglich C, Futschik ME, Lindell D et al. The challenge of regulation in a minimal photoautotroph: non-coding RNAs in Prochlorococcus. PLoS Genet. 2008;4:e1000173. 10.1371/journal.pgen.1000173. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B34] 34. Vioque A. Analysis of the gene encoding the RNA subunit of ribonuclease P from cyanobacteria. Nucleic Acids Res. 1992;20:6331–7. 10.1093/nar/20.23.6331. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B35] 35. Ionescu D, Voβ B, Oren A et al. Heterocyst-specific transcription of NsiR1, a non-coding RNA encoded in a tandem array of direct repeats in cyanobacteria. J Mol Biol. 2010;398:177–88. 10.1016/j.jmb.2010.03.010. [DOI] [PubMed] [Google Scholar]

[B36] 36. Riediger M, Spät P, Bilger R et al. Analysis of a photosynthetic cyanobacterium rich in internal membrane systems via gradient profiling by sequencing (Grad-seq). Plant Cell. 2021;33:248–69. 10.1093/plcell/koaa017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B37] 37. Gonzalez A, Riley KW, Harwood TV. et al. A tripartite, hierarchical sigma factor cascade promotes hormogonium development in the filamentous cyanobacterium Nostoc punctiforme. mSphere. 2019;4:e00231–19. 10.1128/mSphere.00231-19. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B38] 38. Harwood TV, Risser DD. The primary transcriptome of hormogonia from a filamentous cyanobacterium defined by cappable-seq. Microbiology. 2021;167:001111. 10.1099/mic.0.001111. [DOI] [PubMed] [Google Scholar]

[B39] 39. Teikari JE, Hou S, Wahlsten M et al. Comparative genomics of the Baltic Sea toxic cyanobacteria Nodularia spumigena UHCC 0039 and its response to varying salinity. Front Microbiol. 2018;9:356. 10.3389/fmicb.2018.00356. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B40] 40. Kopf M, Möke F, Bauwe H et al. Expression profiling of the bloom-forming cyanobacterium Nodularia CCY9414 under light and oxidative stress conditions. ISME J. 2015;9:2139–52. 10.1038/ismej.2015.16. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B41] 41. Kim D, Paggi JM, Park C et al. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat Biotechnol. 2019;37:907–15. 10.1038/s41587-019-0201-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B42] 42. Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv, 10.48550/arXiv.1303.3997, 16 Mar 2013, preprint; not peer reviewed. [DOI]

[B43] 43. Li H, Handsaker B, Wysoker A et al. The sequence alignment/map format and SAMtools. Bioinformatics. 2009;25:2078–9. 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B44] 44. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30:923–30. 10.1093/bioinformatics/btt656. [DOI] [PubMed] [Google Scholar]

[B45] 45. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550. 10.1186/s13059-014-0550-8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B46] 46. Segel LA, Slemrod M. The quasi-steady-state assumption: a case study in perturbation. SIAM Rev. 1989;31:446–77. 10.1137/1031091. [DOI] [Google Scholar]

[B47] 47. Shearwin KE, Callen BP, Egan JB. Transcriptional interference—a crash course. Trends Genet. 2005;21:339–45. 10.1016/j.tig.2005.04.009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B48] 48. Sneppen K, Dodd IB, Shearwin KE et al. A mathematical model for transcriptional interference by RNA polymerase traffic in Escherichia coli. J Mol Biol. 2005;346:399–409. 10.1016/j.jmb.2004.11.075. [DOI] [PubMed] [Google Scholar]

[B49] 49. Carpenter B, Gelman A, Hoffman MD et al. Stan: a probabilistic programming language. J Stat Soft. 2017;76:1–32. 10.18637/jss.v076.i01. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B50] 50. Gabry J, Češnovar R, Johnson A. cmdstanr: r interface to ’CmdStan’_. 10.1101/2022.05.06.490971. 2024. [DOI]

[B51] 51. Muro-Pastor AM. The heterocyst-specific NsiR1 small RNA is an early marker of cell differentiation in cyanobacterial filaments. mBio. 2014;5:e01079–14. 10.1128/mBio.01079-14. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B52] 52. Schindelin J, Arganda-Carreras I, Frise E et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012;9:676–82. 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B53] 53. Lowry OH, Rosebrough NJ, Farr AL et al. Protein measurement with the Folin phenol reagent. J Biol Chem. 1951;193:265–75. 10.1016/S0021-9258(19)52451-6. [DOI] [PubMed] [Google Scholar]

[B54] 54. Chen H, Shiroguchi K, Ge H et al. Genome-wide study of mRNA degradation and transcript elongation in Escherichia col i. Mol Syst Biol. 2015;11:781. 10.15252/msb.20145794. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B55] 55. Ochoa de Alda JAG, Lichtlé C, Thomas JC et al. Immunolocalization of NblA, a protein involved in phycobilisome turnover, during heterocyst differentiation in cyanobacteria. Microbiology. 2004;150:1377–84. 10.1099/mic.0.26992-0. [DOI] [PubMed] [Google Scholar]

[B56] 56. Kaneko T, Nakamura Y, Wolk CP et al. Complete genomic sequence of the filamentous nitrogen-fixing cyanobacterium Anabaena sp. strain PCC 7120. DNA Res. 2001;8:205–13.; 10.1093/dnares/8.5.205. [DOI] [PubMed] [Google Scholar]

[B57] 57. O’Connor NJ, Bordoy AE, Chatterjee A. Engineering transcriptional interference through RNA polymerase processivity control. ACS Synth Biol. 2021;10:737–48. 10.1021/acssynbio.0c00534. [DOI] [PubMed] [Google Scholar]

[B58] 58. Sanmartín Á, Iturbe P, Rodríguez-Beltrán J et al. ExcludonFinder: mapping transcriptional overlaps between neighboring genes. Nucleic Acids Res. 2025;53:gkaf686. 10.1093/nar/gkaf686. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B59] 59. Sesto N, Wurtzel O, Archambaud C et al. The excludon: a new concept in bacterial antisense RNA-mediated gene regulation. Nat Rev Microbiol. 2013;11:75–82. 10.1038/nrmicro2934. [DOI] [PubMed] [Google Scholar]

[B60] 60. Fischer M, Hoffmann S. Transcriptional interference revisited. Nat Genet. 2026;58:974–83. 10.1038/s41588-026-02536-8. [DOI] [PubMed] [Google Scholar]

[B61] 61. Sudzinová P, Knežová Balgová T, Schwarz M et al. Bacteria sense the antibiotic rifampicin through a widespread dual-promoter based alarm system. Nucleic Acids Res. 2026;54:gkaf1407. 10.1093/nar/gkaf1407. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B62] 62. Hoffmann SA, Hao N, Shearwin KE et al. Characterizing transcriptional interference between converging genes in bacteria. ACS Synth Biol. 2019;8:466–73. 10.1021/acssynbio.8b00477. [DOI] [PubMed] [Google Scholar]

[B63] 63. Levine E, Hwa T. Small RNAs establish gene expression thresholds. Curr Opin Microbiol. 2008;11:574–9. 10.1016/j.mib.2008.09.016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B64] 64. Kopf M, Klähn S, Pade N et al. Comparative genome analysis of the closely related Synechocystis strains PCC 6714 and PCC 6803. DNA Res. 2014;21:255–66. 10.1093/dnares/dst055. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B65] 65. Kopf M, Klähn S, Scholz I et al. Comparative analysis of the primary transcriptome of Synechocystis sp. PCC 6803. DNA Res. 2014;21:527–39. 10.1093/dnares/dsu018. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Antisense RNA controls the degradation of phycobilisomes in heterocyst-forming cyanobacteria by a transcriptional interference mechanism

Isidro Álvarez-Escribano

Manuel Brenes-Álvarez

Jens Georg

Agustín Vioque

Alicia M Muro-Pastor

Roles

Abstract

Graphical Abstract

Graphical Abstract.

Introduction

Materials and methods

Strains and growth conditions

Figure 1.

Figure 4.

Construction of Nostoc derivative strains

Figure 3.

RNA isolation and analysis

RNA bioinformatic analysis

Bayesian ODE model of RNA dynamics

Differential equations

Condition-specific initial conditions

Estimated parameters

Fitting data sources and connection to model variables

Figure 5.

Figure 2.

Inference and convergence

Fluorescence microscopy

Cell fractionation and western blotting

Growth curve and whole cell spectra

Isolation and analysis of suppressors of the yellow phenotype of NblA overexpressing strains

Results

An antisense transcript to the nblA mRNA is produced in heterocyst-forming cyanobacteria

Expression of nblA and as_nblA in nitrogen-fixing filaments of Nostoc

Transcription of an antisense to nblA affects NblA accumulation and degradation of phycobiliproteins

The integrity of phycobilisomes is altered in strains lacking as_nblA

Accumulation of NblA in strains lacking as_nblA is deleterious

as_nblA regulates nblA expression through transcriptional interference

A mathematical model of nblA regulation

Discussion

Figure 6.

Supplementary Material

Acknowledgements

Contributor Information

Supplementary data

Conflict of interest

Funding

Data availability

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases