Some cyanobacteria, such as marine Prochlorococcus and Synechococcus, are of paramount importance as primary producers and for the global biogeochemical carbon (C) and nitrogen (N) cycles1,2,3. Other cyanobacteria, such as Synechococcus sp. PCC 7942 (Synechococcus 7942) or Synechocystis sp. PCC 6803 (Synechocystis 6803), have become model species for the CO2-neutral production of chemical feedstock and biotechnological processes driven by photosynthesis4,5. The two most relevant nutrients for cyanobacteria are inorganic carbon (Ci) and N. Most cyanobacteria assimilate combined inorganic N forms, such as ammonium (NH4+), nitrate (NO3), nitrite (NO2), and urea; in addition, diazotrophic species can utilize N2 gas.

Surprisingly, several different cyanobacteria excrete nitrite under photoautotrophic growth conditions. Certain isolates of the marine cyanobacterium Prochlorococcus were reported to release up to 30% of their N uptake as extracellular nitrite when grown on nitrate6. The released nitrite is cross-fed by other microbes that cannot consume nitrate or produce nitrite. Moreover, it was demonstrated that cooccurring microbes could be beneficial for Prochlorococcus7,8. In Synechococcus 7942, as well, approximately 30% of the produced nitrite were excreted in shifts to low Ci (LC) conditions, while no nitrite was excreted under high Ci (HC) conditions9.

The coordination of C and N metabolism is regulated by several transcription factors and other regulatory proteins. An acute N scarcity is sensed by two central regulators of N assimilation in cyanobacteria, PII and NtcA, by binding the key metabolite 2-oxoglutarate (2-OG)10,11,12,13,14. 2-OG connects C and N metabolism because it represents the main precursor for ammonia assimilation through the glutamine synthetase/glutamine oxoglutarate aminotransferase (GS/GOGAT) cycle in cyanobacteria and plants. Initially, GS incorporates ammonia into glutamate, thereby producing glutamine. The amino group from glutamine is then transferred onto 2-OG by GOGAT, yielding two glutamate molecules. As a consequence, the intracellular level of 2-OG starts to increase once these reactions slow down due to insufficient N supply; thus, 2-OG is an excellent indicator of the N status15. Binding of 2-OG stimulates the activity of NtcA, the main transcriptional regulator of genes encoding proteins involved in N assimilation10,11. Depending on the 2-OG level, the PII interaction protein X (PipX) switches from binding to PII to interacting with NtcA, further enhancing the binding affinity of this complex to target promoters12,14. In Synechocystis 6803, NtcA directly activates 51 genes, including transporters for N sources and GS/GOGAT, and represses 28 other genes after 4 h of N starvation16.

In addition to transcription factors, multiple different proteins have been identified that interact with these factors or the involved enzymes and govern their functions directly. Archetypical examples for these proteins are the inhibitory factors IF7 and IF17, as their binding to GS leads to the inactivation of the enzyme when N supply exceeds demand17. The PII-interacting regulator of arginine synthesis (PirA) regulates the flux into the ornithine-ammonia cycle and therefore arginine synthesis18,19, while the PII-interacting regulator of C metabolism (PirC) functions as an inhibitor of 2,3-phosphoglycerate–independent phosphoglycerate mutase (PGAM), the enzyme that shifts newly fixed CO2 toward lower glycolysis20. PirC becomes sequestered by PII at low 2-OG levels and released for PGAM inhibition at high 2-OG levels and therefore connects the control of C and N assimilatory pathways in a particular way20.

We have previously analyzed the primary transcriptomes of Synechocystis 6803 and the closely related strain Synechocystis sp. PCC 671421,22,23. Based on these analyses, many more genes encoding unknown small, potentially regulatory proteins were computationally predicted. Experiments using a 3xFLAG epitope tag fused in-frame to the 3′ ends of the respective reading frames validated five of these to encode small proteins24. Subsequent experimental analyses using coimmunoprecipitation (co-IP) demonstrated that one of these small proteins, NblD, facilitates the degradation of phycobilisomes under N starvation conditions25, while another one, called AtpΘ, is an inhibitor of the ATP synthase back reaction under low energy conditions26. Collectively, these findings suggest that cyanobacteria use a broad collection of small protein modulators to regulate metabolism according to external clues.

Here, we report a small, regulatory protein involved in the control of N assimilation in cyanobacteria in an unexpected way. This protein is 81 amino acids long in Synechocystis 6803, and its expression is upregulated following shifts from HC to LC conditions and downregulated upon N downshifts. Co-IP experiments identified the enzyme nitrite reductase (NiR) as the primary target of this protein. Following this lead further, we characterize the protein as a regulatory factor in the primary assimilation of N from nitrate/nitrite. Based on the observed phenotype, we named the regulatory protein NirP1, for nitrite reductase interacting protein 1.


The ncr1071 transcript exhibits coding potential

The NtcA regulon in Synechocystis 6803 was reported to comprise 79 genes, including eight non-coding RNAs (ncRNAs)16. One of these ncRNAs, with the transcript ID ncr1071, was strongly downregulated upon N step-down16. This transcript extends from position 2215954 to 2217039 on the forward strand of the Synechocystis 6803 chromosome (GenBank accession no. NC_000911) and was assigned to the transcriptional unit (TU) 229621. When 10 different cultivation conditions were compared, TU2296 was found to be maximally transcribed under LC, while it was strongly repressed under N starvation conditions21. TU2296 overlaps the coding sequence of the sll1864 gene encoding a potential chloride channel protein on the reverse strand by more than 500 nt (Fig. 1a).

Fig. 1: Genomic location and expression of nirP1 in Synechocystis 6803.
figure 1

a The transcriptional unit TU2296 encompassing the coding sequence of nirP1 (ncr1071) overlaps gene sll1864 annotated as encoding a chloride ion channel protein. Plot at low CO2 (blue) versus N starvation (green). Transcriptional units (TUs) are indicated according to the previous annotation of the transcriptome and genome-wide mapping of transcriptional start sites21. The potential open reading frame of ncr1071 coding for the NirP1 protein is indicated by a black arrow. b Alignment of selected potential NirP1 homologs from cyanobacteria belonging to four morphological subsections28. The last and most conserved part of Synechocystis 6803 NirP1 from position 25 to the end is shown. The cysteine and tryptophan residues conserved in all 485 potential homologs (Supplementary Dataset 1) are marked by arrows. The alignment was generated using ClustalW and visualized by Jalview. c Time course of nirp1 mRNA accumulation in Synechocystis 6803 after cells precultivated in medium supplemented with 10 mM NaHCO3 (HC) for 3 h were shifted to medium without a source of Ci (LC). The nirp1 transcript was detected by a 32P-labeled single-stranded RNA probe. The membrane was rehybridized to a 5S rRNA probe as a loading control. The RiboRuler Low Range RNA ladder (Thermo Fisher Scientific) was used as the molecular mass standard. The relative amount of nirp1 transcript normalized to the 5S rRNA is shown in the bar chart. The HC condition was set to 1, and all other signals were normalized to the HC condition. Two independent biological replicates were used and averaged. d NirP1 was detected in the presence or absence of Ci by Western blotting using anti-FLAG antiserum against tagged NirP1 under the control of its native promoter in two biological replicates (R1 and R2). The wild type was used as a control. Cultures were grown in medium supplemented with 10 mM NaHCO3 (+), washed, and cultivated in carbonate-free medium (-) for 24 h. Prestained PageRulerTM (Thermo Fisher Scientific) was used as a molecular mass marker. The blot has been performed in two independent experiments (n = 2). e NirP1 structure predicted by AlphaFold31,32 for the complete Synechocystis NirP1 protein, indicating the presence of four beta folds in the most conserved part. The four totally conserved amino acids as well as Cys27 and Cys81 are highlighted, tryptophan is shown in magenta, and cysteine in yellow. The Met1 residue is also indicated for orientation. Source data are provided as a Source Data file.

Upon closer inspection, we discovered that the first segment of this transcript exhibits potential to encode an 81-residue protein with a calculated IEP of 4.59 and a molecular mass of 9.18 kDa. Evidence for the possible existence of this protein was provided independently by mass spectrometry data from a deep proteogenomics dataset27.

Based on the results below, we named this protein NirP1, for nitrite reductase interacting protein 1. Database searches identified 485 potential homologs of nirP1 (Supplementary Dataset 1) in species belonging to all morphological subsections of cyanobacteria28, including the well-studied model Synechococcus 7942, heterocyst-forming diazotrophic species and Section V strains, such as Fischerella. Homologous genes were detected in marine species with deviating pigmentation, such as Acaryochloris marina and Halomicronema hongdechloris, but not in Prochlorococcus and not in the majority of marine picoplanktonic Synechococcus strains. Homologs were also not detected in the genomes of diatom-associated symbionts Calothrix rhizosoleniae and Richelia intracellularis nor in the endosymbiont chromatophore genomes of photosynthetic Paulinella species or the UCYN-A Candidatus Atelocyanobacterium thalassa endosymbionts, which can perform N2 fixation but lack photosystem II and phycobilisomes29,30. These data suggest that homologs of nirP1 were advantageous but not essential for free-living cyanobacteria with very different physiologies.

The vast majority of the 485 putative homologs identified in this work (Supplementary Dataset 1) are between 76 and 86 aa in length. The alignment of selected NirP1 homologs highlights the conservation in the last ~50 aa and the presence of a centrally located, absolutely conserved single cysteine and three tryptophan residues (Fig. 1b).

To verify the inducibility of nirP1 transcription by shifts to LC conditions, cultures of Synechocystis 6803 precultivated in medium supplemented with 10 mM NaHCO3 (high Ci, HC) were centrifuged and resuspended in medium lacking any source of Ci. In Northern blot analysis, samples taken over 24 h showed a single signal of approximately 580 nt (Fig. 1c) matching the length from the transcription start site (TSS) to the steep decline in the number of reads approximately 80 nt after the nirP1 stop codon (Fig. 1a). A strong signal was obtained 1 h after the shift, with a slightly decreasing intensity at the later time points and increasing again at the 24 h sampling point. At a weaker intensity, the nirP1 mRNA was also detectable in the sample from high Ci (HC)-grown cells.

To directly validate the presence of NirP1, strains expressing NirP1 −3xFLAG fusions were constructed (Supplementary Fig. S1). Protein extracts were prepared and analyzed by SDS–polyacrylamide gel electrophoresis (SDS‒PAGE) and Western blotting using anti-FLAG antiserum. Two signals of slightly different molecular masses were detected. The calculated molecular mass of ~12 kDa for the prominent signal in the samples from HC-grown cultures was consistent with the sum of the calculated molecular masses of 9.18 kDa for the monomeric NirP1 and 2.86 kDa for the 3xFLAG tag (Fig. 1d). After the shift to LC conditions, a band migrating with a ~1.5 kDa larger molecular mass became more intense, possibly indicating a post-translational modification. No FLAG-tagged NirP1 was detected in the wild-type strain used as a control. From these experiments, we conclude that TU2296 encodes the small protein NirP1 and that nirP1 transcription responds to shifts in Ci supply.

To generate a model for the NirP1 structure, we used AlphaFold31,32, which modeled NirP1 as a monomer and predicted four beta folds in the most conserved part of the protein. No structure was predicted for the N-terminal section of the protein, consistent with a lack of sequence conservation (Fig. 1b). Two cysteine residues (Cys27 and Cys81) in Synechocystis 6803 NirP1 were predicted to be in close proximity to each other, potentially forming a disulfide bond (Fig. 1e), whereas a third, highly conserved Cys55 was found between two beta strands in the folded part, exposed to the outside. However, no functional domains could be predicted based on the primary sequence or modeled structure.

Two transcription factors are involved in the regulation of nirP1 expression

Based on literature reports16, we speculated that nirP1 expression could be negatively regulated by NtcA binding to a conserved binding site over the TSS. Moreover, we observed a tandem repeat element located further upstream consisting of two octameric repeats TTTGT(T/C)AA separated by a dinucleotide. In promoter sequence analyses of 456 nirP1 homologs (Supplementary Dataset 2), the NtcA-binding motif was present and conserved in 435 cases. The tandem repeat was found in 135 homologs (Supplementary Dataset 2; Supplementary Fig. S2). To investigate the possible regulation through these sequence elements, the promoter of nirP1 (PnirP1; from −70 to +30, TSS at +1) was engineered in a transcriptional fusion to drive the luxAB reporter genes. We used the native PnirP1 sequence or promoter variants with mutated, likely relevant nucleotides either in the NtcA-binding site (PNtcA-Mut) or in the upstream promoter element (PRepeat-Mut; Fig. 2a).

Fig. 2: NirP1 expression is mediated through a Ci-sensitive promoter and the transcription factor NtcA.
figure 2

a Top: Native PnirP1 promoter sequence from position −70 to +30 (TSS at +1) used in a bioluminescence reporter strain harboring a PnirP1-luxAB transcriptional fusion. Functional promoter elements are colored blue. The previously determined21 transcription start site (TSS) is indicated. Middle row: Mutated nucleotides in the NtcA-binding site yielding promoter PNtcA-Mut are colored red. Bottom: Mutated nucleotides in the repeat motif yielding promoter PRepeat-Mut are colored red. Both mutated promoters were fused to luxAB and introduced into a neutral site in parallel with the PnirP1-luxAB construct. b Bioluminescence of the Synechocystis 6803 PnirP1-luxAB, PNtcA-Mut-luxAB reporter, and PRepeat-Mut-luxAB strains and a promoter-less (Pless) negative control. Cells were grown under HC conditions (BG11 supplemented with 10 mM NaHCO3) for 2 h and transferred to BG11 medium without a CO2 source to induce LC conditions for 24 h. c The strains were cultivated in BG11 (-N), and the chlorotic cultures were transferred back to standard BG11 (17.6 mM NaNO3) medium to start the recovery process seven days after N starvation was initiated. d Strains were cultivated in standard BG11 and then 10 mM NH4Cl was added for 24 h. Bioluminescence data are presented as the means ± SDs of 3 independent measurements with three biological replicates each (n = 9). Significance was calculated with a two-tailed t-test with unequal variance (Welch’s t-test; *P < 0.05; **P < 0.01; ***P < 0.001) between the strains at corresponding time points (details of statistical analysis Supplementary Fig. S3 and in Supplementary Dataset 3). Source data are provided as a Source Data file.

In strains PnirP1 and PNtcA-Mut, transcription was activated 1 h after a shift to medium lacking Ci, followed by a decrease and a light increase again at the later time points (Fig. 2b, Supplementary Fig. S3). The increase at 24 h was not as pronounced as in the Northern blot (Fig. 1c), pointing at the possible involvement of post-transcriptional regulation. This activation was significantly higher in the PNtcA-Mut strains with the mutated NtcA-binding site. This effect was even more pronounced after a shift to N starvation conditions (Fig. 2c; Supplementary Dataset 3). These results indicated the repressor function of NtcA for nirP1, consistent with previous ChIP-seq16 and transcriptome data21. However, repression was still observed 6 h after the shift to N starvation, even when the binding motif was mutated rendering NtcA binding impossible, indicating the presence of additional regulatory mechanisms. It is possible that a second transcription factor is involved in the repression of nirP1 during N starvation.

If the tandem repeat was mutated (PRepeat-Mut), a signal close to baseline was detected in all tested conditions (Fig. 2b–d, Supplementary Dataset 3; Supplementary Fig. S3). We conclude that this motif is essential for the transcription of nirP1, consistent with its position centered at −51.5 from the TSS at +1 (Fig. 2a), a region where frequently transcription factors bind to activate transcription.

Finally, transcriptional regulation was tested in response to the addition of ammonium (Fig. 2d). The signal increased ~threefold after 1 h and then decreased, similar to the results obtained after the shift in LC conditions (Fig. 2b). These data agree with the results of a recent transcriptome analysis in which the ncr1071 transcript was upregulated ~threefold during acclimation to ammonium stress33, indicating that the C/N balance is pivotal for the transcription of nirP1.

Collectively, these results demonstrated the complex transcriptional regulation of nirP1. We validated the transcriptional regulation of nirP1 by NtcA and obtained new insight into the regulation of nirP1 in shifts to N deprivation and with different C/N ratios. Due to the observed strong regulation, we hypothesized that NirP1 might be a regulator and is likely involved in the acclimation to changing Ci and N supply.

Deletion and ectopic expression of NirP1 leads to phenotypical changes

To test the potential functional importance of NirP1, we constructed the following strains: a gene deletion strain ΔnirP1; two overexpression strains in which the coding sequence of nirP1 was fused to a C-terminal 3xFLAG tag under control of the copper-inducible PpetE promoter on a plasmid pVZ322 introduced in a wild type background (strain NirP1oex) and in the ΔnirP1 background (strain ΔnirP1::oex); and a complementation strain with the same plasmid as in the overexpression strains, except that nirP1 was controlled by its native promoter (ΔnirP1::nirP1).

Under nutrient-replete conditions, the ΔnirP1 deletion mutant showed linear growth similar to the wild type. In contrast, both strains expressing nirP1 under control of the PpetE promoter showed severely reduced growth in liquid medium, irrespective of whether the plasmid was in the wild type or the ΔnirP1 background (Fig. 3a). The reduced growth observed at high nirP1 expression compared to wild type and ΔnirP1 was also visible under LC, while the strain grew normally if nirP1 was controlled by its native PnirP1 promoter (Fig. 3b). Therefore, we concluded that the high expression of nirP1 from the PpetE promoter led to reduced growth, in particular, to a prolonged lag phase (Supplementary Dataset 4). In addition, we noticed a pigmentation change, i.e., the cultures became chlorotic if nirP1 was overexpressed from the PpetE promoter (Fig. 3c). This effect was strictly dependent on the nirP1 expression level because cells of the non-induced mutant showed the same pigmentation as the wild type in the absence of added Cu2SO4. (Supplementary Fig. S4a). This phenotype was verified in room temperature absorption spectra. Compared to the spectra of the wild type and deletion mutant, the phycobilisome peak at approximately 625 nm as well as the chlorophyll peaks at 440 and 680 nm were decreased in the two strains that expressed nirP1 from PpetE after addition of Cu2SO4 (Fig. 3d, Supplementary Dataset 4).

Fig. 3: Phenotypical differences between nirP1 mutant strains and the wild type in BG11 medium.
figure 3

a Growth of wild type (black), the ΔnirP1 deletion mutant (red), and the strains expressing nirP1 under control of the PpetE promoter in the wild-type background (dark blue, NirP1oex) and ΔnirP1 background (light blue, ΔnirP1::oex). Data points represent the mean ± SD of 3 biological replicates (n = 3; details with growth rates and doubling time are provided in Supplementary Dataset 4). b Growth in liquid medium after shifting to LC conditions. Strains were cultivated for 2 h in medium containing 10 mM NaHCO3, collected, and transferred to medium without a Ci source. The complementation strain (orange, ΔnirP1::nirP1) expressed nirP1 under the control of its native PnirP1 promoter in the ΔnirP1 background. Other strains as in (a); n = 3. c Pigmentation phenotype of wild-type and nirP1 overexpressors in the presence of Cu2SO4 expressing nirP1 for 24 h. All cultures were set to an OD of ~0.8. d Room temperature absorption spectra for wild-type and nirP1 mutants expressing nirP1 for 24 h from the PpetE promoter in the presence of Cu2SO4, normalized to wild-type OD750. Same strains and colors as in (a). Spectra were recorded from three biological replicates and averaged (n = 3). Further phenotypical differences in Supplementary Figs. S4 and S5.

To extend these results, drop dilution assays were performed to compare growth on plates. Again, the strain that overexpressed nirP1 from the PpetE promoter showed reduced vitality compared to the wild type and ΔnirP1, irrespective of the addition of 10 mM NaHCO3 (Supplementary Fig. S4b). In conditions where nitrate was replaced by ammonium as the only N source, the phenotype was completely abolished (Supplementary Fig. S5). To summarize, the phenotypical effects due to ectopic NirP1 expression suggested that the amount of NirP1 played a critical role in the cell in presence of nitrate, while its absence did not lead to major phenotypic effects under the tested conditions.

Metabolic changes in the GS/GOGAT and ammonia-ornithine cycle

Upon nirP1 overexpression, the chlorotic phenotype (Fig. 3c, d) resembled the pigmentation changes characteristic of N-starved cells. Together with the observed expression control by NtcA (Fig. 2), this suggested a primary relationship between NirP1 and processes related to the assimilation of N, albeit the regulatory effect of LC.

Therefore, we tested whether the ectopic expression of NirP1 influences amino acid levels in the cell. Metabolites and amino acids were extracted in a time course experiment from wild type, ΔnirP1, and NirP1oex following a shift to LC. The total amount of soluble amino acids was significantly higher in NirP1oex at 24 h after transfer to LC than in the other two strains (Fig. 4a, full data in Supplementary Dataset 5). However, a closer look at the data revealed that the higher amino acid levels in NirP1oex were mainly caused by the accumulation of a single amino acid, glutamate (Fig. 4b). The glutamate level in Synechocystis 6803 is normally approximately 10 times higher than that of any other amino acid. Therefore, the increased accumulation of glutamate alone was responsible for the measured higher total amino acid amount in NirP1oex. However, for the direct product of glutamate amination by GS, glutamine, we measured lower levels in NirP1oex at 3 and 24 h after the shift to LC (Fig. 4b). This result is consistent with previous analyses in which a decrease in glutamine content was observed in LC shifts34. In addition, we found that the concentrations of citrulline and arginine, which are produced from glutamine via the ornithine-ammonium cycle (OAC), were lowered in NirP1oex at 3, 12, and 24 h after the induction of expression (Fig. 4d). In contrast, the aspartate levels were increased, consistent with its synthesis from glutamate by glutamate aspartate aminotransferase. These tendencies were confirmed in a repeat experiment (Supplementary Dataset 6). Furthermore, the concentration of 2-oxoglutarate (2-OG) was significantly higher in NirPoex than in the other strains (Fig. 4c). This finding indicates an impaired regulation of the GS/GOGAT cycle and a reduced N assimilation caused by high NirP1 and is consistent with effect the of pigmentation (Fig. 3c) and glutamine and glutamate concentration (Fig. 4b). In the wild type and ΔnirP1 deletion mutant, 2-OG levels increased after 12 and 24 h, which was more pronounced in the deletion mutant pointing to a further regulation when NirP1 is absent (Fig. 4c). We conclude that the upregulation of NirP1 caused several changes in amino acid levels related to the synchronization between C and N metabolism. As the OAC is involved in N storage and remobilization35, these results support that NirP1 plays a role as a regulator in N metabolism.

Fig. 4: Differences in the accumulation of key intermediates in C/N primary metabolism due to the availability of NirP1 during shifts to LC.
figure 4

a Total amino acid content. b Glutamate and glutamine concentrations. c 2-oxoglutarate concentrations. d Concentrations of ornithine-ammonium cycle intermediates and of aspartate. Metabolite content was measured for ΔnirP1, NirP1oex, and wild type (WT). All concentrations are given in ng * mL−1 * OD750 nm−1. Two biological replicates (n = 2) were used for all strains and metabolites (except in c where the means ± SD of three biological replicates are shown, n = 3) and averaged (details and raw data in Supplementary Dataset 5). Significance was calculated with the two-tailed t-test with unequal variance (Welch’s t-test; *only P ≤ 0.05 are shown) for each strain and between the strains at corresponding time points. The measurements were independently repeated in triplicates (n = 3) and the results confirmed (details in Supplementary Dataset 6).

NirP1 interacts with ferredoxin-nitrite reductase

Coimmunoprecipitation (co-IP) was successfully used in the past to identify the interacting partners of Synechocystis 6803 small proteins NblD, AtpΘ, and SliP4, thereby providing essential hints for their specific functions25,26,36. Therefore, we used a strain-expressing epitope-tagged NirP1 to obtain insight into the molecular mechanism that involves NirP1. After 24 h of induction, total cell extracts of three replicates of NirP1oex were analyzed by SDS‒PAGE and Western blotting using anti-FLAG antisera. The size of the NirP1-3xFLAG signal in the Western blot matched the sum of the calculated molecular masses of 9.81 kDa for NirP1 and 2.86 for the 3xFLAG tag. No signal was detected in the wild type used as a negative control (Fig. 5a).

Fig. 5: NirP1 expression and pull-down analysis.
figure 5

a Detection of NirP1 after 24 h of induction with 2 µM Cu2SO4 by Western blotting using anti-FLAG antiserum against tagged NirP1 in three biological replicates (R1–R3). The wild type was used as a control. b NirP1 pull-down analysis of two biological, independent replicates (R1 and R2). Upper panel: Coomassie-stained SDS gel, the NirP1 signal, and a band indicating a prominent coeluting protein of ~60 kDa are marked with arrows. Fractions are numbered and labeled CE cell extract, FT flow-through, W wash, E elution. Lower panel: Immunological identification of the lower band as NirP1. c Mass spectrometry-based analysis of NirP1 co-IP. Scatter plot of log2-transformed LFQ protein ratios (NirP1-3xFlag co-IP/cell extract) from two independent replicates. Displayed are proteins with a higher abundance in NirP1-3xFlag co-IP elution fractions compared to the initial cell extract. The main interacting protein was identified as ferredoxin-nitrite reductase (NiR), which is encoded by the nirA/slr0898 gene. d Western blot of NirP1 co-IP fractions using native, denaturing, or reducing conditions. All samples were loaded in biological replicates R1 and R2. e NirP1-NiR (WP_010873675.1) interaction in Synechocystis 6803 predicted by AlphaFold31,32 using their full-length sequences. The interaction modeled for the NiR and NirP1 homologs from Synechococcus 7942 is shown in Supplementary Fig. S7b. Source data are provided as a Source Data file.

NirP1-3xFLAG and potential bound interaction partners were coimmunoprecipitated from the lysate using anti-FLAG resin (Fig. 5b). Elution fractions of the protein pull-down showed NirP1 in Coomassie-stained protein gels and an additional band of higher molecular mass in both replicates. Because signals were obtained in the Western blot at these two and a third position, this result pointed at potential interaction partners of NirP1 that were copurified with the bait protein. These additional proteins were still present after 8 washing steps, indicating a strong interaction with NirP1 (Fig. 5b).

Next, we subjected whole-cell lysates and eluate fractions to liquid chromatography-tandem mass spectrometry (LC‒MS/MS)-based proteome analysis for the identification of coimmunoprecipitated proteins. In addition to NirP1, a single other protein was strongly enriched, which was identified as the 56 kDa ferredoxin-nitrite reductase (NiR), encoded by nirA (slr0898) (Fig. 5c). Overall, eight and 72 nonredundant peptides of the NirP1-3xFLAG protein and of NiR were detected, yielding high respective sequence coverages of 94% and 98% (Supplementary Datasets 7, 8, and 9). The combined molecular masses of NirP1 and NiR are compatible with the uppermost signal in the Western blot in Fig. 5b. These results indicated that NiR was a specifically enriched protein that coimmunoprecipitated with the NirP1-3xFLAG protein under the conditions used.

To validate the potential complex formation between NirP1 and NiR, Western blot experiments were performed under different reducing conditions. Under native conditions, two bands appeared. The signal at the lower molecular mass matched the predicted molecular mass of monomeric NirP1, while the upper signal was consistent with the sum of NirP1 and NiR of ~70 kDa. After the samples were denatured by heating or reduced through an addition of dithiothreitol (DTT) and β-mercaptoethanol, the ~70 kDa signal disappeared. The single remaining band migrated higher than the band of the native form, indicating that it was the monomeric and completely unfolded NirP1 protein (Fig. 5d).

Modeling by AlphaFold31,32 predicted NirP1 binding to the NiR substrate binding pocket next to its catalytic center with the iron-sulfur cluster where the reduction of nitrite takes place (Fig. 5e). Interestingly, none of the NirP1 cysteine residues, but two of the three totally conserved tryptophan residues (W36 and W65), were predicted close to the catalytic center. Thus, these amino acids potentially play a functional role.

Excretion of nitrite is triggered by NirP1

To test if the action of NirP1 might lead to the excretion of nitrite in LC conditions, as previously reported for Synechococcus 79429, we first verified if Synechocystis 6803 showed this effect as well. Indeed, after 2 h at LC, the Synechocystis 6803 wild type, but not the ΔnirP1 deletion mutant, started to excrete nitrite into the medium (Fig. 6a, b). After 24 h at LC, the concentration of nitrite in the medium of the wild-type strain was ~5 µM. However, in the nirP1-overexpressing strain the nitrite concentration was much higher with ~75 µM (Fig. 6a, b). All three results (ΔnirP1, wild type, NirP1oex) suggested that the excretion of nitrite into the supernatant was related to the presence of NirP1 in the cell. This effect was even more pronounced by the ectopic overexpression of NirP1 in BG11 medium aerated with 5% (v/v) CO2 leading to an ~150-fold higher accumulation of nitrite in the supernatant of NirP1oex compared to wild type and ΔnirP1 grown samples (HC in Fig. 6a). After the cells were washed and resuspended in fresh medium, the overexpressor immediately excreted nitrite again (time point 0 in Fig. 6a). These data establish that NirP1 triggers the excretion of nitrite.

Fig. 6: Excretion of nitrite is triggered in the presence of NirP1.
figure 6

a Synechocystis 6803 strains were grown photoautotrophically in BG11 medium with 5% (v/v) CO2 in air as high Ci (HC), harvested and resuspended in new medium and aerated with 0.04% (v/v) CO2 in air to initiate LC conditions. The wild type (WT) started to excrete nitrite in LC conditions when nirP1 was expressed (see Fig. 1c). b Same data as in (a) but at different scales to include the results from the NirP1oex strain. Accumulation of nitrite in the supernatant was followed over time. c The consumption of nitrate, measured as the decrease in the concentration of nitrate in the supernatants after pelleting the cells by centrifugation. Strains were grown in BG11, washed with N-free BG11 (BG11 –N), set to OD = 1 and resuspended in BG11 containing 200 µM nitrate. d The production and excretion of nitrite into the supernatant measured in the same samples as in (c) for nitrate. Data are presented as the means ± SD of measuring biological triplicates in duplicates each (n = 6). Significance was calculated using a two-tailed t-test with unequal variance (Welch’s t-test; ns no significance; *only P < 0.05 are shown) between the strains at corresponding time points. Full details are provided in Supplementary Dataset 10.

To ensure that the accumulated nitrite was a product of absorbed and reduced nitrate, the uptake of nitrate and subsequent excretion of nitrite were measured in parallel. The cultures were first grown in N-free BG11 medium and then 200 µM nitrate was added. In the control experiment, an additional 2 mM ammonia was added because nitrate uptake in Synechocystis is inhibited in the presence of ammonia and the transcription of nitrate and nitrite reductase genes is repressed. After 2 h, the overexpressor led to significant differences in nitrite concentration in the medium compared to wild type and ΔnirP1. In addition, the overexpression seemed to slow the uptake of nitrate (Fig. 6c, d). Here, the wild type, although less pronounced than the overexpressor, also excreted nitrite into the medium, as observed in the LC shift experiment before. In contrast, in the control experiment with additional ammonium, no nitrate uptake and no excretion of nitrite into the medium were observed (Supplementary Fig. S6a, b). These results demonstrated that the accumulation of nitrite in the medium of the wild type and overexpression strains resulted from nitrate uptake, the subsequent reduction to nitrite by NR and the alteration of NiR activity, with the involvement by NirP1.


Nitrate is one of the most abundant N-containing nutrients for cyanobacteria, while the preferred source for nitrogen assimilation is ammonium. Therefore, complex regulatory mechanisms occur to regulate nitrogen assimilation and uptake, e.g., genes of nitrate assimilation are repressed if ammonium is available37,38. When nitrate is the only N source available, it is taken up and reduced in two steps to nitrite and ammonium, which is then incorporated into glutamate in the GS/GOGAT cycle (Fig. 7). The reduction of nitrate to nitrite is catalyzed by the enzyme NR and the subsequent reduction to ammonium by NiR. NirP1, a small regulatory protein, is upregulated in LC conditions in Synechocystis 6803 (Fig. 1a). Its elevated expression, as shown in this study, leads to the accumulation of nitrite in the medium. We identified NiR as the primary binding partner for NirP1 (Fig. 5c). Therefore, the results can be explained if the NirP1:NiR interaction directly leads to the accumulation of nitrite in the cell, which is toxic and therefore becomes rapidly exported (Fig. 7). Excess internal nitrite has been shown to inhibit photosynthesis, particularly the activity of photosystem II in Synechocystis 680339. Comparable responses were reported for N-starved Synechocystis 6803 mutants defective in glycogen synthesis that excreted 2-OG and pyruvate34.

Fig. 7: NirP1 is a regulator at a central position in the assimilation of inorganic nitrogen.
figure 7

Cyanobacteria can assimilate N from simple compounds, such as ammonium, nitrate, nitrite, and urea. During nitrate assimilation, intracellular nitrate is reduced in two steps to nitrite and ammonium, which is then incorporated into glutamate in the GS/GOGAT cycle or into carbon skeletons during the synthesis of amino acids by transaminases. The reduction of nitrate to nitrite is catalyzed by the NR and the subsequent reduction to ammonium by the NiR. Transcription of the regulatory factor NirP1 is controlled by an unknown transcription factor (TF) and by NtcA. 2-OG is a corepressor of NtcA. Therefore, an increased 2-OG level leads to the repression of nirP1 transcription, while a low level of 2-OG leads to its activation. NirP1 binds to the enzyme nitrite reductase (NiR), leading to the inhibition of nitrite reduction. The accumulating nitrite is then rapidly exported by an unidentified export system. Transport systems for ammonium (AMT) and nitrate/nitrite (NRT) uptake are shown.

Several factors control primary N metabolism, such as the transcription factor NtcA or the PII regulator and proteins interacting with them10,11,12,13,14,17,18,19. However, unlike several other regulatory proteins that modulate the activity of other regulators (e.g., PirA and PipX), NirP1 targets one of the key enzymes of N assimilation metabolism directly. This is reminiscent of the direct inhibitory role of the small GS inhibitory factors IF7 and IF1717. These findings suggest that a higher number of relatively small proteins might exist that modulate or modify the activity of relevant enzymes and await characterization.

To examine the possible mode of action, we further predicted the interactions between NiR and other proteins. In Synechocystis 6803, ferredoxin Fd1 (ssl0020, petF) acts as an electron donor for the reduction of nitrite40. Therefore, we modeled the interaction of NiR with NirP1 (Fig. 5e) and Fd1 (Supplementary Fig. S7a). The results suggested that both proteins compete for the binding pocket of NiR. By binding NirP1 instead of Fd1, the binding pocket is blocked for electron transfer and the subsequent reduction of nitrite to ammonium. Thus, this could be a possible mechanism by which NirP1 inhibits NiR, leading to the excretion and accumulation of nitrite in the medium during NirP1 expression.

Transcription of nirP1 is controlled by an unknown transcription factor and by NtcA. If 2-OG cannot be aminated efficiently in the GS-GOGAT cycle due to N starvation, Ci oversupply, or a combination of both, it will accumulate in the cell. The increased 2-OG level is sensed by PII and transferred to NtcA, leading to stronger promoter binding and the repression of nirP1 transcription. Therefore, NirP1 is at the confluence of C and N assimilation and is effectively a regulator of the C/N balance.

In addition to the NtcA-binding site, we found another putative promoter element at position −43 to −60 (Fig. 2a), the mutation of which rendered the nirP1 promoter inactive. Therefore, this element is likely recognized by a positively acting transcription factor, although its identity was not further investigated in this work and should be studied in the future.

It was previously reported that during photoautotrophic growth in shifts to LC conditions, Synechococcus 7942 excreted nitrite into the medium9. As the fundamental reason for this observation, a decline in the enzymatic activity of NiR was identified that started within 30 min. This decline was not a consequence of changed nirA expression because the observed decrease in NiR activity under LC conditions was substantially larger than that observed when the cultures were treated with rifampicin to inhibit transcription. Instead, a mechanism was postulated that directly inactivated NiR under LC conditions9. Our data were obtained in Synechocystis 6803, but a close NirP1 homolog is found in Synechococcus 7942 (Fig. 1b), and an almost identical interaction with NiR was predicted (Supplementary Fig. S7b, c). Therefore, we have likely identified the mechanism underlying previously observed NiR inhibition and nitrite excretion. We calculated that approximately 5% of assimilated nitrate was excreted as nitrite into the medium by Synechocystis 6803 in the first 2 h (Fig. 6c, d). Compared to Prochlorococcus strains MIT0917 and SB, and Synechococcus 7942, which were reported to excrete up to 30% of the assimilated nitrate as nitrite6,9, the here-measured nitrite excretion of Synechocystis 6803 was less pronounced. Previous studies reported for Synechocystis 6803 a higher excretion, of up to 60% of the reduced nitrate under mixotrophic conditions41, which clearly shows that Synechocystis 6803 is capable to release high nitrite amounts into the medium. The nirP1 overexpression strain was obviously not significantly inhibited by the remaining nitrite, which could indicate that it had a way to avoid nitrite poisoning by storing it at a physiologically neutral place (or the regulation by NirP1 was not completely inhibiting the NiR activity). Indeed, there is circumstantial evidence for such a nitrite storage, possibly in the periplasm. We observed a very fast immediate release of nitrite from washed cells of the NirP1oex strain (see time point 0 in Fig. 6a), which then was followed only by a slower further increase. Thus, under photoautotrophic conditions used here, the cells may have a mechanism to store most of the nitrite in order to resume normal N assimilation when the Ci supply increases again. This points at a possible mechanism for deciding whether nitrite is released or kept within the cell, depending on the physiological conditions. These points are interesting topics for future research.

According to our data, NirP1 expression is upregulated under LC conditions and then participated in the inhibition of NiR, leading to nitrite excretion and accumulation. Surprisingly, the regulation occurs at an unexpected point (Fig. 7), as controlling N assimilation during nitrate uptake or during nitrate reduction seems more intuitive. Here, uncontrolled nitrate uptake led to light-dependent nitrite excretion under low light conditions42. Therefore, the availability of electrons could play a role because the reduction of nitrite requires six moles of reduced ferredoxin, while the reduction of nitrate requires only two. Another consideration is the possible beneficial effect of nitrite excretion on cooccurring microbes in the natural environment. Certain Prochlorococcus release up to 30% of their N uptake as extracellular nitrite when grown on nitrate6. Prochlorococcus typically contains the most compact genomes of any free-living cyanobacteria, which was linked to its quasi-homeostatic environment43,44,45. Because we did not identify NirP1 homologs in Prochlorococcus, the underlying mechanism probably operates without NirP1. In other strains of cyanobacteria, exposed to greater variations in the environment, the involvement of NirP1 has likely led to a more accurate regulation. In any case, the release of an N cycle intermediate, such as nitrite, should have tremendous relevance for certain other members of the microbiome. Our results show that homologs of NirP1 are widely distributed among cyanobacteria, expanding our understanding of how C and N metabolism become integrated in these organisms.


Cultivation conditions and spectrometric measurements

Strains were maintained in BG11 or in copper-free BG1128 supplemented with 20 mM TES pH 7.5 under continuous white light of 50 µmol photons m−2 s−1 at 30 °C. Mutant strains containing pVZ322-derived plasmids were cultivated in the presence of 50 µg/mL kanamycin, while ΔnirP1 was maintained at 20 µg/mL streptomycin. The cultures for overexpression of NirP1 using the PpetE promoter for pull-down experiments and MS measurements were grown either in BG11 or under CellDeg® conditions supplemented with 2 µM CuSO4. Cultivation with CellDeg® conditions was performed for 4 days in 100 mL FOM (freshwater organism medium; 50 mM NaNO3, 15 mM KNO3, 2 mM MgSO4, 0.5 mM CaCl2, 0.025 mM H3BO3, supplemented with 0.15 mM FeCl3 /Na2EDTA, 1.6 mM KH2PO4, 2.4 mM K2HPO4, 10 mM NaHCO3, 10 nM MnCl2, 0.1 nM ZnSO4, 2 nM Na2MoO4, 10 pM CuSO4, and 30 pM CoCl2) inoculated with culture with a starting OD of ~0.8 and 150 μmol photons m2 s1. High CO2 concentrations were delivered to the flasks from a carbonate buffer (0.3 M K2CO3 and 2.7 M KHCO3) through a highly gas-permeable polypropylene membrane. Light intensity was increased during growth according to the cell density. 24 h prior harvesting the cells (5000 × g, 4 °C, 10 min), protein expression was started by adding 2 µM CuSO4. Cultures in growth experiments and phenotypic assays were grown in BG11 medium supplemented with copper but without antibiotics to prevent any possible side effects.

Whole-cell absorption spectra were measured using a Specord® 250 Plus (Analytik Jena) spectrophotometer at room temperature and were normalized to the wild type at 750 nm. Cultures were measured in triplicate. Cultures at an OD750 > 1 were diluted with 1× BG11 prior to taking the absorption spectra.

Construction of mutant and overexpression lines

Synechocystis 6803 PCC-M46 was used as the wild type and background for the construction of mutants. Knockout mutants were generated by replacing the nirP1 (ncr1070) coding sequence with a streptomycin resistance cassette (aadA)47 by homologous recombination and using pUC19 as a vector for subcloning. The construct for gene replacement was generated by AQUA cloning48. Details of primer sequences and plasmids are provided in Supplementary Tables S1 and S2. Segregation of nirP1 and ΔnirP1/aadA alleles was checked by colony PCR using the primers P_AK9 and P_AK10 (Supplementary Fig. S1).

To complement the knockout and create a NirP1-overexpressing strain, nirP1 was inserted into the self-replicating pVZ322 plasmid under control of the Cu2+-inducible PpetE promoter or under control of its native promoter (PnirP1). The inserts for the overexpression of nirP1 were assembled into pVZ322 predigested by XbaI and PstI for 16 h at 37 °C by AQUA cloning48. The constructs were checked using the primers P_AK16 and P_AK17 (Supplementary Fig. S1). The sequence-verified plasmids were introduced into Synechocystis 6803 by electroporation49. For the electroporation, culture was cultivated to exponential phase with OD ~ 0.7–1.0, 25 mL washed three times with ice-cold HEPES buffer (1 mM, pH = 7.5) and resuspended in 0.5 mL HEPES buffer. 1 µg of plasmid was mixed with 100 µL of cells and transferred into a precooled Gene Pulser/MicroPulser Electroporation Cuvettes (0.2 cm gap, BIO-RAD). A pulse of 12.5 kV and 4 ms was executed (MicroPulser Electroporator, BIO-RAD), 1 mL of BG11 (BG11 lacking copper for constructs using the petE promotor) was added to the cells and transferred to 50 mL BG11 for 24 h cultivation under continuous light of 50 µmol photons m−2 s−1 at 30 °C. The culture was harvested (5000 × g, 4 °C, 10 min), resuspended in 100 µL BG11, and plated on 0.75% Kobe-agar plate with 50 µg/mL kanamycin and cultivated for 5 days. Single colonies were transferred on a new plate and verified by PCR (Supplementary Fig. S1, Supplementary Table S1). The expression of nirP1 was checked by Western hybridization using ANTI-FLAG antisera (Sigma–Aldrich, #SAB4301135).

RNA preparation and northern blot verification

Cultures were grown in BG11 supplemented with 10 mM NHCO3 (HC), washed three times with BG11 lacking any Ci source (LC), and grown for 1, 3, 6, 12, and 24 h in LC conditions to induce nirP1 expression. The time immediately before Ci removal (0 h, HC) served as the negative control. Cells were collected by filtration through hydrophilic polyethersulfone filters (Pall Supor®−800, 0.8 μm), transferred into PGTX buffer, and instantly frozen in liquid N250. Samples were incubated for 10 min at 65 °C in a water bath, one volume chloroform:isoamyl alcohol (24:1) was added, and were incubated for 15 min at room temperature with several vortexing cycles. After centrifugation for 3 min at 3250 × g, the supernatant was transferred to a fresh tube, one volume of chloroform:isoamyl alcohol was added. This step was repeated twice. RNA was precipitated by the addition of one volume of isopropanol and incubated for 16 h at −20 °C. For northern blot verification, 10 μg of RNA per well was loaded on a 10% polyacrylamide-8.3 M urea gel and electroblotted onto Hybond N nylon+ membranes (Amersham) with 1 mA per cm2 for 1 h. nirP1 transcript accumulation was analyzed by Northern hybridization using single-stranded radioactively labeled RNA probes transcribed in vitro from PCR-generated templates using primer P_AK30 and P_AK31 (Supplementary Table S1). The radioactively labeled probes were generated using [α-32P]-UTP and the Maxiscript T7 In vitro transcription kit (Thermo Fisher Scientific). Hybridizations were performed in 50% deionized formamide, 7% SDS, 250 mM NaCl, and 120 mM Na2HPO4/NaH2PO4 pH 7.2 overnight at 62 °C. The membranes were washed in buffer 1 (2× SSC (3 M NaCl, 0.3 M sodium citrate, pH 7.0), 1% SDS), buffer 2 (1× SSC, 0.5% SDS) and buffer 3 (0.1× SSC, 0.1% SDS) each for 10 min. All wash steps were performed at 57 °C. Signals were detected with a Laser Scanner Typhoon FLA 9500 (GE Healthcare). Signal intensities of nirP1 were determined using Quantity One software (BIO-RAD) and normalized to the signal of the 5S rRNA (Fig. 1c).

Promoter analysis

To verify nirP1 promoter activity in vivo in Synechocystis 6803, the nucleotides −70 to +30 (TSS at +1) were fused to luxAB reporter genes and integrated into the genome. Initially, cultures were set to an OD 0.7 and grown under standard BG11 or HC conditions for 2 h and subsequently transferred to NO3- or NaCO3-free BG11 or BG11 medium supplemented with additional NH4Cl to induce transcription. Measurements were carried out for 24 h in LC or high NH4Cl conditions or for 7 days under N deprivation. To recover from N starvation, the bleached cells were transferred back to standard BG11 medium after 7 days and observed for 3 days. Bioluminescence was measured in vivo as total light counts per second by using a VICTOR3 multiplate reader (PerkinElmer). Prior to the measurement, cells were diluted to an OD750 = 0.7 and 200 μL of the suspension were filled into a white 96-well plate (CulturePlate™−96, PerkinElmer). To increase the signal, 2 µL of Decanal was added to the samples. In the multiplate reader the cell suspensions were shaken for 10 s and subsequently total light emission was measured for 1 s. A strain carrying the promotorless luxAB genes served as a negative control51. The raw data, the calculation, and statistical evaluations are provided in Supplementary Dataset 3.

Co-IP, LC-MS/MS analyses, and data processing

The overexpression of NirP1-3xFLAG was induced in exponentially growing cultures of the Synechocystis 6803 strain NirP1oex in the CellDeg system by adding Cu2SO4 to a final concentration of 2 µM. 50 mL of cells were harvested by centrifugation (5000 × g, 4 °C, 10 min) after 24 h of NirP1-3xFLAG overexpression, resuspended in 2 mL FLAG buffer (50 mM HEPES-NaOH pH 7; 5 mM MgCl2; 25 mM CaCl2, 150 mM NaCl; 10% glycerol; 0.1% Tween-20) supplemented with Protease Inhibitor (cOmplete, Roche) mixed with 100 μL of glass beads (Ø 0.1–0.25 mm, Retsch) and lysed mechanically in a prechilled Precelly24 (Bertin Instruments) with three rounds of shaking at 6000 rpm (10 s) followed by a pause (5 s) per cycle, for a total of 5 cycles. To remove cell debris and glass beads, extracts were centrifuged (1000 × g, 5 min, 4 °C). Supernatants were collected for further processing. Remaining cell debris was removed by centrifugation (15,000 × g, 4 °C, 30 min). The cleared total cell lysate was subjected to co-IP using Anti-FLAG M2 affinity magnetic beads (Sigma–Aldrich) rotating for 1 h at 4 °C. The beads were washed 8 times with FLAG buffer (2 min, 4 °C, rotating). Bound proteins were eluted with FLAG peptide (Sigma–Aldrich, 100 ng/µL) in TBS buffer (150 mM NaCl in 20 mM Tris/HCl, pH = 7.6) at 4 °C. Two independent replicates were prepared. Protein concentrations of eluate fractions and respective whole-cell lysates were measured in the Coomassie Plus Bradford Assay (Thermo Fisher Scientific) through the plate photo reader 1420 multilabel counter (PerkinElmer) in duplicate, and 500 µL of each was subsequently snap frozen. Proteins in whole-cell lysates were purified by acetone/methanol precipitation52, and co-IP eluates were lyophilized. For further analysis, samples were resolubilized in denaturation buffer (6 M urea, 2 M thiourea in 100 mM Tris/HCl; pH 8.0) at a final protein concentration of 1 µg/µL. Protein disulfide bonds were reduced with 1 mM dithiothreitol for 45 min and the resulting thiol groups were alkylated with iodoacetamide at a final concentration of 5.5 mM for 45 min. Predigestion of proteins with Lys-C (Santa Cruz Biotechnology) for 3 h was followed by dilution with 4 volumes of 20 mM ammonium bicarbonate buffer, pH 8, and overnight digestion with trypsin (sequencing grade, Promega), both at protease/protein ratios of 1/100. The resulting peptide solutions were acidified with trifluoroacetic acid to pH 2.5, and an aliquot corresponding to 10 µg peptides was purified by stage tips53. For LC‒MS/MS-based protein analysis, 40 ng of each sample was loaded onto an in-house-made 20 cm column with 75 µm ID, packed with ReproSil-Pur 1.9 µm C18 material (Dr. Maisch, Germany) and separated by RP chromatography on an EASY-nLC 1200 system (Thermo Fisher Scientific, USA) using 60 min gradients. Eluting peptides were analyzed on an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific, USA) as described elsewhere54.

All raw spectra were processed with MaxQuant software (version at default settings. The match between run option and label-free quantification (LFQ) was enabled. Peak lists were searched against an in-house modified target-decoy database of Synechocystis 6803 with 3,681 protein sequences, including the NirP1-3xFLAG fusion protein sequence. False discovery rates were limited to 1% at the peptide and protein levels. LFQ protein intensities of two independent co-IPs were normalized by corresponding values from whole-cell extract measurements, log2 transformed, and plotted against each other using the Perseus software suite (version

SDS‒PAGE, native gel electrophoresis and western blotting

Proteins were mixed with denaturing or reducing loading dye (5× concentrated: 250 mM Tris-HCl pH 6.8, 25% glycerol, 10% SDS, 500 mM DTT, and 0.05% bromophenol blue G-250). Moreover, 6% ß-mercaptoethanol (v/v) was added fresh to the sample in reducing conditions before heating for 5 min at 95 °C. The protein samples were separated by 12% SDS‒PAGE using the PageRulerTM molecular weight marker (Thermo Fisher). To run samples under nonreducing conditions, a native loading dye (5× concentrated: 30% glycerol, 0.05% bromophenol blue G-250, 150 mM Tris-HCl pH 7.5) was used, and boiling was omitted. To make proteins visible, the gel was stained with Coomassie dye. The gel was incubated for 30 min under slight shaking, covered with Instant Blue Coomassie Protein Stain (Abcam), and washed overnight with ddH2O. For immunoblot analysis, separated proteins were transferred to nitrocellulose blotting membrane (AmershamTM Protan, pore size 0.4 5 µm). Membranes were blocked overnight at 4 °C with 5% milk powder in TBS-T, washed three times (10 min), and subsequently probed with monoclonal ANTI-FLAG® M2-Peroxidase (HRP) antibody (Sigma–Aldrich, #SAB4301135 at a dilution of 1:5000) in TBS-T for 1 h at room temperature in the dark and washed again three times (10 min). All washing steps were performed with TBS-T (20 mM Tris pH 7.6, 150 mM NaCl, 0.1% (v/v) Tween-20) at room temperature. Signals were detected with WesternBrightTM ECL-spray (Advansta) on a chemiluminescence imager system (Fusion SL, Vilber Lourmat) and subsequently visualized using FUSION-CAP (Vilber Lourmat) and Quantity One software (BIO-RAD).

Nitrate and nitrite measurements

Wild type, ΔnirP1, and nirP1-overexpressing cultures were cultivated photoautotrophically in BG11 medium with 5% (v/v) CO2 in air as the high CO2 condition (HC), harvested and resuspended in new medium and aerated with 0.04% (v/v) CO2 in air to start low CO2 (LC) conditions. Samples were taken under HC conditions before cells were shifted to LC conditions and then after 0, 2, 4, 6, and 24 h of growth in LC conditions and measured in biological triplicates.

To measure nitrate uptake and nitrite accumulation, cells were washed three times and resuspended in BG11 without NaNO3 (BG11-N) lacking CuSO4. After incubation for 1 h, only 200 µM NaNO3 as the nitrogen source or NaNO3 together with 2 mM NH4Cl and a final concentration of 2 µM CuSO4 were added. Samples were collected at the indicated time points and centrifuged (16,800 × g, 5 min, room temperature). Concentrations in the resulting supernatants were determined by ion chromatography measurements or with the Griess Reagent Kit (Promega). For IC measurements, samples were mixed with methanol (1:3 (v:v)) and centrifuged three times (16,800 × g, 10 min, room temperature), and 10 µL was loaded and separated by a column (DionexTM IonPacTM AS11-HC, Thermo Fischer). Ions were identified, and concentrations were quantified using the area (µS min−1) given in the software (Chromeleon® 7). Standards at various concentrations were used for calibration. The raw data, calculations, and statistical evaluations are given in Supplementary Dataset 10. Nitrite was measured after conversion into an azo dye57. The determination limit was 0.09 µmol L−1. The combined standard uncertainty was 4.9%. Samples higher than the measuring range were diluted with ultrapure water.

Extraction and metabolite analysis

Wild type, ΔnirP1, and nirP1 overexpressor cultures were set in duplicates or triplicates to an OD of 0.5 and grown for 24 h without antibiotics before cells were pelleted by centrifugation (3,210 × g, 10 min, room temperature), washed three times in copper-free BG11 medium and induction of NirP1 was started by adding 2 µM CuSO4 after 1 h. After 0, 1, 3, 12, and 24 h of incubation, 2 mL of each culture was collected by centrifugation (16,800 × g, 5 min, room temperature) and immersed in liquid N2. For extraction, 630 µL of methanol containing carnitine (internal standard, 1 mg per extraction) was added, mixed for 1 min, and incubated for 5 min in a sonic water bath. After another 15 min of shaking at room temperature, 400 µL of chloroform was added, and the sample was incubated at 37 °C for 5 min. Next, 800 µL of ROTISOLV LC‒MS grade H2O (Roth) was added, and the sample was mixed thoroughly. Precipitation was enabled by storage at 22 °C for at least 2 h, and phase separation was subsequently achieved by centrifugation for 5 min at room temperature (16,800 × g). The upper polar phase was collected and dried in a speed vac for 30 min followed by lyophilization overnight.

Absolute metabolite contents were quantified on a high-performance liquid chromatography‒mass spectrometer LC‒MS-8050 system (Shimadzu)58. Dried samples were dissolved in 200 μL LC‒MS grade water and filtered through 0.2-mm filters (Omnifix-F, Braun, Germany), and 5 μL of the cleared supernatant was separated on a pentafluorophenylpropyl column (Supelco Discovery HS FS, 3 mm, 150 3 2.1 mm).

The compounds were identified and quantified using the multiple reaction monitoring (MRM) values given in the LC‒MS/MS method package and the LabSolutions software package (Shimadzu). Authentic standard substances (Merck) at various concentrations were used for calibration, and peak areas were normalized to signals of the internal standard (carnitine). The data were further normalized to the OD750 and volume measured for each sample. The raw data and estimations as absolute values in ng per mL per OD750 are given in Supplementary Datasets 5 and 6, with all the statistical evaluations.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.