Introduction

Cyanobacteria are photosynthetic microorganisms essential for harvesting solar energy, producing oxygen, and contributing to Earth’s oxygenation, thus creating conditions that led to aerobic life forms1,2,3. Their ability to consume CO2 and convert solar energy into chemical energy through photosynthesis has allowed them to succeed in diverse environments. These metabolic attributes make them valuable for carbon-neutral energy production, carbon sequestration, and engineering of energy-rich biomolecules4. Crocosphaera subtropica sp. ATCC 51142 (formerly known as Cyanothece sp. ATCC 51142, hereafter referred as Crocosphaera 51142) is a unicellular diazotrophic strain that performs biological nitrogen fixation (BNF), while protecting nitrogenase, which are sensitive to O2, from oxygen toxicity through an anoxic cytoplasmic environment5,6,7. Its genome reveals extensive metabolic potential, active photosynthesis, and efficient CO2-uptake8. Notably, it can perform BNF in darkness or prolonged light after light-dark entrainment9,10, aligning with its circadian clock. This versatility makes Crocosphaera 51142 an excellent model for studying photosynthesis, BNF, carbon sequestration, and circadian regulation11,12.

The intensity and duration of light as well as nitrogen availability strongly influence cyanobacterial physiology and time-dependent adaptations to the light-dark cycle9,13. Under nitrogen-fixing conditions, cells accumulate glycogen granules, crucial for energy production and reducing intracellular oxygen during nitrogen fixation14. Gaining deeper insights into the mechanistic processes that link light-dark sensing, protein translation, membrane organization, and signaling can provide key information on how Crocosphaera 51142 can be optimized for improved renewable energy production. In Crocosphaera under LD growth, nitrogenase expression is restricted to nitrate-depleted growth and typically occurs during the dark cycle without an external carbon source15,16,17. However, in the presence of organic carbon (e.g., glycerol), nitrogenase is expressed during both light and dark cycles, likely due to enhanced light-phase respiration creating an anoxic environment15.

Advances in mass spectrometry have significantly enhanced sensitivity, resolution, and accuracy of measurements, allowing for more detailed analyses of complex biological processes. Leveraging these advancements, we performed quantitative proteomic analysis of Crocosphaera 51142 to investigate the mechanisms driving metabolic shifts during BNF and light-dark transitions. By examining metabolic differences under nitrogen-fixing and non-fixing conditions, we aimed to understand how Crocosphaera 51142 temporally separates mutually exclusive processes like photosynthesis and BNF within a single cell and adapts to different natural environments. This study linked changes in protein abundance changes and pathway dynamics to key processes such as CO₂ fixation, respiration, and glycogen metabolism, that coordinate the timing of photosynthesis and BNF in cyanobacteria. Our results reveal significant proteomic alterations in response to nitrogen availability and light-dark transitions, shedding light on the metabolic strategies underpinning these essential functions.

Results

Overview of the proteomic results

The minima and the maxima of O2 production in cyanobacteria occurs around 6–8 h of the dark and the light periods, respectively18. Thus, we analyzed the proteome for six hours into both the light and dark cycles to capture the organism’s peak or near peak metabolic activities. Cells were grown as previously described19 (Fig. 1) and harvested 6 h into the light and 6 h into the dark cycles. Four biological replicates per treatment group were separated into soluble and insoluble fractions for in-depth LC-MS/MS analysis. LC-MS/MS data were collected using a 130-minute LC method in DDA identifying 29,946 peptides (Table S1) and 2,688 proteins (Table S2) representing ~ 40% of the proteome after MaxQuant search. Filtering for high-confidence proteins (identified in at least three of four replicates with LFQ intensity >0 and MS/MS counts >2) yielded 2,050 proteins for further bioinformatic analysis. Of these, 1,321 were shared across groups, with the highest protein count under nitrogen-fixing conditions in the dark (Fig. 2a).

Fig. 1
figure 1

Experimental workflow. (A). Culture growth conditions and sample collection timeline. Crocosphaera subtropica ATCC 51142 cultures were grown in ASP2 medium without NaNO3 (-NO3) and with NaNO3 (+NO3) at 30 °C under continuous light (LL) for seven days. Then, cultures were grown for seven days at 12 h light/dark (L/D) cycles before harvesting cells at 6 h into the light (L) and 6 h into the dark (D). (B). Cells were lysed and separated into soluble and insoluble fractions by differential centrifugation. Soluble lysates were acetone precipitated, and insoluble fractions were solubilized with 0.1% Rapigest. Both fractions were digested with trypsin and analyzed by LC-MS/MS with four biological replicates per group. Data processing and statistical analysis were performed using MaxQuant and Perseus, respectively. Sequence-based function prediction of the identified proteins was done using PFP, Phylo-PFP and ESG to assign GO terms to the identified proteins. Please refer to the “Materials and Methods” for details about this analysis.

Full size image
Fig. 2
figure 2

Differentially regulated proteins during diurnal cycles and nitrogen fixation states. (A). Venn Diagram showing protein overlaps between light-dark cycles with and without the presence of nitrate (D-, D+, L- and L+). (B). PCA Plot of all the analyzed replicates of D-, D+, L- and L + samples. The explained variances are indicated in the brackets. (C). Heatmap shows the hierarchical clustering of 361 proteins changing due to interaction of effect of light and dark as well as nitrate. Both rows and columns were clustered by Euclidean distance and average linkage method, column clustering indicates the reproducibility within the replicates for each condition. D-: dark cycle without nitrate present; D+: dark cycle with nitrate present; L-: light cycle without nitrate present; L+: light cycle with nitrate present.

Full size image

Principal component analysis (PCA) showed strong consistency between replicates (Fig. 2b). ANOVA revealed 1,370 proteins changed with nitrate (adj. p value ≤ 0.05), 557 with L/D cycles, and 361 with both (Fig. 2c, Table S3). Hierarchical clustering highlighted nitrate as causing the largest proteome differences (Figure S1a). The Pearson correlation analysis (Figure S1b) shows the relative impact of nitrate availability and L/D cycles on Crocosphaera 51142 proteome and reveals that nitrate availability exerts a more pronounced influence compared to the effects of L/D cycles. Nitrogenases were highly upregulated, while 2 of the 4 flavoproteins (cce_3835; cce_3833) and heme oxygenase (Ho1; cce_2573), were among the most downregulated proteins in the dark under nitrogen-fixing conditions. Key metabolic categories affected included nitrogen fixation, photosynthesis, respiration, CO₂ fixation, and protein turnover. Of Crocosphaera 51142’s 5,269 predicted proteins8, 20% identified here remain uncharacterized, underscoring the need to study these proteins to fully understand its metabolism.

Figure 3 depicts the GO classification of proteins significantly altered by light-dark cycles (blue), nitrate availability (green), and their interaction (red). The circular bar plots highlight changes in biosynthetic processes (BSP) and metabolic processes (MP). Proteins involved in nitrogenase activity, molybdenum cofactor binding, lyase activity, tryptophan synthase activity, oxidoreductase activity, and phycobilisome were enriched under nitrate and light-dark interaction conditions, highlighting their roles in nitrogen assimilation and photosynthetic efficiency. Nitrate primarily influenced proteins involved in nitrogen assimilation, macromolecule biosynthetic process, gene expression, non-membrane bound organelles, and organic substances. Conversely, proteins affected by light-dark transitions were enriched in pathways related to macromolecule biosynthesis, gene expression, organonitrogen compound biosynthesis, primary metabolic processes, and photosynthesis (Fig. 3). Interestingly, the effects of individual growth conditions on these proteins and pathways were more pronounced than their combined interaction effects, suggesting that the responses to these conditions are not simply additive but may involve competing regulatory mechanisms. Detailed protein classifications and associated pathways are listed in Table S11.

Fig. 3
figure 3

GO classification of significantly changing proteins. Circular bar plot shows the gene ontology classification for the significantly changing proteins due to differences in light and dark (blue), effect of nitrate (green) and interaction of both these conditions (red) (Supplementary Table S11). BSP: Biosynthetic process, MP: Metabolic process, EC 3.5.4.26: diaminohydroxyphosphoribosylaminopyrimidine deaminase, EC 1.1.1.193: 5-amino-6-(5-phosphoribosylamino)uracil reductase.

Full size image

Figure 4 uses volcano plots to depict protein differences under nitrogen-fixing and non-fixing conditions. During the dark cycle under nitrogen-fixing growth, nitrogenase-related proteins (NifD, NifK, NifB, NifS, NifU), hydrogen metabolism (HupL), and respiration (PetH) were the most abundant, while bicarbonate transporters (CmpA, CmpC, CmpD) were reduced (Fig. 4a and c). In the light period under nitrogen-fixing conditions, enzymes such as GlpD, Pgi, Pgl, MetE, KaiC2, PsbA4, GlgA1, TalA, AphA, and AcnB, were the most abundant (Fig. 4b and d). Notably, NifH expression during the light suggests nitrogenase activity can occur in anoxic conditions beyond dark periods15,20.

Fig. 4
figure 4

Proteins exhibiting the highest fold change dynamics. Volcano Plots indicate the differential regulation of proteins because of nitrate (A, B) and effect of light and dark (C, D). Significantly changing proteins determined by T-tests with a q-value ≤ 0.05 and threshold log2(fold change) = ± 0.58 are indicated in red (significantly up) and blue (significantly downregulated) in their respective conditions. Proteins exhibiting the highest fold change dynamics in each comparison are specified. Proteins with q = 0 were replaced with the lowest possible q-value (or the highest possible –log10(q-value) for the respective dataset. D-: dark cycle without nitrate present; D+: dark cycle with nitrate present; L-: light cycle without nitrate present; L+: light cycle with nitrate present.

Full size image

Differential regulation of proteins involved in respiration and glycogen metabolism

The respiratory enzyme CoxB1 (cce_1977) was significantly upregulated under nitrogen-fixing conditions, increasing 13-fold in the dark and 3.8-fold in the light (Table 1). Its expression correlated with nitrogenase activity (Fig. 5a and b), suggesting CoxB1’s role in active respiration to create anoxic conditions for nitrogenase activity15. Under nitrogen-fixing conditions, GlgA1 and GlgA2 were more abundant in the light, while glycogen metabolism enzymes GlgP1 and GlgP2 were more abundant in the dark (Table 1). Both glycogen debranching enzymes, GlgP and GlgX, were identified but showed varied responses to light-dark cycles and nitrate. GlgP1 (cce_1629) was upregulated, whereas GlgP2 (cce_5186) and GlgP3 (cce_1603) were downregulated under nitrogen-fixing conditions, regardless of light-dark cycles. GlgX was also upregulated under nitrogen-fixing conditions suggesting a tight regulation of glycogen accumulation and degradation during nitrogen-fixation.

Table 1 List of important proteins changing due to at least one condition (effect of nitrate/light and dark/interaction of both light and dark) with their fold change.
Full size table
Fig. 5
figure 5

Expression of nitrogenase enzyme clusters. (A) Spider chart depicting the differential regulation of enzymes involved in the nitrogenase cluster under various conditions (D+/D-/L+/L-). The indicated log(LFQ intensity) is in increasing order. (B) List of all the identified nitrogenase clusters and associated genes that play an important role in nitrogen-fixation. All the proteins (except the highlighted ones in grey) are significantly changing in at least one condition. D-: dark cycle without nitrate present; D+: dark cycle with nitrate present; L-: light cycle without nitrate present; L+: light cycle with nitrate present. The spider chart was prepared using Origin 2022 software.

Full size image

Regulation of glycolysis, tricarboxylic acid cycle and oxidative pentose phosphate pathway enzymes

Abundance of glycolytic enzymes remained consistent between L/D cycles but showed a strong response to nitrate availability (Table 1, Figure S2a, Table S3). The glucose-oxidizing enzyme Zwf (cce_2536), which generates NADPH for BNF21, was upregulated in the dark under nitrogen-fixing conditions (Table 1). Other OPP pathway enzymes, including Gnd (cce_3746), OpcA (cce_2535), TalA (cce_4686), TalC (cce_4208), and Pgl (cce_4743) also showed higher abundance in the dark during nitrogen-fixing condition (Table 1). Interestingly, TalA levels were affected by light-dark cycles and remained unaffected by nitrate availability, whereas TalC levels were stable across light-dark cycles but increased in the absence of nitrate (Table 1) suggesting different responses of these PPP enzymes to growth conditions.

Enzymes in the TCA cycle were generally more abundant in the dark, particularly in nitrate availability (Table 1, Figure S2a). Icd (cce_3202), involved in nitrogen-fixing growth, was significantly upregulated (Table 1), consistent with previous reports of increased icd transcript levels under nitrogen-fixing conditions in other cyanobacteria22. GabD (cce_4228), which converts 2-OG to succinate, was less abundant in nitrogen-fixing cells but exhibited higher levels in the dark, confirming earlier findings16. AcnB (cce_3280) abundance was influenced by both nitrate availability and L/D cycles (Table 1). During nitrogen fixation, cyanobacteria utilize alternative nitrogen sources23, with ammonia (NH4+) produced by BNF being converted to glutamate via GS (GlnA)-GOGAT (GlsF) cycle. This process results in increased dark-phase abundance of GlnA, GlnB, and GlsA. Additionally, enzymes CarA (cce_0902) and CarB (cce_2038) were upregulated in the dark, highlighting glutamine’s involvement in purine and pyrimidine biosynthesis. PyrFE (cce_0502) and PyrG (cce_2923) also exhibited higher abundance during the dark phase of nitrogen fixation. Urea cycle enzymes, such as ArgB (cce_3224), ArgG (cce_4370), and ArgF (cce_3251), were upregulated, indicating increased production of citrulline and argininosuccinate. These findings suggest that Crocosphaera 51142 utilizes a combination of conventional and alternative pathways to optimize metabolic activities under varying environmental conditions.

Expression of nitrogenase and hydrogenase enzyme clusters

Nitrogenases were among the most abundant proteins, with 14 annotated nitrogenase enzymes identified, all differentially regulated and exclusively expressed under nitrogen-fixing conditions (Fig. 5a and b). These enzymes were absent in non-fixing conditions. While all nitrogenases were present during both light and dark cycles under nitrogen-fixing growth, their abundances were significantly higher in the dark, with changes ranging from a 7-fold for NifN to 450-fold for NifH (Fig. 5b). The structural proteins NifHDK were the most upregulated, followed by NifT (cce_0547), and NifW (cce_0568). Additionally, the 2Fe-2S putative nitrogen-fixing related protein (cce_0571) showed a 134-fold higher abundance in the dark under nitrogen-fixing conditions compared to nitrate supplemented conditions (Fig. 5a and b). NifW, a small nitrogenase stabilizing and protective protein with 116 amino acid residues (13.6 kDa), located within the 35 nitrogenase gene cluster, was 148-fold more abundant in the dark cycle without nitrate compared to with nitrate (Fig. 5b).

Crocosphaera 511422 contains hupSL genes, which encode an uptake hydrogenase induced in the dark under nitrogen-fixing growth15,24. The HupS and HupL subunits facilitate the uptake of molecular hydrogen (H2), which is converted into protons and electrons for ATP production or other reductive reactions. The abundance of HupL was 76 times and HupS was 8 times higher in the dark under nitrogen-fixing conditions compared to non-fixing conditions (Fig. 5b). The kaiABC gene cluster, responsible for encoding circadian clock proteins, exhibited elevated expression during nitrogen fixation. In particular, KaiA, KaiC1, and KaiC2 were upregulated under nitrogen-fixing conditions during L/D cycles. Remarkably, KaiC2 showed nearly a 9-fold increase in light in the absence of nitrate compared to when nitrate was present (Figure S2b).

Photosynthesis and carbon dioxide fixation

CO2 fixation predominantly involves the enzyme RuBisCo which operates during the Calvin cycle to synthesize carbohydrates. RuBisCo can also catalyze a reaction with oxygen (O2), leading to a process known as photorespiration. When RuBisCo reacts with O₂ instead of CO₂, it causes the loss of fixed carbon as CO₂, NH₃, and consumes ATP, thereby reducing the efficiency of photosynthesis. IcfA1 (cce_2257) was upregulated in the light cycle under nitrogen-fixing conditions. Similarly, bicarbonate transport system substrate-binding proteins (CmpA, CmpB, CmpC, and CmpD) were also upregulated in the light cycle, though unlike IcfA1, they were downregulated in the absence of nitrate (Fig. 4c; Table 1). SbtA (cce_2939) was downregulated in the dark compared to the light in both nitrogen-fixing and non-fixing conditions, regardless of nitrate presence. These findings align with previous observations except SbtA, which was shown to have higher abundance in the dark16,25. The RuBisCo large subunit (CbbL) was affected by L/D cycles but not by nitrate, however, the small subunit (RbcS) was affected by both the L/D and nitrate with RbcS downregulated under nitrogen-fixing conditions. Meanwhile, glycolate oxidase (cce_3708) expression depended on nitrate rather than L/D cycles. These observations suggest that carboxysome activity was influenced by the presence or absence of nitrate.

PSI and PSII proteins were downregulated in the dark compared to the light and again downregulated in the absence of nitrate (Table S3). The data in Table S3a for PSII proteins is one of the most comprehensive analyses of PSII protein levels under different environmental conditions. There are 4 copies of PsbA in Crocosphaera 51142 and each is regulated differently under L/D and nitrate concentrations. PsbA3 (cce_0267) is the major PsbA species under all 4 conditions studied and is about 2-fold higher in the presence of nitrate compared to the absence of nitrate under both L/D growth. PsbA1 (and PsbA5) has increased the most and is about 8-fold higher under N2-fixing conditions but is still ~ 5-fold lower in concentration than PsbA3. It is clearly strongly regulated by the nitrate level. On the other hand, PsbA4 is the lowest in abundance of all PsbA species but is upregulated in the dark in the presence of nitrate, as first demonstrated by semi-quantitative PCR26. It is strongly regulated in the dark and then further in the presence of nitrate. PsbA4 features alterations in residues important for binding and stabilizing the Oxygen-Evolving Complex, potentially maintaining PSII integrity without enabling O2 evolution. It is likely that PsbA4 is important in the presence of nitrate but superseded by PsbA1 under N2-fixing conditions. In contrast, most other PSII proteins, such as PsbD2, show minimal variations over the 4 conditions, with a ~ 50% increase in the presence of nitrate. The other key regulatory changes include PsbO and PsbU, which increase when nitrate is absent, linking them to regulation of O2 evolution (Table S3A). The phycobilisome complex was upregulated mostly in the light cycle but was also strongly influenced by nitrate availability with significant downregulation under nitrogen-fixing conditions. The cytochrome b6f complex involved in the electron transport chain, linking the PSII and PSI27 was upregulated in the dark under nitrogen-fixing conditions but was not different between LD cycles under non-fixing conditions (Table 1, Table S3).

Protein translation, folding, degradation and cellular homeostasis

Protein synthesis, chaperones, detoxifying proteins, and proteases play crucial roles in maintaining proteome homeostasis within cells28. Unlike ubiquitin proteasome system in eukaryotes, proteostasis in cyanobacteria is regulated by protein quality control (PQC) system depending on AAA + proteolytic machines29. Most of the identified 30S and 50S ribosomal proteins were downregulated under nitrogen-fixing conditions (Fig. 6a and b).

Fig. 6
figure 6

Response of proteins involved in ribosomal subunits, carbon concentrating mechanism and protease degradation. A. Heatmap of proteins involved in 30 S (A) and 50 S ribosomal subunit (B). Heatmap of proteins involved in carbon concentrating mechanism (C) and proteasomal degradation pathway(D). Hierarchical clustering of the proteins was performed by Euclidean distance measurement and average linkage. The average intensities of all the four replicates for each condition was used to plot these heatmaps.

Full size image

The cyanobacterial ribulose bisphosphate carboxylase (CbbL & RbcS) and CO2 concentrating mechanism (CCM) proteins have been identified as promising gene targets for increasing crop yield by improving photosynthesis and water use efficiency30. Most of these proteins showed higher abundances under nitrate compared to non-nitrate growth independent of L/D cycles (Fig. 6c). However, CcmK3, CcmK4, CcmM, and CcmO were more abundant in the dark than in the light when nitrate was present. Both RbcS and CbbL were more abundant in light than in the dark under both nitrate supplemented and nitrate-depleted growth (Fig. 6c). The SbtA which plays an important role in CO2 concentration process facilitated by CCM proteins31 was also more abundant in light than in the dark, independent of nitrate (Fig. 6c).

Protein homeostasis is tightly controlled by chaperones and proteases32. The small heat shock protein, HSP20 (HspA3; cce_5270), was 3.5-fold more abundant in the dark and 3.0-fold more abundant in the light when compared between nitrogen-fixing and non-fixing conditions (Table 1). The chaperone DnaK2 (cce_4004) also showed more than 7.5-fold higher abundance in the dark and more than 4.5-fold higher abundance in the light when compared between nitrogen-fixing and non-fixing conditions (Table 1). Nitrogen, not the L/D cycle, had a significant impact on the chaperonin GroEL (cce_1344 and cce_3314). While specific roles of chaperones and heat shock proteins in cyanobacteria have not been thoroughly investigated, increased abundances of HSP20 and GroEL under nitrogen-fixing conditions may suggest cellular stress responses.

We identified proteases including ATP-dependent Clp proteases (ClpX, ClpP, ClpP2, ClpP4, ClpC1), ATP-dependent zinc metalloproteases (Ftsh, FtsH2, FtsH4), CAAX- and carboxy-terminal proteases (CtpA, CtpB) (Fig. 6d; Table 1). These proteases were, in general, more abundant under nitrogen-fixing conditions than the non-fixing condition in both the L/D cycles. The CtpA (cce_3991), was more than 5-fold higher in nitrogen-fixing cells compared to non-fixing cells but its expression changed only by 20% between L/D (Table 1). The Clp proteases also showed significant upregulation under nitrogen-fixing condition with ClpP2 showing almost 5-fold higher abundance (Table 1). Among the four zinc metalloproteases, FtsH4 (cce_1593) showed a 2-fold increase in the dark under nitrogen-fixing compared to non-fixing conditions.

Comparison between transcript and protein expression

Gene transcripts alone explain about one-third of a cell’s phenotype, while proteins define around 40%33, showing that transcriptomics and proteomics are complementary. Poor correlations between transcript and protein levels are often due to factors like timing differences, post-translational modifications, stability, translational efficiency, and protein turnover. In cyanobacteria, mRNA and protein correlations are less explored. We compared our proteomic data (a single time point, 6 h into the dark cycle) with published transcriptomic data at three dark cycle time points (1 h, 5 h, and 9 h) under nitrogen-fixing conditions34. Of the 868 proteins identified that matched transcriptomic data, 158 showed significant differences across L/D cycles. Fold changes for transcripts and proteins were calculated and plotted using Pearson’s correlation (Fig. 7).

Fig. 7
figure 7

Comparison of proteomics and transcriptomics data. Scatterplots show the comparison of proteomics and transcriptomics data. The left panel shows the correlation of all the identified proteins in the dark cycle without nitrate (D-) condition that were common between the proteomic and transcriptomic data, and the right panel shows the correlation between significantly changing proteomics data in D- compared to the pooled control of transcriptomics data. Proteomics data at 6 h into the dark was compared with transcriptomics data collected at 1 h, 5 h, and 9 h into the dark respectively. The color of the data points indicates if the proteins were identified with a threshold fold change of ± 1.5 (or log (foldchange = ± 0.58) under both conditions (purple), protein only(blue), transcript only(green), or none (grey).

Full size image

The comparison showed strong correlations between transcriptomic data at one and five hours into the dark cycle34 with our proteomics data, with the strongest correlation observed between one-hour transcript and six-hour protein data (Fig. 7, Table S8). As anticipated, there was no correlation between proteomic data and the nine-hour transcriptomic data. Among nitrogen-fixing proteins, nitrogenase (Nif) enzymes displayed a good correlation with their transcripts across all time points, especially with five-hour transcript data (Figure S3). This analysis highlights that while transcript peaks often occur early in the dark cycle and remain stable, the relationship between transcript and protein levels varies by pathway.

Discussion

We conducted a proteomic analysis of the unicellular diazotrophic cyanobacterium Crocosphaera 51142 to deepen our understanding about its metabolism during light-dark cycles under nitrogen-fixing and non-fixing conditions. We identified 2,050 high-confidence proteins, with the highest numbers under nitrogen-fixing conditions in the dark, highlighting increased metabolic demands of cells during BNF. Key findings include coordination among active respiration, glycogen metabolism, and nitrogen-fixation during L/D cycles. Increased expression of glycogen synthase (GlgA1 and GlgA2) in the light cycle and glycogen phosphorylase (GlgP1 and GlgP2) in the dark under nitrogen-fixing conditions (Table 1) highlight their role to support ATP production and to create low-oxygen environment conducive for BNF6,9. There are two glycogen debranching enzymes in cyanobacteria GlgP and GlgX7,8,35,36. Identification of both enzymes, with differential responses to L/D cycles and nitrate availability (Table 1) suggest fine tuning of glycogen accumulation and degradation in Crocosphaera 51,142 to optimize metabolic activity during nitrogen fixation.

In Crocosphaera 51,142, most of the 34 nitrogen-fixation genes are located on a contiguous 28-kb cluster12,37, indicating a highly coordinated expression of enzymes involved in Nif assembly or nitrogen-fixing activities (Fig. 5). Under N2-fixing conditions, we observed a large increase in HupSL, which is critical for protecting NifHDK under oxic conditions38. HupSL, along with respiratory and hydrogenase enzymes, was significantly upregulated during the dark phase, highlighting tightly integrated regulation of respiration, nitrogen fixation, and redox balance to manage oxygen levels and energy demands. The activation of alternative metabolic pathways, such as oxidative pentose phosphate (OPP) pathway and 2-oxoglutarate (2-OG) metabolism, further underscores the adaptation of Crocosphaera to enhance ATP production for nitrogen fixation. Additionally, significant upregulation of NifT and NifW was observed. While specific function of NifT remains unclear39,40, prior studies suggest it may suppress nitrogenase activity, as deletion of NifT increased nitrogenase function without affecting diazotrophic growth41,42. These results provide insights into proteomic adaptations of Crocosphaera 51142 to nitrogen availability and light-dark cycles, warranting further studies on the role of NifT in cyanobacteria.

The HesA (cce_0569) and HesB (cce_0570), conserved in diazotrophic cyanobacteria and located within the nitrogenase gene cluster40, were more abundant in the dark under nitrogen-fixing conditions compared to nitrogen-sufficient conditions (Fig. 5A, Table S5). While their actual function in nitrogenase activity is unknown, recent study indicates their involvement in the efficient production of the MoFe protein40. The uptake hydrogenase complex, HupSL, crucial for redox balance during BNF, recycles H2 produced by nitrogenase, providing electrons and protecting nitrogenase from O2 inactivation45,46. Circadian clock proteins KaiA, KaiC1, and KaiC2, encoded by the kaiABC cluster, were upregulated under nitrogen-fixing conditions, suggesting that their role in linking circadian regulation to metabolic processes essential for nitrogen-fixing efficiency.

Proteomic and transcriptomic34 comparisons revealed a time delay between transcription and protein expression, with a strong correlation between transcripts at one hour and proteins at six hours into the dark cycle. However, no correlation was observed at nine hours, indicating temporal shifts influence protein expression. Nitrogenase transcripts and proteins showed high correlation throughout the dark cycle, suggesting tightly regulated nitrogenase expression. These findings underscore the value of integrating transcriptomic and proteomic approaches to understand cyanobacterial nitrogen fixation and cellular adaptations to environmental changes.

Interestingly, a closer look at transcripts and proteins responsible for nitrogen fixation showed a high level of correlation between their transcripts and protein levels at all time points, indicating stability of nitrogenase transcripts and consistent protein production throughout the dark cycle. This was particularly evident for nitrogenase enzymes, where transcript levels at five hours strongly correlated with protein abundance, while early and late transcripts showed less correlation. This suggests that nitrogenase expression is tightly regulated and may follow a distinct temporal pattern during the dark cycle. These findings enhance our understanding of the coordination between gene expression and protein abundance in cyanobacteria, especially during nitrogen fixation. The data highlights the importance of integrating both transcriptomic and proteomic approaches to fully capture the complexity of cellular processes under different environmental conditions.

Conclusions

This study investigates how nitrogen and light-dark phases impact the Crocosphaera 51142 proteome, revealing significant modulation of metabolic pathways. Significant changes in abundances of CoxB1 and GlgP1, nitrogenase, and uptake hydrogenase suggests that cells maintain suboxic conditions for nitrogenase activities through increased respiration and glycogen metabolism. While previous studies have reported 10-40fold upregulation of Nif proteins during the dark cycle, our data revealed an unprecedented dynamic range in Nif protein expression from a 7fold to a striking 450-fold increase, after six hours into the dark cycle under nitrogen-fixing growth condition. Nif protein levels sharply declined during the light phase, indicating an active degradation to maintain proteostasis by protease activities. However, the mechanisms by which cyanobacteria maintain NIF protein homeostasis under highly dynamic growth conditions remain unclear. Increased protease activity under nitrogen-fixing conditions suggests their critical role in proteome regulation, yet the specific proteases involved in Nif protein degradation are unknown. Future research on protease activity, substrate specificity, and post-translational modifications will advance our understanding of proteome regulation and nitrogenase dynamics.

Materials and methods

Cell growth and experimental conditions

The original Crocosphaera sp. subtropica ATCC 51,142 strain was isolated from the intertidal sands of the Texas Gulf coast, Port Aransas, Texas and subsequently maintained in the Sherman lab1. For this study, cells were obtained from the ATCC (American Type Culture Collection) and maintained in ASP2 medium with 17.6 mM NaNO3 and 30 µmol photons m−2 s−1 of continuous light in our lab. One-twentieth (5 ml) of the Crocosphaera 51,142 stock was first inoculated to each 250-ml flask containing 100 mL ASP2 medium with or without 17.6mM NaNO3 and allowed to grow for 7 days on a shaker at 125 rpm, 30 °C, and 30 µmol photons m−2 s−1 of continuous light (LL). After seven days, cultures were transitioned to alternative 12 h-light/12 h-dark (30 µmol photons m−2 s−1) diurnal cycles and grown for additional seven days before harvesting. Cells were harvested at 6 h into the light period (L) or 6 h into the dark period (D). For each sample, four replicates were harvested. For example, for cells collected 6 h into the dark in media supplemented with nitrate (D+), the replicates were called “D1+”, “D2+”, “D3+” and “D4+” respectively, and the replicates for the cells grown in media without nitrate (D-) were denoted as “D1-”, “D2-”, “D3-”, “D4-” respectively. Similarly, cells collected six hours into the light cycle, with nitrate denoted as L+: L1+, L2+, L3+, L4 + and without nitrate L-: L1-, L2-, L3-, L4- respectively.

Fifteen15 ml of cell cultures were collected by centrifugation at 3,220 × g in 15 ml tubes (Corning), washed with 1 ml of 50 mM HEPES-KOH buffer (pH7.5), and then centrifuged again at 10,000 × g at 4 °C for 15 min to collect cell pellets. The pellets were resuspended in 200 µL of HEPES/KOH (pH7.8) buffer, supplemented with 1 mM phenylmethylsulfonylfluoride (PMSF) protease inhibitor and homogenized in Precellys VK 0.5 tubes (Bertin Corp., Rockville, MD, USA), 3× at 6000 rpm for 3 × 20 s in each cycle followed by probe sonication. Protein concentration was determined by bicinchoninic acid (BCA) assay (Pierce Chemical Co., Rockford, IL, USA). Following BCA assay, cell lysates corresponding to 200 µg of total protein (equivalent volumes) were ultracentrifuged at 150,000 × g for 20 min to divide proteins into soluble and insoluble fractions. The soluble fractions were acetone precipitated with four volumes of cold acetone and incubated overnight at −20 °C, while the insoluble pellets were resuspended in 200 µl of HEPES/KOH (pH 7.5), bath sonicated for 5 min and then acetone precipitated with four volumes of cold (−20 °C) acetone. The soluble and insoluble protein pellets were collected by centrifuging at 17,200 × g for 20 min at 4 °C, washed 3× with 80% cold (−20 °C) acetone and prepared for LC-MS/MS analysis as described below.

Protein extraction and proteolysis

Both soluble and insoluble pellets were resuspended in 20 µl buffer of 8 M urea and solubilized by incubating at room temperature with continuous vortexing for an hour, and 0.1% Rapigest (Waters, MA) was added to the insoluble pellets only with the urea solution. The samples were reduced with 10mM dithiothreitol (DTT) at 37 °C for 45 min and then cysteines alkylated with iodoethanol mix (195 µL acetonitrile, 4 µL iodoethanol and 1 µL triethyl phosphine) for 45 min at 37 °C in a dark. After reduction and alkylation, samples were dried in a vacuum centrifuge (Vacufuge Plus, Eppendorf, Enfield, CT) at 45 °C, reconstituted in 150 µL of 50 mM ammonium bicarbonate and then digested with trypsin at a 1:25 enzyme to substrate ratio. High pressure digestion was performed using a Barocycler (Pressure Bioscience INC., Easton, MS, USA) at 50 °C with 60 cycles, each cycle consisting of 50 s at 20,000 PSI and 10 s at 1 atm) as described before43,44. Digested peptides were cleaned using Pierce Peptide Desalting Spin Columns (Thermo Fisher Scientific, Waltham, MA, USA). Eluted clean peptides were dried in a vacuum centrifuge and reconstituted in 20 µL 0.1% formic acid (FA) in 3% acetonitrile. The peptide concentration was measured using a nanodrop spectrophotometer (ThermoFisher Scientific), adjusted the final concentration in each sample to 1 µg/µl, and 1 µg (1 µl) was used for proteomics analysis of the highest concentrated fraction of each sample (either soluble or insoluble), and 0.5 µg was injected for the lower concentrated fraction.

Liquid chromatography tandem mass spectrometry analysis

One µg (1 µl) and 0.5 µg (of the other fraction) of clean peptides were analyzed by reverse-phase HPLC separation using a Dionex UltiMate 3000 RSLC nano system, coupled to a Orbitrap Fusion Lumos mass spectrometer via Nanospray Flex™ electrospray ionization source (Thermo Fisher Scientific) as described previously44,45. Briefly, peptides were first loaded into a PepMap C18 trap column (3 μm × 75 μm ID × 2 cm) (Thermo Fisher Scientific, Waltham, MA, USA), and then separated using a reverse phase 1.7 μm 120 Å IonOptics Aurora Ultimate C18 column (75 μm x, 25 cm). The column was maintained at 50 °C, mobile phase solvent A was 0.1% FA in water, solvent B was 0.1% FA in 80% ACN. The loading buffer was 0.1% FA in 2% ACN. Peptides were loaded into the trap column for 5 min at 5 µl/min, then separated with a flow rate of 400 nl/min using a 130 min linear gradient. The concentration of mobile phase B was increased linearly to 8% in five minutes, 27% B in 80 min, and then 45% B at 100 min. After 100 min, it was subsequentially increased to 100% of B at 105 min and held constant for another 7 min before reverting to 2% of B in 112.1 min and maintained at 2% B until the end of the run. The mass spectrometer was operated in positive ion and standard data dependent acquisition (DDA) mode. The spray voltage was set at 2.8 kV, the capillary temperature was 320 °C and the S-lens RF was set at 50. The resolution of Orbitrap mass analyzer was set to 60,000 and 15,000 at 200 m/z for MS1 and MS2, respectively, with a maximum injection time of 100 ms for MS1 and 20 ms for MS2. The full scan MS1 spectra were collected in the mass range of 350–1600 m/z and the MS2 first fixed mass was 100 m/z. The automatic gain control (ACG) target was set to 3 × 106 for MS1 and 1 × 105 for MS2. The fragmentation of precursor ions was accomplished by higher energy C-trap collision dissociation (HCD) at a normalized collision energy setting of 27% and an isolation window of 1.2 m/z. The DDA settings were for a minimum intensity threshold of 5 × 104 and a minimum AGC target of 1 × 103. The dynamic exclusion was set at 15 s and accepted charge states were selected from 2 to 7 with 2 as a default charge. The exclude isotope function was activated.

Data analysis

LC–MS/MS data were processed with MaxQuant software (Ver 2.0.3.0)46,47. Raw spectra were searched against the Crocosphaera 51,142 protein sequence database obtained from the UniProt (release-2022_04/2022-12-05) containing 5403 protein sequences, for protein identification and MS1 based label-free quantitation. The minimum length of the peptides was set at six AA residues in the database search. The following parameters were edited for the searches: precursor mass tolerance was set at 10 ppm, MS/MS mass tolerance was set at 20 ppm, enzyme specificity of trypsin/Lys-C enzyme allowing up to 2 missed cleavages, oxidation of methionine (M) as a variable modification and iodoethanol of cysteine as a fixed modification. The decoy reverse database was considered for data analysis and to control false discovery rate (FDR) and was set at 0.01 (1%) both for peptide spectral match (PSM) and protein identification. The unique plus razor peptides (non-redundant, non-unique peptides assigned to the protein group with most other peptides) were used for peptide quantitation. Only proteins detected with at least one unique peptide and MS/MS ≥ 2 (spectral counts) were considered valid identifications. Label-free quantitation (LFQ) intensity values were used for relative protein abundance comparisons.

Bioinformatics data analysis

We mainly performed data analysis in Perseus (version 1.6.0.9)48, Microsoft Excel and data visualized using OriginPro (Version 2022, OriginLab Corporation, Northampton, MA, USA), InteractiVenn49, Morpheus (https://software.broadinstitute.org/morpheus). Log2-transformed LFQ intensities (protein intensities) were used for further analysis. Coefficients of variations were calculated for raw protein intensities of Hela digest using triplicate runs to determine the reproducibility of LC-MS/MS analysis and label-free quantitation. Data sets were filtered to make sure that identified proteins showed expression in at least three out of four biological replicates of at least one treatment group and the missing values were subsequently replaced by imputation in Perseus that were drawn from a normal distribution. Principal component analysis of treatment effects and biological replicates was performed as described in50. Multi-sample test (ANOVA) for determining if any of the means of differentiation stages were significantly different from each other was applied to protein data set. For hierarchical clustering and heatmap generation of significant proteins, mean protein abundances of biological replicates were z-scored and clustered using Euclidean as a distance measure for row clustering. Significantly upregulated or downregulated proteins between the treatment groups (± nitrate, L/D, and nitrate × L/D) were determined by ANOVA and a two tallied student’s t-test. Differentially expressed proteins were determined with p-value < 0.05.

Gene ontology analysis

We used three sequence-based function prediction methods: PFP51, Phylo-PFP52, and Extended Similarity Group method (ESG)53 to assign Gene Ontology (GO) terms54 to protein-coding genes. The PFP algorithm scores GO terms based on Expect (E)-values of sequences with those GO terms retrieved from the UniProt sequence database by PSI-BLAST55. It then propagates the scores to parental terms on the GO Directed Acyclic Graph (DAG) according to the number of database sequences annotated with parent (e.g., metabolic process) and child terms (e.g., carbohydrate metabolic process), with a confidence score assigned from benchmark validation results.

Phylo-PFP, which is an improvement over PFP, improves the performance by incorporating phylogenetic information. The ESG method performs iterative sequence database searches and annotates a query sequence with GO terms. Each annotation is given a probability based on how similar it is to other sequences in the protein similarity graph. To ensure the inclusion of meaningful GO term annotations, we focused exclusively on predictions characterized by high confidence, and all GO terms exceeding confidence score cutoffs of 20,000 for PFP, 0.7 for Phylo-PFP, and 0.7 for ESG were incorporated into our analysis. To improve high-confident prediction, we combined results from three prediction methods and presented the consolidated GO term annotations. Each result file has information about Protein ID, GO ID, Depth, Class, and GO Description. The depth refers to the depth of GO ID in the GO DAG, and class refers to GO functional category (f - molecular function, p- Biological process, c- Cellular Component), and GO Description describes the predicted GO term. The Gene Ontology release 2021-11-16 was used for this analysis.

Correlation of proteomics and transcriptomics data

Transcriptomics data was obtained from Stockel et al.34 and matched with our proteomics data. Like the transcriptomics “pooled control”, the proteomics “pooled control” was the average of the intensities of all nitrate-depleted samples. The ratio of D- to the pooled control was then used to calculate the fold change (FC) of proteins. The log2(FC) of these values were plotted for both the proteomics and transcriptomics data using scatterplots in OriginPro (Version 2022, OriginLab Corporation, Northampton, MA, USA). A threshold of ± 1.5 (or log (foldchange = ± 0.58) was used to decide if the fold change was significant enough. The data was divided into various categories from the plots, and colored accordingly, such as those indicated in both proteomics and transcriptomics data with a threshold fold change of ± 1.5, or in any one of these, or none. Then, the GO enrichment was performed for these various categories to provide insights into the molecular regulation of several proteins of interest, such as those of the nitrogenase gene cluster, proteases, and ribosomal proteins. The spider chart in Fig. 5a was created using Origin 2022 software by selecting the “Radar chart” option from the “Plot” menu. The x-axis represents different nif genes, while y-axis corresponds to their log-transformed LFQ Intensities of proteins.