If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
PPIs in two types of cells of Anabaena sp. 7120 were systematically identified.
10,302 and 8557 high-confidence PPIs were obtained and over 80% were novel.
About 438 proteins showed significant changes in vegetative cells and heterocysts.
Protein Alr4359 was found to influence the diazotrophic growth of filaments.
The filamentous cyanobacterium Anabaena sp. PCC 7120 can differentiate into heterocysts to fix atmospheric nitrogen. During cell differentiation, cellular morphology and gene expression undergo a series of significant changes. To uncover the mechanisms responsible for these alterations, we built protein–protein interaction (PPI) networks for these two cell types by cofractionation coupled with mass spectrometry. We predicted 280 and 215 protein complexes, with 6322 and 2791 high-confidence PPIs in vegetative cells and heterocysts, respectively. Most of the proteins in both types of cells presented similar elution profiles, whereas the elution peaks of 438 proteins showed significant changes. We observed that some well-known complexes recruited new members in heterocysts, such as ribosomes, diflavin flavoprotein, and cytochrome c oxidase. Photosynthetic complexes, including photosystem I, photosystem II, and phycobilisome, remained in both vegetative cells and heterocysts for electron transfer and energy generation. Besides that, PPI data also reveal new functions of proteins. For example, the hypothetical protein Alr4359 was found to interact with FraH and Alr4119 in heterocysts and was located on heterocyst poles, thereby influencing the diazotrophic growth of filaments. The overexpression of Alr4359 suspended heterocyst formation and altered the pigment composition and filament length. This work demonstrates the differences in protein assemblies and provides insight into physiological regulation during cell differentiation.
The filamentous cyanobacterium Anabaena sp. PCC 7120 (hereafter Anabaena sp.) is a model organism for studying cell differentiation, cell–cell adhesion, and intercellular communication. In the presence of combined nitrogen, vegetative cells exhibit phenotype of the filaments and fix CO2 through oxygenic photosynthesis. In the absence of combined nitrogen, some vegetative cells differentiate into heterocysts for nitrogen fixation (
In order to fix nitrogen, heterocysts need an efficient nitrogenase enzyme, which is extremely sensitive to oxygen. Therefore, heterocysts have evolved different strategies to prevent oxygen from the environments interfering with intracellular reactions (
). However, the mechanisms responsible for thylakoid membrane reorganization and whether additional proteins participate in this process require further investigation.
In the diazotrophic filament, vegetative cells provide heterocysts with a carbon source in the form of sucrose, and heterocysts provide vegetative cells with combined nitrogen in the forms of glutamine and β-aspartyl-arginine (
). The septal junctions contain the proteins SepJ, FraC, and FraD. The structure of the septal junction was recently recovered by cryo-EM as containing a cap, a plug, and tube modules, and it was found to undergo reversibly controlled material communication under stress (
Generally, intracellular biological processes rely on a series of physical associations among molecular substances, especially proteins. The PPI network is a powerful tool for exploring the fundamental metabolism of living organisms (
). The PPI network may also help us better understand the difference between vegetative cells and heterocysts and find more functional proteins related to cell differentiation. Several approaches have been developed for the identification of PPIs at the proteome scale (
), such as yeast two-hybrid (Y2H), affinity purification followed by mass spectrometry (AP–MS), and cofractionation coupled with mass spectrometry (CoFrac–MS). The similarity of protein's elution profiles was the principal character in the CoFrac–MS experience. Different computational approaches were applied to distinguish elution profiles and generate the predicted complexes, such as hierarchical clustering and machine learning (
In this study, we constructed PPIs of vegetative cells and heterocysts in Anabaena sp. using CoFrac–MS. We compared the protein interaction networks to provide a better understanding of cell differentiation between vegetative cells and heterocysts. Furthermore, we found the redistribution of the hypothetical protein Alr4359 and its function in heterocyst formation.
Cell Culture and Heterocyst Purification
The Anabaena sp. PCC 7120 was cultured in liquid BG11 (with combined nitrogen) or BG110 medium (without combined nitrogen) under continuous illumination of 30 to 40 μmol m−2 s−1 at 28 °C. To induce heterocyst, cyanobacterial cells were harvested at the exponential growth phase in BG11 and then incubated in BG110 medium for 48 h after being washed twice using BG110 medium. To purify the heterocysts, the filaments containing heterocysts were harvested by centrifugation (3000g at 4 °C for 5 min) after 48 h of induction in BG110 medium. The isolation process was based on previous studies (
). The pellet was resuspended in 8% sucrose, 5% Triton X-100, 50 mM EDTA, pH = 8.0, 50 mM Tris–HCl, and pH = 8.0 containing lysozyme (1 mg/ml) at 4 °C and vortexed vigorously for 2 to 3 min at room temperature. The suspension was mildly sonicated for approximately 2 min on ice to break the vegetative cells, whereas the heterocyst remained intact during this procedure. The heterocysts were collected by centrifugation at 3000g for 5 min at 4 °C and washed twice in 8% sucrose, 50 mM EDTA, pH = 8.0, 50 mM Tris–HCl, and pH = 8.0 at 4 °C. The isolated heterocysts were assessed by microscopy, and contamination by vegetative cell contents was tested by measuring the concentration of ribulose-1,5-bisphophate carboxylase/oxygenase.
Protein Extraction and Separation
Cells from vegetative cells or heterocysts were suspended in lysis buffer containing 20 mM Tris-Cl (pH = 7.5), 150 mM NaCl, 1% dodecylmaltoside, and Complete Protease Inhibitors EDTA-free (Roche), and sonicated on ice with an output of 135 W. The whole-cell lysate was centrifuged (10,000g at 4 °C for 10 min) to remove the cell debris. The protein concentration was determined using the Bradford assay. Then, 300 μg of protein mixtures of vegetative cells and heterocysts were individually fractionated by size-exclusion chromatography (SEC) using a Thermo Scientific Ultimate 3000 HPLC system. The lysates separated by SEC were injected into a Superose 6 10/300GL column (GE Life Sciences) equilibrated with PBS (pH = 7.2) and exposed to 120 min of isocratic elution. The total collection time was 55 min, as the first fraction was collected at 20 min and the last fraction finished at 75 min. In total, 55,300 μl fractions were collected, with a flow rate of 0.3 ml min−1.
Protein Digestion and Desalting
The proteins from all the HPLC fractions were denatured at 95 °C for 10 min. For the LC–MS/MS analysis, each fraction was reduced with 10 mM freshly prepared DTT at 37 °C for 45 min, and the cysteines were alkylated with 15 mM iodoacetamide at 37 °C in the dark for 30 min. Trypsin (Promega) was added, and the samples were incubated overnight at 37 °C. Each fraction was desalted using ZipTip C18 plates (Millipore) following the manufacturer’s protocol. Peptides were dried using a LABCONCO evaporate and resuspended in 0.1% formic acid.
LC–MS/MS Analysis and Data Processing
The peptides were analyzed by online nanoflow LC–MS/MS using an Easy-nLC 1200 system connected to a Q-Exactive Plus hybrid quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific). Peptides in 0.1% formic acid were injected onto a C18 column (75 μm × 15 cm, 3 μm, 100 Å) and eluted at a flow rate of 300 nl/min with a 100 min gradient from 10% solvent B (90% acetonitrile/0.1% formic acid, v/v) to 80% solvent B. The peptides were ionized by nanoelectrospray at 2.0 kV and analyzed with higher-energy collisional dissociation fragmentation. The MS/MS spectra of the top 20 most-abundant precursor ions were acquired using a data-dependent method. The dynamic exclusion duration was 40 s with a repeat count of 1 and a ±10 ppm exclusion window. Automatic gain control was used to prevent the overfilling of the ion trap, and 5 × 104 ions were accumulated for the generation of MS/MS spectra.
The RAW data files were analyzed using Proteome Discoverer 2.1 (Thermo Fisher Scientific). The database containing 6175 entries supplied with the Anabaena sp. PCC 7120 proteome (UP000002483) was from the UniProt database. The mass tolerance of the precursor ions was set to 10 ppm, and the MS/MS mass tolerance was set at 0.02 Da. The enzyme was set as trypsin, allowing up to two missed cleavages. The carbamidomethyl modification of cysteines was set as a fixed modification, and methionine oxidation and protein N-terminal acetylation were set as variable modifications. The false discovery rate for protein-level and peptide-level identification was set at 1%, using a target-decoy–based strategy.
Machine learning was performed using EPIC (elution profile–based inference of complexEPIC) following the manufacturer’s instructions. The command line is: python/EPIC/src/main.py -s 1110011 -c ../complexes.txt. -r 0.75 -R 0.5 -e 3 -E 5 ../InputFolder/ ../OutputFolder/. The interpretation of each parameter can be found at https://github.com/BaderLab/EPIC. The standard true-positive protein complexes (supplemental Table S3) were manually collected using the experimental data in the STRING database. The input data are mass spectrometry (MS) results in supplemental Table S1. We created candidate protein-pair datasets in which only the protein pairs with at least one elution profile similarity score greater than 0.75 were retained. Then finally, PPIs (supplemental Table S4) were generated from these candidate protein-pair datasets using the random-forest machine-learning classifier. ClusterONE served as the clustering algorithm to predict the final protein complexes (supplemental Table S5). The visualization of protein complexes was built by Cytoscape 3.8 (https://cytoscape.org/).
Construction of Overexpression Strains
The full-length target genes were amplified by PCR using Anabaena sp. chromosome DNA as the template, and they were cloned into the plasmid pRL25N at the SmaI site, which contains the Cu2+-inducible promotor petE and GFP ORFs. The plasmid pRL25N was donated by Professor Xiang Gao. The resultant plasmids were validated by PCR and sequencing analysis and then introduced into the Anabaena sp. via conjugative transformation from an HB101 carrying the cargo plasmid and the helper plasmid.
The primers used in this study are as follow:
The cell lysates of the GFP-tag strains were subjected to affinity purification using an anti-GFP antibody (catalog no.: ab290; Abcam). The antibody purification was performed using Protein A MagBeads following the manufacturer’s instructions (GenScript). In brief, 2 μl GFP antibody was first incubated with 100 μl magnetic beads for 30 min. The magnetic beads were washed by 1 ml binding buffer (20 mM Na2HPO4, 0.15 M NaCl, pH = 7.0) twice to remove the free antibody. The cell lysate that contained about 500 μg proteins was mixed with magnetic beads binding with GFP antibody and was incubated for 1 h at room temperature. After removing the dissociated proteins from the solution and washing three times with PBS, the target protein and its interaction partners were released using 0.1 M glycine solution (pH = 2–3). The neutralization buffer (1 M Tris, pH = 8.5) was added to each eluate to neutralize the pH at the ratio of 1:10. Finally, the sample was denatured at 95 °C for 10 min, digested with 0.3 μg trypsin, and analyzed by MS. The LC/MS/MS and database search were the same as the fractionation samples mentioned previously, except that the phosphorylation modifications of serine, threonine, and tyrosine were set as variable modifications. The SAINT software (https://sourceforge.net/projects/saint-apms/files/) analyzed all the results with default settings, and the command line is SAINTexpress-spc inter.dat prey.dat bait.dat.
Yeast Two-Hybrid Assays
Full-length fraH and alr4359 were amplified from Anabaena’s genome by PCR with the NdeI restriction site at their N-terminal region, the XhoI restriction site at the fraH C-terminal region, and the PstI restriction site at the alr4359 C-terminal region. Then, the gene fragments of fraH and alr4359 were cloned into the AD-vector pGADT7 and the BD-vector pGBKT7 to generate pGADT7-fraH and pGBKT7-alr4359, respectively. The plasmids of pGADT7-T and pGBKT7-53 were used as positive controls. pGADT7-T and pGBKT7-lam were negative controls. The recombinant plasmids were cotransformed into Saccharomyces cerevisiae AH109 and grown on Synthetic Dropout Medium/-Trp-Leu-His agar plates at 28 °C for 3 days. This experiment was repeated twice. The primers used in this experiment are as follows:
Confocal Microscopy and Electron Microscopy
Anabaena cells were visualized with a Leica SP8 confocal microscope. GFP was excited using 488 nm laser irradiation. The fluorescent emission was monitored by collection across windows of 498 to 541 nm for GFP imaging and 630 to 700 nm for cyanobacterial autofluorescence. For electron microscopy imaging, the Alr4359-GFP overexpression strain cultured in BG11 was harvested, washed twice with fresh BG11, and prepared by the method presented by Merino-Puerto et al. (
). The samples were examined with a Hitachi HT-7700 electron microscope at 120 kV.
Experimental Design and Statistical Rationale
Three independent SEC separation experiments were conducted in vegetative cells and heterocyst. In each experiment, 300 μg of protein mixtures were injected into the SEC column, and a total of 55 fractions were collected for one fraction per minute. The SEC reproducibility of two cell types was evaluated based on the Pearson correlation coefficient. The false discovery rate for protein-level and peptide-level identification was set at 1%. After LC/MS/MS analysis and database search, the results for each experiment were summarized into one matrix, where each row represents a protein and each column contains the spectral counts of proteins for the corresponding fraction. The peak shift of proteins was determined by fold change >1.5 with t test p value <0.05, which reflected shifting of average elution peak position in heterocyst to vegetative in three replicates. The protein elution profiles based on spectra counts were put together and run in EPIC software for protein complex prediction. The final protein complex datasets were generated by combining three independent experiments based on the in-build algorithm.
For AP–MS experiments, three independent replicates for one bait protein were performed in vegetative cells and heterocyst. The empty vector served as a negative control was also treated three times in each cell type. The prey proteins in vegetative cell’s experiment or heterocyst’s experiment were analyzed by SAINT software, which prey proteins with SAINTscore more than 0.6 were reserved and used for further analysis
Identification of Proteins in Vegetative Cells and Heterocysts
The filaments of Anabaena sp. consist of vegetative cells, which specialize into heterocysts with no red spontaneous fluorescent signals in the absence of combined nitrogen (supplemental Fig. S1). Heterocysts were purified from the filaments without vegetative cell contamination, as verified by fluorescence imaging (supplemental Fig. S2A) and Western immunoblotting (supplemental Fig. S2B). Cellular extracts of vegetative cells or heterocysts containing native proteins and protein complexes were fractionated by SEC and quantified by LC–MS/MS (Fig. 1A). Most of the fractions had a decent reproducibility between biological replicates in two cell types with Pearson correlation coefficients over 0.85 (supplemental Fig. S3, A and B). The correlation coefficients of fractions between two types of cells were slightly lower than in the same cell type (supplemental Fig. S3C). Besides, the correlation coefficients of each protein in two replicates were also calculated (supplemental Fig. S4). High abundant proteins presented higher correlation coefficients in both types of cells.
Finally, a total of 2770 Anabaena sp. proteins were identified in both cell types with the peptide-spectrum matches ≥2. Of these, 532 proteins were specific to vegetative cells, and 351 proteins were specific to heterocysts (Fig. 1B and supplemental Table S1). We analyzed the over-representation in the GO annotations of these proteins. For the proteins identified in the vegetative cells, only one biological process—methylation—was enriched, with 21 proteins in this term (Fig. 1C). We note that five of these 21 proteins contained a predicted tetrapyrrole methylase domain that was able to catalyze the methylation of different porphyrin compounds and participate in chlorophyll metabolism in cyanobacteria (Fig. 1D). These results indicate that protein methylation plays an important role in regulating the pigment composition in vegetative cells. The nitrogen fixation process was significantly enriched for the 351 proteins identified in heterocyst (Fig. 1C). The biological process of DNA-templated transcription regulation was also found in heterocysts. By analyzing the protein compositions associated with the term, we found five proteins belonging to group 2 sigma factor of RNA polymerase, containing the RNA_pol_sigma70 domain (Fig. 1D). The group 2 sigma factors are known as alternative sigma factors, and it has been reported that they are required for normal growth under nitrogen stress (
In addition, we observed that high-affinity potassium ion transport proteins, such as the Ktr/Trk-type K+ uptake transporter families (Ktr/Trk) and ATP-dependent transporters specific for K+ families (Kdp), were activated in heterocysts (Fig. 1, C and D). Potassium ions are the dominant intracellular cations and play essential roles in turgor homeostasis and pH regulation (
). The extra expression of K+ uptake transporters indicated a massive demand for K+ in heterocysts, but the biological function associated with this has not been elucidated.
Comparison of Protein Elution Profiles Between Vegetative Cells and Heterocysts
To thoroughly compare the elution profiles of the proteins in the two cell types, we performed an unsupervised hierarchical clustering analysis of the average protein intensities of relatively highly abundant proteins (peptide-spectrum matches ≥3) (Fig. 2A). Generally, most of the proteins in both cell types showed similar elution profiles, but some proteins showed a distinct peak shift. For example, most chaperone proteins or 30S ribosomal proteins were found in two types of cells with similar elution profiles, and the elution peak remained unchanged, but proteins such as DnaJ or uS10 (also known as RpsJ) (
). To identify elements related to variations in function, we searched for protein inconsistencies between vegetative cells and heterocysts. After calculating the peak shifts, we detected 438 proteins that showed significant peak changes in two cell types (Fig. 2C and supplemental Table S2). GO analysis revealed that these proteins were mainly involved in transcription and translation (Fig. 2D). DNA-directed RNA polymerase complexes are stable and tightly linked in both vegetative cells and heterocysts (supplemental Fig. S5A). Their elution peaks shifted toward a higher molecular weight range, and thus they may bind with new components in heterocysts. For translation processes, the main peak-shifted proteins were a series of amino acid-tRNA ligases (supplemental Fig. S5B), such as AlaS, GlyS, and ValS. It suggests that the regulation of amino acid-tRNA ligases is necessary for the differences in protein synthesis between the two cell types.
In addition to proteins with different elution profiles between the two cell types, we also observed that certain protein complexes were only identified in either vegetative cells or heterocysts (Fig. 2E). In vegetative cells, two zeta-carotene desaturases, All7255 and CrtQ, were identified and presented similar elution profiles that regulated the carotene biosynthetic pathway (
). An analysis of the whole-cell absorption spectra of vegetative cells and heterocysts shown that the ratio of carotenoids to chlorophyll differed significantly, and the absorption peak of phycobilin almost disappeared in heterocysts (Fig. 2F). Nitrogenases and hydrogenase formation/maturation proteins were identified in heterocysts (Fig. 2E). Most of the components of the nitrogenase or Hup families had similar elution profiles, but several proteins, such as NifV1 and HupD, differed from others, indicating loose and weak bonds with the complex. The hydrogenase expression/formation/maturation proteins are usually required for the maturation of hydrogenase, a nickel metalloenzyme that catalyzes the reversible oxidation of molecular hydrogen (
) based on the elution profile dataset. The training set of “gold standard” protein complexes was derived from experimental data in the STRING database, together with information in the literature (supplemental Table S3). In total, 10,302 and 8557 high-confidence protein pairs were obtained in vegetative cells and heterocysts, respectively. Among them, 1429 protein pairs containing 808 proteins were present in both types of cells (supplemental Fig. S6A and supplemental Table S4). These 808 proteins had a higher level of degree than the proteins that only existed in the protein interaction network of one type of cell (supplemental Fig. S6B). High-degree proteins can interact with many proteins and play an important role in the protein network (
Although most proteins and complexes are found in both cell types, some well-known protein complexes showed differences between vegetative cells and heterocysts (Fig. 3A). For example, the 50S and 30S ribosomes in heterocysts contain more heterocyst-specific PPIs and components such as uL2, uL23, uS2, uS5 (also known as RplB, RplW, RpsB, and RpsE), which are distinct peak-shifted proteins. This result is consistent with previous studies, that ribosome is susceptible to environmental fluctuations with variations in structure and components (
). We also observed that most photosynthetic complexes, including PSI, PSII, and phycobilisomes, might remain intact in both cell types, indicating that the photosystems have the ability to absorb light energy and perform electron transfer (
). However, some components, such as PsaL in PSI, PsbA1 in PSII, and PecC in phycobilisomes, presented distinct elution peak shifts and participated in different binary interactions between two types of cells (supplemental Fig. S7 and Fig. 3A). In particular, most phycobilisome components remained after cell differentiation, and the elution peak had no obvious shift (supplemental Fig. S8). However, the absorption spectrum of phycocyanobilin declined dramatically at 630 nm in heterocysts (Fig. 2F). The light absorption of phycobiliprotein depends on phycobilin (
), so we propose that phycocyanobilin may not covalently couple with phycobiliproteins, or its number may be reduced in heterocysts.
We also observed that other known complexes recruited new members in heterocysts, such as the diflavin flavoprotein and cytochrome c oxidase (Cox) (Fig. 3A). The transcriptions and quantitative proteomics results have shown that these complexes were dramatically upregulated, and some specific components were induced, leading to efficient electron transport and energy generation (
). The diflavin flavoprotein complex recruited more components to participate in the electron transfer process in heterocysts (Fig. 3B), meeting the enormous energy requirement. In addition, the elimination of oxygen is indispensable for the proper functioning of nitrogenase, and several mechanisms of elimination to prevent oxygen’s detrimental effects in heterocysts have arisen (
). The elution profile of Cox was similar in the two types of cells. However, CoxB shifted toward the higher molecular weight range (Fig. 3C), indicating the complex became larger in heterocysts.
Continuous Genomic Regions Tend to Be Induced and Expressed as Protein Complexes
The 10,302 and 8557 high-confidence protein pairs were clustered into 280 and 215 protein complexes of vegetative cells and heterocysts, respectively, which removed interactions among complexes and remained 6322 and 2791 PPIs (Fig. 4A and supplemental Table S5). Among these, we found well-known protein complexes, such as ribosomes and phycobilisomes. Except for conserved PPIs, nearly 88% of the identified PPIs were novel. To our knowledge, this is the first study to obtain a large protein complex dataset for Anabaena sp.
By analyzing the annotations of the protein complexes, we found that most of the components in the predicted complexes were localized in continuous genomic regions of chromosomes or plasmids. It has been reported that gene transcription is physically clustered to form “expressed islands,” and the structure of chromatin can change and be remodeled during heterocyst differentiation in Anabaena sp (
). We observed more continuous genomic regions in vegetative cells than in heterocysts (Fig. 4B). This difference became apparent when observing the protein identified in the two types of cells. For example, the continuous DNA region, alr3602–alr3071, encodes a probable glycosyl transferase, which was only identified and clustered into a complex in vegetative cells. Furthermore, the DNA region, alr5340–alr5355, which encodes a glycolipid synthase, was only expressed in heterocysts. That protein complex is essential for heterocyst maturation (
). We also found that some continuous DNA regions, such as the 50S ribosome, existed in both cell types, indicating conserved and indispensable functions (Fig. 4B).
We selected several hypothetical protein complexes from our dataset for AP–MS experimental validation to verify the protein complexes from continuous DNA regions. The hypothetical proteins of All1475, All7197, and Alr4359 were amplified from the Anabaena sp. genome and cloned into plasmid pRL25N for overexpression, which contained a Cu2+-induced petE promotor and a GFP ORF on the C-terminal region of the target gene (
). From AP–MS results, it was clear that their interacting partners were located upstream and downstream of the genes (Fig. 4C). It is worth noting that some gene segments, including ORFs with the same or reverse origin, could form protein complexes, such as Alr4359, which could interact with All4357 and All4358 (Fig. 4C), which are ATP-dependent Clp proteases that regulate protein degradation within the cell (
Alr4359 Is Located on Cellular Poles, and Its Overproduction can Influence the Diazotrophic Growth of Filaments
We observed that the hypothetical protein, Alr4359, interacts with proteins from the continuous DNA region all4118–all4120 in heterocysts (Fig. 4A). The elution profiles of Alr4359 and Alr4119 were more similar in heterocysts than in vegetative cells (Fig. 5A). The strong interaction between Alr4359 and Alr4119 in heterocysts was validated by AP–MS (supplemental Table S6). In addition, we found that Alr4359 interacted with FraH (Fig. 5A), which was validated by Y2H (supplemental Fig. S9A). Previously, an interaction between these two proteins was also found by Y2H in Synechocystis sp. PCC 6803 (
). The AP–MS experiment further confirmed that FraH–Alr4359–Alr4119 protein complex exists in heterocysts (supplemental Fig. S9B). Alr4119 belongs to the CURVATURE THYLAKOID 1 family and contains the homologous CAAD domain, which has the membrane-bending capacity and influences thylakoid organization (
). Interestingly, both FraH and Alr4119 have been shown to be dynamically located, changing from peripheral localization in the vegetative cells of nitrate-grown filaments to the heterocyst poles in diazotrophic filaments (
We used the Alr4359–GFP overexpression strain to trace the intracellular location of the Alr4359 protein and determine whether it had the same location pattern as FraH and Alr4119 (Fig. 5B). The fluorescence intensity of Alr4359 was obviously less than that of the control and FraH, indicating that the level of Alr4359 is strictly regulated. Alr4359 also interacts with ATP-dependent Clp proteases All4357 and All4358, which might regulate the rapid turnover of Alr4359. However, similarly to FraH, Alr4359 was also clearly located on the cell membrane and vegetative cell poles after nitrogen deprivation for 24 h. It is worth noting that no heterocysts were formed in the Alr4359 overexpression strain after nitrogen deprivation for 24 h or more (Fig. 5B). To observe the localization of Alr4359 in heterocysts, we regulated the expression of alr4359 by controlling the copper concentration in BG11 medium to reduce the activity of the petE promoter in the pRL25N plasma. When the filaments were cultured in BG11 without copper sulfate, the petE promoter's transcription was dramatically reduced, and GFP fluorescence was eliminated in the filaments (supplemental Fig. S10). The filaments were then transferred to BG11 medium without soluble copper for 24 h, and heterocyst differentiation was induced without any observation of GFP fluorescence. After returning the copper levels in the BG11 medium to normal levels for 12 h, we observed that Alr4359 was located on heterocyst poles (supplemental Fig. S10).
We observed the morphological characteristics of strains grown with or without combined nitrogen resources (Fig. 6A). Interestingly, thylakoid membranes were dramatically condensed and aggregated in both types of cells in FraH overexpression strain under nitrogen-deficient conditions. Unlike the supercompression of the thylakoid membrane in the FraH overexpressing strain, red spontaneous fluorescence was eliminated along the filaments in the Alr4359 overexpression strain (Fig. 6A). The diazotrophic growth of the Alr4359 overexpression strain was also seriously decreased, as shown by the significant change in the pigment composition (Fig. 6B). In addition, the filament length of the Alr4359 overexpression strain was longer than that of the control under nitrogen-deprivation conditions, even when no heterocysts were formed (Fig. 6C). Our results verify that the Alr4359 was mainly localized on the cell’s poles, and its overexpression can suppress heterocyst development and increase the length of filaments.
Filamentous nitrogen-fixing cyanobacteria served as a model organism to study cell differentiation and nitrogen fixation. It is necessary to explore PPI networks to understand better the molecular mechanisms underlying these processes. We constructed the PPIs of Anabaena sp. PCC 7120 by CoFrac–MS and machine learning, which is the largest protein interaction dataset of Anabaena sp. so far. We also generated a protein interaction map of Synechocystis sp. PCC 6803 recently (
). However, presenting protein interaction is not the only purpose of this work. Most important is to figure out the variation of PPIs between two cell types and then reveal the function variation of proteins.
Generally, oxygenic photosynthesis and CO2 fixation are performed in vegetative cells, whereas nitrogen fixation occurs in heterocysts (
). High-resolution MS was used to predict Anabaena sp. PPIs, and we observed most of the PSII proteins and their interaction with other proteins. In addition, phycobilisomes remained present, and the complex structure did not change dramatically, as most of the components did not shift on chromatography. However, the absorption peak of phycocyanobilin at 630 nm was diminished in heterocysts. Meanwhile, respiration was enhanced in heterocysts, maybe because of energy consumption during nitrogen fixation. In line with these observations, PSI components that serve as part of the electron-transport chain were upregulated in heterocysts (
). The results presented here indicate that intact and functional photosystems, including PSI, PSII, and phycobilisomes, exist in both cell types. However, the pigment components associated with them change during heterocyst differentiation.
Interestingly, the thylakoid membrane structure was reorganized during heterocyst differentiation. FraH and Alr4119 can influence honeycomb formation and protein redistribution in the thylakoid membrane during heterocyst differentiation (
). Here, we found that Alr4359 was a new factor that influences heterocyst differentiation and the diazotrophic growth of filaments. Under nitrogen-deficient conditions, we observed that the Alr4359 overexpression strain could not form heterocysts, and the pigment composition and filament length were also altered in this line. It may cause by twisted and loose thylakoid membrane structure, as shown by scanning electron microscopy (supplemental Fig. S11). A further experiment is needed to confirm the thylakoid membrane structure because of the technical limitations in this experiment. We also found that in heterocysts, FraH can interact with SepJ, which is located in the cell poles to control material communication between adjacent cells (
). Alr4359 was not found to interact with SepJ directly, but it interacts with Alr2947, an interaction partner of SepJ (supplemental Table S6). It is reasonable to speculate that the Alr4359–FraH–Alr4119 complex, located on heterocyst poles, influences material exchange by interacting with SepJ and its partner Alr2947.
CoFrac–MS is a powerful tool to monitor dynamic protein changes and uncover important functional factors when cells are in different life cycles or exposed to different environmental conditions (
). However, it also has some disadvantages, such as cannot distinguish protein aggregation. It is worth noting that both AP–MS and CoFrac–MS tended to identify tight and stable protein complexes in the cell since the weak physical interaction always depolymerized during the protein extraction process. Combining these methods with crosslinking, the weak and instantaneous interactions that have been ignored can be captured easily. Crosslinking combined with CoFrac–MS was also applied to generate protein correlation profiling of global membrane proteins in humans (
). A fully assembled structural information of a certain protein complex can be obtained by utilizing the cryo-EM, crosslinking, and MS in native cell extracts, such as pyruvate dehydrogenase complex (
). Combined with these assays, the active nanostructures of protein complexes involved in heterocyst differentiation can be further investigated in the future.
In addition to protein expression and interaction, post-translational modifications can act as another factor of protein regulation. Spectra of AP–MS revealed that Alr4359, FraH, and their interaction partner, Alr4119, all have phosphorylation modifications. The T362 and T410 phosphorylation on Alr4359 and T177 phosphorylation on FraH are shown in supplemental Fig. S12. Alr4119 has four phosphorylation modification sites on S32, T25, T27, and T29. The phosphokinase Alr0548 and the phosphorylase Alr0547 were also found in the prey proteins of FraH AP–MS result (supplemental Fig. S9B). FraH interacts with phosphokinase Alr0548 in vegetative cells, and it interacts with phosphorylase Alr0547 in heterocysts. Alr0548 was also found in the prey proteins of Alr4359 AP–MS result (supplemental Table S6). The results indicate that phosphorylation modification may affect the function of the Alr4359–FraH–Alr4119 complex during heterocyst differentiation.
There is still follow-up work to be done. The relationships across protein complexes, genomic regions, and upregulation or downregulation in different Anabaena states need to be explored. More experiments are needed to determine the specific function of each protein. However, our dataset provides valuable candidate proteins for further research. Knowledge of dynamic protein interactions can help to explain the functional differences between vegetative cells and heterocysts.
This work was supported by the National Natural Science Foundation of China (grant nos.: 31800647 and 91951210). We thank Dr Juyuan Zhang at the Institute of Hydrobiology Chinese Academy of Sciences for valuable discussions. We also thank Prof Xiang Gao at the Central China Normal University for providing algal, cloning vectors, and technical guidance.
C. W. conceptualization; C. X. formal analysis; C. X., B. W., H. H., and J. H. investigation; C. X. data curation; C. X. writing–original draft; C. W. writing–review & editing; C. W. supervision.