Transcriptome Analysis Reveals IsiA-Regulatory Mechanisms Underlying Iron Depletion and Oxidative-Stress Acclimation in Synechocystis sp. Strain PCC 6803

This study analyzed the impact of isiA deletion on the transcriptomic profile of Synechocystis. The isiA gene encodes an iron stress-induced chlorophyll-binding protein, which is significantly induced under iron starvation. The deletion of isiA affects the expression of various genes that are involved in photosynthesis and ABC transporters. WGCNA revealed three functional modules in which the blue module was correlated with oxidative stress. We further demonstrated that the isi operon contained the following five genes: isiA, isiB, isiC, ssl0461, and dfp by cotranscriptional PCR. Three sRNAs were identified that were related to oxidative stress. This study enhances our knowledge of IsiA-regulatory mechanisms under iron deficiency and oxidative stress.

ABSTRACT

Microorganisms in nature are commonly exposed to various stresses in parallel. The isiA gene encodes an iron stress-induced chlorophyll-binding protein which is significantly induced under iron starvation and oxidative stress. Acclimation of oxidative stress and iron deficiency was investigated using a regulatory mutant of the Synechocystis sp. strain PCC 6803. In this study, the ΔisiA mutant grew more slowly in oxidative-stress and iron depletion conditions compared to the wild-type (WT) counterpart under the same conditions. Thus, we performed transcriptome sequencing (RNA-seq) analysis of the WT strain and the ΔisiA mutant under double-stress conditions to obtain a comprehensive view of isiA-regulated genes. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses showed significant differences between the WT strain and ΔisiA mutant, mainly related to photosynthesis and the iron-sulfur cluster. The deletion of isiA affects the expression of various genes that are involved in cellular processes and structures, such as photosynthesis, phycobilisome, and the proton-transporting ATPase complex. Weighted gene coexpression network analysis (WGCNA) demonstrated three functional modules in which the turquoise module was negatively correlated with superoxide dismutase (SOD) activity. Coexpression network analysis identified several hub genes of each module. Cotranscriptional PCR and reads coverage using the Integrative Genomics Viewer demonstrated that isiA, isiB, isiC, ssl0461, and dfp belonged to the isi operon. Three sRNAs related to oxidative stress were identified. This study enriches our knowledge of IsiA-regulatory mechanisms under iron deficiency and oxidative stress.

IMPORTANCE This study analyzed the impact of isiA deletion on the transcriptomic profile of Synechocystis. The isiA gene encodes an iron stress-induced chlorophyll-binding protein, which is significantly induced under iron starvation. The deletion of isiA affects the expression of various genes that are involved in photosynthesis and ABC transporters. WGCNA revealed three functional modules in which the blue module was correlated with oxidative stress. We further demonstrated that the isi operon contained the following five genes: isiA, isiB, isiC, ssl0461, and dfp by cotranscriptional PCR. Three sRNAs were identified that were related to oxidative stress. This study enhances our knowledge of IsiA-regulatory mechanisms under iron deficiency and oxidative stress.

INTRODUCTION

Cyanobacteria are photoautotrophic prokaryotes that are widely distributed in various environmental conditions, including oceans, land, and freshwater lakes (1). Cyanobacteria also play an important role in the global carbon and nitrogen cycles (2). Synechocystis sp. strain PCC 6803 (hereafter termed Synechocystis) is a nondiazotrophic cyanobacterium and was first fully sequenced to serve as a model organism (3, 4). As with other aerobic photosynthetic organisms, Synechocystis contains two photochemical reaction centers, which are termed photosystem I (PSI) and photosystem II (PSII) (5). The photosynthetic apparatus is mainly located on the thylakoid membrane and is directly immersed in the cytoplasm (6, 7). The PSI is present in the form of a trimer with nine transmembrane protein subunits and three matrix protein subunits (8). The PSII monomer includes 20 protein subunits and approximately 80 cofactors (9). The imbalance of PSI and PSII may lead to increased intracellular reactive oxygen species (ROS) levels, which induces the upregulation of ROS protection mechanisms (10, 11).

Respiration of all aerobic organisms produces ROS in the cells (12). Otherwise, the cyanobacterial photosynthetic electron transport chain is also a source of ROS, especially in high light conditions (13, 14). Abiotic stress agents, including methyl viologen (MV), also cause oxidative stress by inducing cells to generate ROS. MV is thought to accept electrons from PSI and inhibit photosynthetic electron flow (15, 16). Excessive ROS inhibits the photosynthesis of cyanobacteria, which is mainly reflected in the “light suppression” of PSII. The high chemical reactivity of ROS results in the peroxidation of lipids, cleaves nucleic acids, and inactivates proteins in the cells (17). The response of cyanobacteria to oxidative stress includes energy expenditure, expression of nonenzymatic antioxidants and antioxidant enzymes, and the intrinsic ability of photosynthetic electron transport in oxidative-stress tolerance (10). To maintain the balance of intracellular ROS, the cells upregulate the expression of superoxide dismutase (SOD), catalase, and other peroxidase antioxidant enzymes to eliminate excess ROS (17, 18). Therefore, cyanobacteria can regulate the transcription of genes by generating ROS and can reconstruct metabolic networks in the cell to cope with adverse conditions (10, 19).

The assimilation of microbes under various stress conditions in the laboratory has been thoroughly investigated, but microbes in nature are typically subject to multiple stress conditions, including temperature, pressure, salt concentration, nutrition, and other abiotic stress. The PSI complex of Synechocystis contains a large number of iron atoms at an amount that is markedly higher than in Escherichia coli (20, 21). Upon iron limitation, levels of phycocyanin and chlorophyll in the cells are reduced, and the expression of iron-containing protein is decreased in the photosynthetic system, such as the cytochrome complex and all the Fe-S cluster proteins (22, 23). In addition, the cyanobacterial photosynthetic system is reconstructed and induces the expression of IsiA (iron stress-induced chlorophyll-binding protein) and IsiB (flavodoxin), which share the similar functions with iron-containing ferredoxin (23, 24). The peroxide-responsive repressor (perR) regulator in Synechocystis plays a pivotal role in coordinating iron homeostasis and oxidative stress (25). Importantly, oxidative stress is closely related to iron metabolism, and iron excess or deficiency can catalyze the formation of intracellular ROS; however, the antioxidant enzymes that scavenge ROS contain iron cofactors (26, 27).

In most cyanobacteria, IsiA proteins are induced in the thylakoid membrane upon iron limitation (28). Moreover, the isiA gene transcript can also be induced under oxidative stress, heat stress, or salt stress, while the IsiA protein can only be expressed in the absence of iron (29–32). A cyclic structure formed by 18 IsiA proteins surrounds a PSI trimer to form a protein supercomplex (33, 34). IsiA can also form an empty loop and has no direct correlation with the photosynthetic system under certain stress conditions (35). The IsiA has a protective function with induced expression to protect organisms from photooxidative damage under high light conditions (36). In the context of long-term iron deficiency, the dynamic change of the IsiA complex can actualize the distribution of light energy, which can balance the electron transfer between PSI and PSII and further reduce photooxidative damage (32). Although all genes of the isi operon, which includes isiA, isiB, and isiC, are cotranscribed, the expression level of isiA is significantly higher than those of the other genes (29, 31). The isi operon in Synechocystis is further regulated by ferric uptake regulation (fur) and isrR, a cis-encoded antisense RNA that is transcribed from the isiA noncoding strand (37, 38). These results demonstrate the complexity and diversity of the functions of the IsiA protein. However, the regulation of IsiA in acclimation of the photosynthetic apparatus to iron starvation and oxidative stress remains unclear.

This study investigated how Synechocystis adapts to combined stresses (iron deprivation and oxidative stress) at the transcriptome level. We compared the early stress responses of the wild-type (WT) strain and ΔisiA mutant to MV-induced oxidative stress under iron-deprived conditions. The transcriptome sequencing (RNA-seq) analysis results indicate that IsiA regulates specific cellular functions such as photosynthesis and transporters. Weighted gene coexpression network analysis (WGCNA) further demonstrated the regulation mode of the strain and the function of IsiA under oxidative-stress conditions. Cotranscription verification demonstrated that isiA (sll0247), isiB (sll0248), isiC (sll0249), ssl0461, and dfp (sll0250) belong to the isi operon. Finally, three sRNAs were identified using the transcriptome data, and their expression was detected using quantitative PCR (qPCR) analysis. Meanwhile, their target genes were predicted, and their possible roles in oxidative stress were contemplated.

RESULTS

Characterization of the phenotype of the isiA-deleted mutant.

To reveal the function of IsiA in Synechocystis, we deleted the isiA gene by inserting a spectinomycin cassette and replacing the coding region (Fig. S1A in the supplemental material), to which a fully segregated mutant was obtained (Fig. S1B). In this experiment, after being subjected to iron-depleted conditions for 24 h, quantitative real-time PCR (qRT-PCR) analysis demonstrated that isiA transcriptional level was significantly increased for the WT strain, but isiA was not induced for the mutant under the same conditions (Fig. S1C). To further investigate the function of IsiA under oxidative-stress conditions, the growth curves and plate experiments were first done for the WT strain and ΔisiA mutant in BG11 medium supplemented with 2 μM MV (Fig. 1A and D). There were no significant growth differences between the WT strain and ΔisiA mutant before and after the addition of MV. Next, the growth of the WT strain and ΔisiA mutant under MV and iron deficiency conditions was further examined (Fig. 1B and D). After adding MV, the growth of the ΔisiA mutant was weaker than that of the WT strain. Subsequently, the chlorophyll a content was further tested, and its content in the ΔisiA mutant was found to be significantly lower than that of the WT strain at 96 h (Fig. 1C). SOD activity was measured to verify that oxidative stress was induced under experimental conditions. Since the cells could not grow normally with a high concentration of MV, the SOD activity was detected within a short period (Fig. 1E). Results show that the SOD activity of the WT strain was comparable to that of the ΔisiA mutant when cultured under Fe-replete medium, and both significantly increased in iron deficiency conditions for 24 h. However, a significant decrease in SOD activity of the WT strain was found under iron deficiency for 24 h followed by MV treatment for 4 h, while the SOD activity of the ΔisiA mutant exhibited only a slight decrease. These data indicated that under the double-stress conditions, a high concentration of MV may inhibit SOD activity and aggravate the oxidative-stress reaction in the bacteria. To further investigate the regulation of oxidative-stress differences between the WT strain and ΔisiA mutant, we conducted transcriptome sequencing.

FIG 1.

Phenotype of the WT strain and ΔisiA mutant. (A) Growth curves of the WT strain and ΔisiA mutant under oxidative-stress conditions. (B) Growth curves of the WT strain and ΔisiA mutant under iron-depleted and oxidative-stress conditions. (C) Chlorophyll a content of the WT strain and ΔisiA mutant at 96 h under iron-depleted and oxidative-stress conditions. Asterisks indicate significant differences from the WT strain (*, P < 0.05, determined by Student's t test). (D) Drop-plate experiments of WT strain and ΔisiA mutant are shown. The up-and-down lanes indicate the WT strain and ΔisiA mutant, respectively. (E) SOD activity of the WT strain and ΔisiA mutant under Fe-replete medium (yellow), iron deficiency for 24 h (magenta), and iron deficiency for 24 h followed by MV treatment for 4 h (cyan). Asterisks indicate significant differences from strains that were cultured in the BG11 medium (*, P < 0.05; **, P < 0.01, determined by Student's t test).

Overview of transcriptomic analysis of the WT strain and ΔisiA mutant for iron-depleted and MV response.

A total of 383 million raw sequencing reads were obtained from the transcriptomics analysis of 12 samples, as mentioned in Materials and Methods, with average reads of 32 million reads per sample (Table 1). After removal of the low-quality reads, 382 million clean data were obtained. Of these clean reads, 376 million were mapped on the genome of Synechocystis. The mapping rates of each sample were more than 97%. The sequence reads matched to all 3,594 coding genes in the Synechocystis genome, which demonstrates that the sequencing depth is sufficient to cover all transcripts in the cell.

TABLE 1.

Basic statistics of the transcriptome data

Sample	Total reads	Mapped reads	Mapped ratio (%)
WT_0h_1	38.54	38.01	98.63
WT_0h_2	31.04	30.84	99.34
WT_0h_3	29.28	28.89	98.67
WT_4h_1	26.72	26.33	98.53
WT_4h_2	33.27	32.99	99.15
WT_4h_3	29.16	28.99	99.43
ΔisiA_0h_1	34.70	33.84	97.5
ΔisiA_0h_2	35.28	34.54	97.9
ΔisiA_0h_3	36.82	36.02	97.82
ΔisiA_4h_1	29.52	28.9	97.89
ΔisiA_4h_2	30.97	30.35	98
ΔisiA_4h_3	27.26	26.73	98.03
Total sum	382.56	376.41	98.11

Comparative analysis of transcriptional profiles.

To understand the molecular mechanism of physiological differences between the WT strain and ΔisiA mutant under stress conditions, the stress-responsive genes were examined by enrichment analysis using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) terms (Fig. 2). A Venn diagram shows the number of differentially expressed genes (DEGs) in response to iron limitation and MV stress. A total of 1,113 DEGs (572 upregulated and 541 downregulated) and 285 DEGs (168 upregulated and 117 downregulated) were detected in the WT strain and ΔisiA mutant, respectively, of which 227 DEGs were detected in both strains (Fig. 2A). These genes were mainly enriched in rRNA binding, constituents of ribosome and translation, and the metabolic pathways that are involved in the ribosome (Fig. 2B and C). Nevertheless, 886 and 58 DEGs were unique to the WT strain and ΔisiA mutant, respectively (Fig. 2A). These genes were mainly enriched for photosynthesis and phycobilisome as well as metabolic pathways that are involved in photosynthesis for the WT strain (Fig. S2A and B). Meanwhile, regarding the ΔisiA mutant, the 58 DEGs are related to heme binding and iron-sulfur cluster binding and metabolic pathways that are involved in carbon fixation and the citrate cycle (Fig. S2C and D). However, only 90 DEGs (73 upregulated and 17 downregulated) were detected prior to adding MV, and these genes were mainly related to the integral component of membranes and iron-sulfur cluster binding (Fig. S2E and F). This suggests that the depleted iron was not sufficient to elicit more significant transcriptomic changes in both strains, as a high concentration of MV more strongly affects gene expression and metabolic processes under iron-limited conditions.

FIG 2.

Venn diagrams of the detected changes in the transcriptome. Graphs indicate the number of significantly expressed genes. (A, Left) Indicates the overlapping DEGs among the comparison of “WT-4h” versus “WT-0h” and “ΔisiA-4h” versus “ΔisiA-0h.” (A, Right) Represents the overlapping DEGs among the comparison of “ΔisiA-0h” versus “WT-0h” and “ΔisiA-4h” versus “WT-4h.” The GO terms analysis (B) and KEGG analysis (C) refer to the overlapping part of the left Venn diagram of panel A, which represents genes that are differentially expressed in the WT strain and ΔisiA mutant after adding MV.

Quantitative RT-PCR validation of differentially expressed transcripts from RNA-seq.

To confirm the accuracy of the RNA-seq results, the expression levels of 20 genes, including up- and downregulated genes, were randomly selected for qPCR verification. As shown in Fig. 3, the qPCR results of the 20 selected genes are in accord with the RNA-seq data, which indicates the reliability and reproducibility of the sequencing results.

FIG 3.

Real-time qPCR analysis of 20 randomly selected genes to validate the RNA-seq data. The x axis indicates that the RNA-seq data were calculated by the FPKM method. The y axis represents the qPCR results by using the 2^−ΔΔCt (cycle threshold) method. The qPCR experiments were performed independently in triplicate. The fold change values from RNA-seq and qPCR obeyed a linear correlation (y = 0.9578x + 0.5298; R² = 0.9604).

The genes related to photosynthesis and oxidative stress.

The PSI complex in Synechocystis contains 11 subunits, while the PSII complex is composed of 20 subunits. Further, the genes that encode these subunits are dispersed throughout the genome (4, 9). We found that the expression levels of most photosynthetic system genes were lower for the WT strain compared to the ΔisiA mutant after adding MV (Table S1 and S2). The expression levels of these genes also demonstrated evident downregulation in the ΔisiA mutant, but the degree of reduction was significantly lower than that of the WT strain. No significant differences in gene expression levels between the WT strain and ΔisiA mutant were found when MV was not added (Table S1 and S2). After adding MV, the expression levels of genes that encode subunits of phycobilisome (cpc, apc) were also lower in the WT strain, along with higher expression levels of petC, which encodes subunits of the cytochrome b₆f complex iron-sulfur protein. Transcription of the gene that encodes the small heat shock protein hspA was significantly induced by 151.43-fold for the WT strain and 13.26-fold for the ΔisiA mutant. This gene undergoes induction and is expressed under various stress conditions, including osmotic stress, oxidative stress, and temperature stress (39). There are other heat shock protein genes, such as dnaJ and htpG, which are upregulated in both strains but higher in the WT strain. These results indicate that the WT strain is more sensitive to iron-limited and oxidative stress than the ΔisiA mutant.

The genes related to transporters.

ABC transporters use the energy that is released by ATP binding and hydrolysis to transport a variety of substrates across the cell membrane. The presence of ROS in the cells affects the homeostasis of iron in cells, as ROS can oxidize ferrous iron and release harmful peroxide radicals via the Fenton reaction (40). Thus, some proteins that are related to iron transport are induced to change intracellular ferrous levels. The genes futA1 and futA2 (slr1295and slr0513), which encode the iron transport protein, were also found to be induced by oxidative stress (41). However, the expression levels of both genes in the WT strain and mutant are downregulated under double-stress conditions. Meanwhile, the expression levels of genes that encode iron transport (fecB, fecC, fecE, and fhuA) were also lower under oxidative-stress conditions in both strains. The expression levels of genes that encode the high-affinity nitrate/nitrite transporters (nrtBACD) are significantly inhibited in the WT strain despite minor upregulation in the ΔisiA mutant. The gene kpsM, which encodes the polysialic acid transport protein, also exhibited a 4.56-fold change in the WT strain. The gene amiC, which encodes a negative aliphatic amidase regulator, displayed a 3.23-fold change in the ΔisiA mutant. These results illustrate that iron deprivation may reduce the responses to oxidative stress.

Build and detection of functional modules.

To further investigate the genes that are related to oxidative stress, 1,305 DEGs in both strains were selected to construct a scale-free coexpression network. WGCNA divided the 1,305 DEGs into three different modules—referred to as the blue, gray, and turquoise—that contain 557, 110, and 638 genes, respectively (Fig. 4A and B). The genes in the gray module showed a higher negative correlation with the WT strain and a positive correlation with the ΔisiA mutant, which suggests that these genes may be associated with isiA (Fig. 4C). Further, the genes of the blue module were shown to be related to the regulation of oxidative stress. Lastly, the turquoise module negatively correlates with SOD (r = −0.81; P < 0.05).

FIG 4.

Detection of module-trait relationships in the WT strain and ΔisiA mutant under oxidative-stress conditions based on WGCNA was performed. (A) The dendrogram displays the different genes that were aggregated into coexpressed modules. (B) The numbers of DEGs for the different modules are shown. (C) The relationship between coexpressed modules and oxidative stress in both strains. The numbers in the squares represent the correlation coefficients. A positive correlation is colored in red, and a negative correlation is colored in green.

GO enrichment analysis was performed on the genes in the three constructed modules that were detected by WGCNA (Fig. S3). The genes in the blue module were mainly involved in rRNA binding, photosynthesis, and phycobilisome. The gray module-related genes were enriched in ATPase activity and regulation of transcription, DNA templated. Lastly, in the turquoise module, DNA binding and transposase activity were detected as the most enriched GO terms. These results indicated that the coexpressed gene modules have different functions in response to MV stress.

Networks that display relationships among genes within coexpressed modules.

To identify the key hub genes within each module, we visualized the gene network using Cytoscape software. The node circle size is positively correlated with the number of genes that are partnered within interactions. The hub genes refer to the genes with the biggest node size. In the blue module, we observed three main clusters of genes that were connected by the genes sll0822 (hypothetical protein) and slr0121 (hypothetical protein) (Fig. 5). Moreover, we identified several hub genes, including glpX (d-fructose-1,6-bisphosphatase), nifS (cysteine desulfurase), apcc (phycobilisome LC linker polypeptide), tktA (transketolase), apqZ (water channel protein), etc. The key hub genes in the turquoise module that were detected are kpsM (polysialic acid transport protein), smpB (SsrA-binding protein), jag (hypothetical protein), sll0005 (ABC1 like), synA (cation-transporting ATPase), etc. The gray module contained several small clusters of genes and several hub genes, which include fus (elongation factor G), ppsA (phosphoenolpyruvate synthase), and ilvD (dihydroxy-acid dehydratase).

FIG 5.

Coexpression network analysis of blue (A), gray (B), and turquoise (C) modules. The size of the node circle is positively correlated with the number of interacting gene partners. The genes marked in red indicate the connection point in the blue module. The genes marked in yellow present the hub genes of each module.

Prediction of the isi operon and verification.

In the previous study, the isi operon includes isiA, isiB, and isiC, which could be induced under various stress conditions (31). Two open reading frames, ssl0461 and dfp (sll0250), are located immediately downstream of isiC and encode a hypothetical protein and pantothenate metabolism flavoprotein, respectively. This study found that the genes ssl0461 and dfp may belong to the isi operon. First, the correlations of adjacent genes were calculated based on the fragments per kilobase of exon per million mapped fragments (FPKM) values of the five genes of the transcriptome data (Table 2). The correlations between neighbor genes among these five genes by Pearson correlation coefficients (PCCs) analysis showed that all exceeded 0.9. Since isiA can achieve the highest level of transcription at 24 h in iron deficiency conditions (Fig. S1C), based on RNA-seq data, the “WT-0h” sample represents iron deficiency for 24 h, and thus, this sample was analyzed by Integrative Genomics Viewer (IGV) software (Fig. S5). From the results of the reads alignment, no breakpoint among the five genes was found, and the initial breakpoint is approximately 1517268, which means that the transcription start site of the isi operon, and the start code of the isiA gene, is 1517073. This finding is consistent with those of previous reports (29).

TABLE 2.

The correlation analysis of adjacent genes

Characteristic	Gene pair
	dfp-ssl0461	ssl0461-isiC	isiC-isiB	isiB-isiA	isiA-sll0245
PCC	0.925	0.992	0.996	0.920	0.665
P value	1.095E−05	0.0403	0.0135	0.029	0.027

Next, a cotranscriptional PCR experiment is further required to validate the isi operon. The RNA sample of the WT strain at 24 h of iron deficiency conditions was selected. The genomic DNA, RNA, and cDNA were found to all amplify bands of corresponding sizes, though digest RNA could not expand the bands (Fig. 6A). At the same time, two genes, sll0245 and galE (sll0244), that were located upstream of isiA were selected as negative controls (Fig. S6). Only genomic DNA and RNA amplified bands, while digested RNA and cDNA could not expand the bands. This further proved that these five genes belong to a unique operon. To further investigate the function of the isi operon, the interaction networks among the genes of the isi operon (isiA, isiB, isiC, ssl0461, and dfp) and their coexpressed genes are shown in Fig. 6B. The gene slr1958 encodes a hypothetical protein that is coexpressed with the isiA gene. The genes sll1825 and slr1456 are coexpressed with isiB, isiC, and ssl0461 and encode aklaviketone reductase and general secretion pathway protein G, respectively.

FIG 6.

Cotranscription verification analysis of the isi operon. (A) The cotranscription PCR of the isiA, isiB, isiC, ssl0461, and dfp genes in the WT strain. The primer pairs that are represented by 1 to 10 are as follows: q-sll0250-F versus q-ssl0461-R (354 bp), q-sll0250-F versus Co-sll0249-R (1,043 bp), q-sll0250-F versus Co-sll0248-R (1,523 bp), q-sll0250-F versus Co-sll0247-R (2,434 bp), q-ssl0461-F versus Co-sll0248-R (1,347 bp), q-ssl0461-F versus Co-sll0247-R (2,258 bp), q-ssl0461-F versus Co-sll0249-R (867 bp), Co-sll0249-F versus Co-sll0248-R (326 bp), Co-sll0249-F versus Co-sll0247-R (1,237 bp), and Co-sll0248-F versus Co-sll0247-R (487 bp). (B) Coexpression network of the isi operon.

Prediction of sRNA targets and coexpression networks.

Three new sRNAs that may be associated with oxidative stress were identified using the RNA-seq data. The qPCR verification analysis indicated that the expression levels of these three sRNAs were consistent with the RNA-seq data for the WT strain (Fig. 7A). However, the expression level of Novo_0002 did not appear to be upregulated in the ΔisiA mutant. Moreover, while the expression level of Novo_0003 was upregulated, the expression was downregulated in RNA-seq data. This may be because the FPKM value is not fully representative of its expression. Next, the expression levels of sRNAs in the WT strain under iron deficiency or oxidative-stress conditions were further assessed (Fig. 7B). The expression of all three sRNAs was inhibited under oxidative-stress conditions, and only Novo_0002 was induced under iron deficiency. Thus, all the sRNAs are related to iron deficiency and oxidative stress.

FIG 7.

qPCR analysis of the sRNA expression levels. (A) The sRNA expression levels according to the RNA-seq data. (B) The expression levels of sRNAs in the WT strain. (C) The sRNA-mRNA regulatory network. The yellow-colored nodes represent the sRNAs; the blue-colored nodes represent the target genes. A positive correlation is represented by a solid line, and a negative correlation is represented by a dotted line.

The target genes of sRNA were predicted by miRanda and PCCs (PCCs > 0.9). The regulating network of sRNA and their target mRNAs were visualized using Cytoscape (Fig. 7C). The sRNAs can regulate the expression of target genes posttranscriptionally by complementing with mRNA sequences (42). sRNA and mRNA exhibited a negative correlation, as sRNAs negatively regulate the expression of their target mRNAs via cleavage of target mRNA. Further, Novo_0002 showed negative correlations with cpcD, psaB, and psaA, all of which are related to photosynthetic systems. In total, 42 negative correlations exist between Novo_0004 and target mRNAs. The three encoded gene products with the highest negative correlations are adenylate cyclase, hypothetical protein, and quinolinate synthetase, in that order.

DISCUSSION

This study generated the gene-deleted mutant of isiA in Synechocystis to obtain a comprehensive view of the function of IsiA under oxidative-stress conditions using RNA-seq analysis. We successfully obtained the fully segregated isiA-deleted mutant from the WT strain (Fig. S1B in the supplemental material). Previous studies show that isiA transcript can be induced under various stress conditions, especially iron deficiency (29, 30). The qPCR analysis revealed that the isiA expression level reduced to below the detection limit in our mutant after iron deficiency was applied (Fig. S1C). Relative transcript levels were approximately 250 to 300 times higher for the isiA after 48 h of iron starvation in the cyanobacterium Leptolyngbya sp. strain JSC-1 (43). However, the transcript levels of isiA were approximately 700 times higher after 24 h of iron depletion in Synechocystis (Fig. S1C). Further, a significant decrease in SOD activity in the WT strain was found (Fig. 1E). However, in the transcriptome data, the transcription levels of sodB in both strains were upregulated and increased 9-fold in the WT strain and 3.6-fold in the ΔisiA mutant. This indicates that cells may induce antioxidant expression given oxidative stress, but high concentrations of MV undergo inhibition of enzyme activity.

The photoreaction center is mainly composed of PSI and PSII, the cytochrome b₆f complex, and ATP synthase (44). PSII is most vulnerable to oxidative damage. An orange carotenoid protein-related energy dissipation mechanism and induction of another iron stress protein (encoded by idiA) in the absence of iron can protect PSII from oxidative damage (45). The orange carotenoid protein can act as both the photoreceptor and mediate photoprotective energy dissipation via interactions with the phycobilisomes’ core (45). The dynamic change of the IsiA complex in the absence of iron can balance the electron transfer between PSI and PSII to reduce oxidative damage (32). However, from the RNA-seq data, the expression levels of psa and psb genes in the ΔisiA mutant are higher than in the WT strain (Table S1). This suggests that overexpression or deletion of the IsiA protein in cells may cause different regulatory patterns for photosynthetic apparatus. In microorganisms, the metalloprotein perR regulates the expression of many genes in response to ROS changes in the external environment (46). In the present study, the expression levels of perR in both strains are significantly upregulated under MV-induced oxidative stress (Table S1). Moreover, the coexpression network analysis identified 302 DEGs that are coexpressed with perR. These genes were mainly related to photosynthesis pathways and were significantly enriched with the following GO terms: photosynthesis and phycobilisome (Fig. S7). It is presumed that the perR regulators help to withstand oxidative stress by regulating photosynthetic gene expression.

Both hypothetical and unknown proteins are widely distributed in the cyanobacterial genome, representing an incomplete understanding of cyanobacterial metabolism and molecular structure (47, 48). In this study, we found 144 (“WT-4h” versus “WT-0h”) and 25 (“ΔisiA-4h” versus “ΔisiA-0h”) proteins with unknown functions with more than 4-fold transcriptional changes (|log₂FoldChange [FC]| > 2; P < 0.05). The hypothetical protein wecG (slr5055) was the most upregulated gene (178.01-fold) in the WT strain, and a significant increase (14.07-fold) was observed in the ΔisiA mutant (Table S1). Among these hypothetical proteins, the fold changes in the WT strain were more pronounced than in the ΔisiA mutant. These unmapped or indeterminate transcripts are key specific responses to MV stress in Synechocystis, and the function and putative roles of these proteins should be further investigated.

WGCNA was performed to decompose the DEGs into three functional modules (Fig. 4). The functional enrichment analysis results indicate that the differences in the interactions between different modules are due to their different functions. Thus, the blue module was found to be mainly enriched in pathways that were associated with photosynthesis. Further, the gray module genes were mainly enriched for a two-component system and the nitrogen metabolism process (Fig. S4). A two-component system is a signal transduction system and is a mechanism for bacteria to adapt to selection pressure (49). Since cyanobacteria can reduce the relative number of PSI complexes upon iron starvation (50), the enrichment pathway that involves the blue module may regulate the reconstruction of photosynthetic apparatus (Fig. S4). Thus, we speculated that the blue and gray modules are the most important modules of Synechocystis in response to iron deficiency and oxidative stress.

We have initially verified that the isi operon contains five genes (isiA, isiB, isiC, ssl0461, and dfp) (Fig. 6), and the function of IsiA has been investigated in recent years. The isiB gene is not essential for the photoautotrophic iron-limited growth of the cyanobacterium (51). To further explore the functions of isiC and ssl0461, which encode hypothetical proteins, and to determine their role in the operon, gene deletion and overexpression strains should be constructed separately, followed by observing the phenotype of the mutant strains under various stress conditions.

Combined with RNA-seq data and sRNA prediction analysis, we identified three sRNAs and predicted their target genes. The gene gidA, which encodes glucose, inhibited the division protein A, which is the common target of these three sRNAs. However, gidA is only negatively correlated with Novo_0004. The htpG that encodes a heat shock protein is positively correlated with Novo_0003 and negatively correlated with Novo_0004. Thus, the regulation mechanisms of sRNAs and their targets under oxidative-stress and iron depletion conditions require further investigation.

In conclusion, global transcription profiling of the WT strain and ΔisiA mutant showed differences when the cultures were grown in iron deficiency and oxidative-stress conditions. Differences in GO and KEGG enrichment in the WT strain and ΔisiA mutant indicate that the deletion of isiA alters the regulatory pattern of strains in response to stress. Using a WGCNA method, the present study demonstrates that the functional modules of isiA relate to nitrogen metabolism and the two-component system. Further, we proved that the isi operon contains five genes, and three new sRNAs were identified that were associated with iron deficiency and oxidative stress. These findings provide a critical theoretical framework for further research on the regulation of IsiA under oxidative stress conditions.

MATERIALS AND METHODS

Strains and culture conditions.

In this study, the wild-type Synechocystis and its derivatives were grown in BG11 medium, as described previously (52). Synechocystis cells were cultivated at 30°C with rotary agitation at 150 rpm under illumination (∼40 microeinsteins m⁻² s⁻¹). Spectinomycin (10 μg ml⁻¹) was added to the culture medium when necessary. Oxidative stress was induced by adding MV (methyl viologen). When preparing the iron-deficient medium, it was not necessary to add ferric ammonium citrate. Further, the culture vessel was soaked with 1 mmol/liter HCl overnight and then washed with deionized water and sterilized. The cell density was measured to be optical density at 730 nm (OD_730nm). All strains and plasmids that were used in this study are described in Table 3, and all the primers that were used in this study are described in Table 4. Chlorophyll a content was determined spectrophotometrically at OD_663nm and OD_645nm in N,N-dimethylformamide. Chlorophyll a content (μg ml⁻¹) was calculated using the following equation: chlorophyll a = 12.7 × Abs663 − 2.35 × Abs645.

TABLE 3.

Strains and plasmids used in this study

Strain/plasmid	Description	Source
DH5α	F−, conventional cloning host strain	State Key Laboratory of Agricultural Microbiology
PRL271	Cmr, cyanobacterial integrated shuttle vector	State Key Laboratory of Agricultural Microbiology
PHP45Ω	Spr, containing Ω fragments	State Key Laboratory of Agricultural Microbiology
PRL271-sll0247-Ω	Spr Cmr, integrated plasmid for isiA deletion	This study
Synechocystis sp. PCC 6803	Unicellular cyanobacteria, photoautotrophic, or heterotrophic	State Key Laboratory of Agricultural Microbiology
ΔisiA mutant	Spr, transformation of plasmid PRL271-sll0247-Ω to Synechocystis sp. PCC 6803	This study

TABLE 4.

List of oligonucleotides used in this study

Name of oligonucleotide	Sequence (in 5′–3′ direction)	Application
sll0247-up-F	ATTCGATATCTAGATCTTTTGATTCGTCAAAATCA	Construction of PRL271-sll0247-Ω plasmid
sll0247-up-R	CAATCACCGGATCCCCGTCATTAGCCATCAGCT	Construction of PRL271-sll0247-Ω plasmid
sll0247-dn-F	CAATCACCGGATCCCCCAGAATTGCCTCCTTAATT	Construction of PRL271-sll0247-Ω plasmid
sll0247-dn-R	ACGTTGTTGCCATTGCAATGGCCTTAACCCCG	Construction of PRL271-sll0247-Ω plasmid
sll0247-P5	TCTGTTAACTCTGGCTGATTA	Verification of ΔisiA mutant
sll0247-P6	CTCTCTACCACCTTACGC	Verification of ΔisiA mutant
Pomega18	GCTTGCTCAATCAATCAC	Verification of ΔisiA mutant
q-psb28-F	CGGTTGGGAATTGAGGAT	Verification of RNA-seq data
q-psb28-R	AGGGAAGTAAAGGGCAAAT	Verification of RNA-seq data
q-psbA1-F	CAGTTTGTTGTTTCGCTTTC	Verification of RNA-seq data
q-psbA1-R	GTTGGCACGGTTAATCAC	Verification of RNA-seq data
q-sodB-F	CGCACTACCTAACTTACCTTA	Verification of RNA-seq data
q-sodB-R	TTGTCGTGATGGAACTCC	Verification of RNA-seq data
q-psbX-F	AACGGGTGATTTTGTCTTTT	Verification of RNA-seq data
q-psbX-R	CCCTTCTTTAGCAAACTTTCT	Verification of RNA-seq data
q-atpI-F	GGAATAGACCCAGACAACA	Verification of RNA-seq data
q-atpI-R	CGGTAGTGGATAACGGAAT	Verification of RNA-seq data
q-nrtA-F	GAGCGGTCAATAATATCTTCAA	Verification of RNA-seq data
q-nrtA-R	CCAGGCAGAGAATAAGGAA	Verification of RNA-seq data
q-petC-F	ATCGGCATTGTTGATAGTTG	Verification of RNA-seq data
q-petC-R	TGGCAGGGACATTGAAAT	Verification of RNA-seq data
q-sbt-F	ATTCGGAACTCCAACTTGA	Verification of RNA-seq data
q-sbt-R	CATTGTAGAGCCACTGACT	Verification of RNA-seq data
q-bcp-F	ACCACTGAAGCGATAGATT	Verification of RNA-seq data
q-bcp-R	GACTAACACCGACAACAAC	Verification of RNA-seq data
q-cpcc-F	ACAAATTCCCGCACACTA	Verification of RNA-seq data
q-cpcc-R	TTAGGCAACGATTACATTATGG	Verification of RNA-seq data
q-hypA-F	CACGAAGTTAGTCTGATGGA	Verification of RNA-seq data
q-hypA-R	GGGTTAAACGATGGATTTGG	Verification of RNA-seq data
q-sthk-F	CATCGGAGGAGATACTAACC	Verification of RNA-seq data
q-sthk-R	ATGGTGAACTGCCTATCG	Verification of RNA-seq data
q-slr1919-F	GCCTCCATACCGATGAAG	Verification of RNA-seq data
q-slr1919-R	CGACATTGACCAGTTCCT	Verification of RNA-seq data
q-ycf16-F	AATCTCACTGCCTCTGTTG	Verification of RNA-seq data
q-ycf16-R	ATAATGGCGTGGACTTCC	Verification of RNA-seq data
q-hsp17-F	TTCGGTGCTATGGGTATC	Verification of RNA-seq data
q-hsp17-R	AGAACTAACTGAAACTGAAGAAG	Verification of RNA-seq data
q-slr1311-F	GACAACGACTCTCCAACA	Verification of RNA-seq data
q-slr1311-R	GCGATGAAGGCAATGATG	Verification of RNA-seq data
q-sll1621-F	CCGAACTCAGGCTCAATA	Verification of RNA-seq data
q-sll1621-R	ATGGGTATGTTGGTGGAAA	Verification of RNA-seq data
q-sll0947-F	TTCCAGAGCGTCTTCAAT	Verification of RNA-seq data
q-sll0947-R	GCATCCAAGATAAGCAACAT	Verification of RNA-seq data
q-psaC-F	ACACAATCCTCGGTTCTAG	Verification of RNA-seq data
q-psaC-R	ACGATACCTGTATTGGTTGTA	Verification of RNA-seq data
q-psaD-F	TGTTTATCCCAGTGGTGAA	Verification of RNA-seq data
q-psaD-R	ACTTGATGGTTACAGGTTCT	Verification of RNA-seq data
q-Rnpb-F	CACCAATTTCCCAAGACTAC	Verification of RNA-seq data
q-Rnpb-R	TGCCATTGATTAGAGCCATA	Verification of RNA-seq data
q-Novo-0002-F	AATTAGAGAAGGAATAGGAGGTAAGC	Verification of the transcription level of sRNA
q-Novo-0002-R	AGTGCAGGGTCCGAGGTATT	Verification of the transcription level of sRNA
q-Novo-0003-F	CCTCACTAAAGCTCCGGCA	Verification of the transcription level of sRNA
q-Novo-0003-R	AGTGCAGGGTCCGAGGTATT	Verification of the transcription level of sRNA
q-Novo-0004-F	TCAGACAATTACACAAAATTACAACAG	Verification of the transcription level of sRNA
q-Novo-0004-R	AGTGCAGGGTCCGAGGTATT	Verification of the transcription level of sRNA
Rever-Novo0002	GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACGGGAAT	Stem loop primer
Rever-Novo0003	GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCAAACT	Stem loop primer
Rever-Novo0004	GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACGCAACA	Stem loop primer
q-sll0247-F	TAATGTTGGCTGACTGACTA	Verification of the transcription level of sll0247
q-sll0247-R	TCATCATCTCCTGTTCCTG	Verification of the transcription level of sll0247
q-sll0248-F	CTCTGGCTGATTATCTTCGT	Verification of the transcription level of sll0248
q-sll0248-R	CAGTGGATTAGGCGGTAA	Verification of the transcription level of sll0248
q-sll0249-F	GATATACACCAACGGAGAGT	Verification of the transcription level of sll0249
q-sll0249-R	TTCACCAACCTGTCCATT	Verification of the transcription level of sll0249 and cotranscription verification
q-ssl0461-F	CGTCATAGGTCTCATTATCCA	Verification of the transcription level of ssl0461 and cotranscription verification
q-ssl0461-R	GATGTAGCTGCACTCACT	Verification of the transcription level of ssl0461 and cotranscription verification
q-sll0250-F	GGTAAAGGTCAAGGGAGTAA	Verification of the transcription level of sll0250 and cotranscription verification
q-sll0250-R	AATCTGTGAGGTGGTGTC	Verification of the transcription level of sll0250 and cotranscription verification
Co-0247-F	CCAAAATGCTGTTAAAGC	Cotranscription verification
Co-0247-R	TCAAACGGGTCGAACAA	Cotranscription verification
Co-0248-F	GTTGCCGGTTTGAGTAC	Cotranscription verification
Co-0248-R	GAGTAAAGACATGGGTAAGTG	Cotranscription verification
Co-0249-F	AAGCCAAATGATAGGTGTTT	Cotranscription verification
Co-0244-R	ACATACCATCATTGACCA	Cotranscription verification
Co-0245-R	CAAGGAAAAGGGATTAGTC	Cotranscription verification

Generation of the isiA-deleted mutation.

The coding region of isiA (sll0247) (1,029 bp, from nucleotides 1516045 to 1517073 according to numbering in NCBI) was completely deleted by inserting a spectinomycin resistance (Ω) cassette. Upstream and downstream fragments of approximately 801 bp, not including the isiA coding sequence, were amplified using PCR from the WT genomic DNA. The following primer sets were used: sll0247-up-F and sll0247-up-R for amplification of 801 bp of upstream fragments and sll0247-dn-F and sll0247-dn-R for amplification of 801 bp of downstream fragments. The plasmid pRL271 is digested with XhoI-PstI. The plasmid pHP45Ω is digested with SmaII, PstI, and NdeI to obtain the Ω cassette. Then, the Exnase cloning kit (Vazyme) was applied for introduction into pRL271. The WT strain was transformed with the cloning product, and transformants (ΔisiA mutant) were selected in the presence of spectinomycin.

SOD activity assay.

Superoxide dismutase (SOD) activity was quantified with the Total SOD assay kit with WST-8 (catalog no. S0101; Beyotime Biotechnology, Shanghai, China) and cell homogenates according to the product manual. The absorbance was measured at 450 nm using a microplate reader. Total SOD activity in cell homogenates was normalized to the total protein.

Extraction of RNA, library construction, and sequencing.

The WT strain and ΔisiA mutant samples for transcript analysis were collected at different time points. RNA-seq analysis was implemented using cultures at −Fe/−MV (iron-depleted) and −Fe/+MV (iron depleted-MV treated) with three biological replicates. The WT strain and ΔisiA mutant were cultured in iron-depleted media for 24 h; then 5 μM MV was added to the culture to be harvested for 4 h. Total cellular RNA was extracted using TRIzol reagent (Invitrogen, USA) according to a method that was previously described (53). RNA quality was assessed using the Nanodrop 2000 (Thermo Fisher Scientific). Total RNA was isolated from the samples for RNA sequencing, and the RNA quality was analyzed with a minimum integrity number (RIN) value of 7. Transcriptome measurement was completed by the Personalbio Company (Nanjing, China) and the RNA sequencing by the Illumina HiSeq Xten platform.

Quantitative real-time PCR.

Primers for qRT-PCR were designed using the Beacon Designer 7.0 software. For each RNA sample, the removal of the genomic DNA and reverse transcription were completed using the HiScript II Q Select RT supermix for qPCR (+gDNA wiper) kit (Vazyme). For specific steps, refer to the product manual. The qPCR was implemented according to a method that was previously described (54). Three biological repetitions were measured with three technical replicates for each gene. The Synechocystis gene rnpB, which is an effective reference gene for qRT-PCR normalization (55), was used as a reference.

RNA-seq analysis.

The raw data were first preprocessed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc) to obtain a high-quality sequence. After removing the adapter sequences, low-quality sequences were collected with an N percentage of over 5% and those with more than 50% bases with a false-discovery rate (FDR) (q value) ≤ 5. Further, for sequences shorter than 35 bases, the clean reads from 12 different samples were obtained. Next, the processed sequence was aligned to the reference genome of Synechocystis sp. PCC 6803 (GCF_000340785.1_ASM34078v1_genomic.fna). Genes were abundantly normalized using the FPKM values. We used DESeq (version 1.18.0) to analyze the differential expression of genes. The absolute value of log₂FoldChange >1 and the false discovery rate (FDR) <0.05 were used to identify the significance of different gene expression.

Construction of coexpression modules based on WGCNA.

The coexpressed gene was identified by calculating Pearson correlation coefficients (PCCs) between a pair of genes in the 12 samples. To build a reliable gene coexpression network, the PCCs of each pair of genes should be higher than 0.9, and genes with a P value of less than 0.05 are considered significantly coexpressed genes. The coexpression network analysis was performed using the weighted gene coexpression network analysis (WGCNA) package version 1.61 in the R software package (56) (http://www.r-project.org/). This analysis was based on the conserved DEGs between the WT strain and ΔisiA mutant under MV treatment. The networks were performed with an established cutoff weight parameter at 0.3 and visualized using Cytoscape version 3.6.0 (57). A proper power value was determined using the soft-thresholding method with the degree of independence at 0.8. Once the power value was determined, the adjacency was transformed into a topological overlap matrix (TOM). Genes with similar expression profiles were classified into gene modules by performing average linkage hierarchical clustering based on TOM-based dissimilarity measurements. The correlation between the module gene and stress treatments can be defined as a module-trait association.

Analysis of operon and verification.

The degree of correlation between adjacent genes was determined based on the gene expression levels in the RNA-seq data, and then it was inferred whether the genes were in the same operon. With regard to potential multiple-gene operons, PCCs were calculated (58). If two adjacent genes showed a higher correlation under certain conditions (PCC ≥ 0.9 and P value ≤ 0.05), these two genes were proven to belong to a unique operon; otherwise, the neighbor genes were considered to belong to two adjacent operons. The coverage plot and read alignments were produced using the Integrative Genomics Viewer (IGV) (59). Primers were designed between the operon genes and amplified by genomic DNA, RNA, digested RNA, and cDNA.

Prediction of sRNA targets.

The identification of sRNA in bacteria included new transcripts from the Rockhopper analysis system (http://cs.wellesley.edu/~btjaden/Rockhopper) and secondary structure prediction via RNA fold (http://www.tbi.univie.ac.at/) (Fig. S8). The three identified sRNAs were used to determine their respective target genes. The miRanda was applied to predict the interaction of sRNA-mRNA in the present study (60). Cytoscape was used to conduct regulatory network visualization for all the sRNAs and their target genes.

Data availability.

The transcriptome sequencing data were submitted to the NCBI SRA database (accession no. SRP256110).