Tandem gene amplification restores photosystem II accumulation in cytochrome b559 mutants of cyanobacteria.

Cytochrome (Cyt) b559 is a key component of the photosystem II complex (PSII) that is essential for its proper functioning and assembly. Site-directed mutants of the model cyanobacterium Synechocystis sp. PCC6803 with mutated heme axial ligands of Cyt b559 have little PSII and are therefore unable to grow photoautotrophically. Here we describe two types of Synechocystis autotrophic transformants that retained the same mutations in Cyt b559 but are able to accumulate PSII and grow photoautotrophically. Whole-genome sequencing revealed that all of these autotrophic transformants carried a variable number of tandem repeats (from 5 to 15) of chromosomal segments containing the psbEFLJ operon. RNA-seq analysis showed greatly increased transcript levels of the psbEFLJ operon in these autotrophic transformants. Multiple copies of the psbEFLJ operon in these transformants were only maintained during autotrophic growth, while its copy numbers gradually decreased under photoheterotrophic conditions. Two-dimensional PAGE analysis of membrane proteins revealed a strong deficiency in PSII complexes in the Cyt b559 mutants that was reversed in the autotrophic transformants. These results illustrate how tandem gene amplification restores PSII accumulation and photoautotrophic growth in Cyt b559 mutants of cyanobacteria, and may serve as an important adaptive mechanism for cyanobacterial survival.

Introduction

Photosystem II (PSII) is a large multisubunit pigment–protein complex located in the thylakoid membranes of cyanobacteria, algae, and plants (Barber, 2006; Umena et al., 2011). This complex utilises light energy to extract electrons from water and drive the transfer of electrons to plastoquinone, which is accompanied by the release of oxygen as a byproduct.

The PSII reaction centre complex (RCII) consists of the protein subunits D1, D2, cytochrome b 559 (Cyt b 559), and PsbI (Barber, 2006; Umena et al., 2011). The chlorophyll (Chl)a‐binding proteins CP47 and CP43 associate with RCII and function as the core antenna. In addition, several (at least 14) other protein subunits (mostly low molecular weight) are associated with fully assembled PSII. The biogenesis of PSII in cyanobacteria has been described as the stepwise assembly of four preassembled pigment–protein modules (D2mod, D1mod, CP47mod, and CP43mod) into intermediate complexes, which ultimately form the functional PSII holoenzyme (Komenda et al., 2004; Nixon et al., 2010; Nickelsen & Rengstl, 2013; Kiss et al., 2019; Zabret et al., 2021). The assembly of PSII is facilitated by a number of accessory or auxiliary protein factors that are not present in the fully assembled functional PSII (Komenda et al., 2004; Nixon et al., 2010; Nickelsen & Rengstl, 2013; Kiss et al., 2019; Zabret et al., 2021).

Cyt b 559 is a heme‐bridged heterodimer protein comprising one α‐subunit and one β‐subunit (subunits PsbE and PsbF encoded by psbE and psb Fgenes, respectively) (Shinopoulos & Brudvig, 2012; Chu & Chiu, 2016; and references therein). In the model cyanobacterium Synechocystis sp. PCC 6803 (from this point forwards Synechocystis), PsbE and PsbF provide His‐22 residues as axial ligands for the noncovalently bound heme of Cyt b 559. PSII is susceptible to photodamage, and several studies have suggested that Cyt b 559 may participate in secondary electron transfer pathways that protect PSII against photoinhibition (Barber & De Las Rivas, 1993; Bondarava et al., 2003; Shinopoulos & Brudvig, 2012; Chu & Chiu, 2016; and references therein). Cyt b 559 subunits also interact with D2 to form the essential intermediate complex during the early steps of PSII assembly (Komenda et al., 2004; Kiss et al., 2019). This concept has been supported by mutagenesis studies in Synechocystis, the green alga Chlamydomonas reinhardtii, and tobacco (Nicotiana tabacum), all showing that the assembly of active PSII reaction centres requires the presence of both the α‐ and β‐subunits of Cyt b 559 (Pakrasi et al., 1988; Morais et al., 1998; Swiatek et al., 2003).

To investigate the exact physiological function of Cyt b 559 in PSII in Synechocystis, several mutagenesis studies were performed to characterise a series of site‐directed mutants with mutations in the His‐22 residues of PsbE and PsbF (Pakrasi et al., 1991; Hung et al., 2007). Most of these Cyt b 559 mutants did not accumulate a sufficient amount of active PSII and therefore were unable to grow photoautotrophically. These findings suggest that the coordination of the heme cofactor in Cyt b 559 is important for the assembly or stability of PSII in Synechocystis (Pakrasi et al., 1991; Hung et al., 2007, 2010). Interestingly, we previously identified several spontaneously generated autotrophic transformants/pseudorevertants during the segregation of a Cyt b 559 mutant on agar plates (Hung et al., 2007). However, how these strains restored their ability for photoautotrophic growth was unclear.

In the present study, we report the characterisation of several previously identified Cyt b 559 autotrophic transformants as well as newly generated autotrophic transformants using an antenna attenuation method (Fig. 1). We performed whole‐genome sequencing, genomic DNA analysis, RNA sequencing (RNA‐seq), and two‐dimensional protein analysis on both groups of Cyt b 559 transformants to reveal how they regained their ability to accumulate PSII and grow photoautotrophically. This study reveals a novel survival mechanism used by cyanobacteria to overcome mutation‐induced instability of important protein components.

Fig. 1

Procedures for generation of Type A (H22KαPS+ and H22YβPS+) and Type B (H22YαPS+ and H22CβPS+) Cyt b 559 autotrophic transformants.

Materials and Methods

Mutant strains and growth conditions

Wild‐type (WT) and most Cyt b 559 mutant stains used in this study were generated and described previously (Hung et al., 2007). The original H22Kα mutant strain (H22KαPS+) (Hung et al., 2007) carried the tandem amplification of chromosomal segments containing the psbEFLJ operon. We generated the new H22Kα mutant strain by transforming the mutant plasmid into the host strain (ΔpsbEFLJ) as described (Hung et al., 2007). The new H22Kα mutant strain showed the PS− phenotype (Table 1). Synechocystis cells were grown at 30°C under a light intensity of 25–30 μmol photons m−2 s−1 and continuous bubbling with sterile, humidified air in BG‐11 medium (photoautotrophic growth) or BG‐11 medium supplemented with 5 mM glucose and 10 μM DCMU (photoheterotrophic growth). Cultures in the exponential growth phase (OD730: 0.6–1.0) were harvested and used for Chl fluorescence measurements and functional analysis.

Table 1

General properties of two distinct types of Cyt b 559 autotrophic transformants.

Strains	PS growth	Oxygen evolution (% of wild‐type) a	PSII content (% of wild‐type) b	References
Wild‐type	+	100 ± 6	100 ± 2	This study
Spontaneously generated autotrophic transformants (Type A)
H22KαPS+	+	84 ± 11	81 ± 16	This study
H22YβPS+	+	41 ± 13	79 ± 19	This study
H22Kα	−	9 ± 1	28 ± 0	This study
H22Yβ	−	~ 0	13 ± 7	Hung et al. (2007).
Autotrophic transformants generated through the temporary antenna attenuation method (Type B)
H22YαPS+	+	54 ± 9	102 ± 12	This study
H22CβPS+	+	23 ± 7	46 ± 10	This study
H22Yα	−	~ 0	17 ± 3	Hung et al. (2007)
H22Cβ	−	~ 0	13 ± 5	Hung et al. (2007)

Measured as a steady‐state rate of oxygen evolution in cells using a Clark‐type oxygen electrode in the presence of dichlorobenzoquinone and ferricyanide.

Measured as variable fluorescence in the presence of DCMU and hydroxylamine.

Generation of H22α and H22β Cyt b 559 autotrophic transformant mutants by the antenna attenuation method

The pCB(ΔcpcC1C2) kmR and pCK(ΔcpcBAC1C2) kmR plasmids used to delete the corresponding PC rod‐forming protein components were constructed as described previously (Ughy & Ajlani, 2004; Thomas et al., 2006). Both plasmids were transformed into H22Yα and H22Cβ Cyt b 559 mutant cells (Hung et al., 2007). When colonies of transformants appeared on solid agar plates containing the antibiotics erythromycin (Em, 2.5 μg ml−1) and kanamycin (Km, 20 μg ml−1), the colonies were immediately re‐streaked and maintained on solid agar plates containing Em (2.5 μg ml−1) only.

Measuring photosynthetic oxygen evolution

Steady‐state rates of oxygen evolution in cells were measured using a Clark‐type oxygen electrode (Hansatech, Norfolk, UK) fitted with a water‐jacketed cell. A cell suspension containing 10 μg Chl was held at 25°C in a stirred, water‐jacketed cell. Potassium ferricyanide and 2,6‐dichloro‐p‐benzoquinone were added to the samples as artificial electron acceptors (final concentration 2 mM) to a total volume of 1 ml (Huang et al., 2016).

Measuring Chla fluorescence at 295 K

Chla fluorescence measurements at 295 K were performed with a Dual PAM (pulse‐amplitude‐modulation) fluorometer (Walz, Effeltrich, Germany). A cell suspension containing 16 μg Chl was diluted in PSII content buffer (50 mM MES‐NaOH, 25 mM CaCl2, 10 mM NaCl, pH 6.5) to a total volume of 1.6 ml and held in a stirred, water‐jacketed cell. The relative PSII content of cells on a Chl basis was estimated based on the total yield of variable Chla fluorescence (F m − F 0) measured in the presence of DCMU and hydroxylamine, according to Chu et al. (1994). To measure time‐dependent flash‐induced transients of fluorescence yield, WT or mutant cells were dark‐adapted for 5 min. A modulated nonactinic measuring beam (1.6 kHz) was then turned on to obtain the minimal fluorescence value, F 0. Multiple turnover flashes were applied to obtain the fluorescence level of F m, (maximal fluorescence in the dark) and (maximal fluorescence in the light). The stable level of fluorescence (F s) was determined in the presence or absence of red actinic light.

77 K fluorescence measurements

77 K fluorescence emission spectra were recorded with a fluorescence spectrometer (FP‐8300; Jasco, Tokyo, Japan) with two different excitation wavelengths: 435 nm to excite Chla (excitation bandwidth 2.5 nm, emission bandwidth 1 nm) and 580 nm to excite the phycobilisomes (excitation bandwidth 2.5 nm, emission bandwidth 1 nm) (Chu et al., 2012; Chiu et al., 2013). All cell suspension samples contained 10 μg Chl ml−1.

DNA‐seq library construction and whole‐genome sequencing

An amount of 200 ml cells was collected by centrifugation (6000  for 10 min) in 250 ml centrifuge bottles. The pellets were frozen in liquid nitrogen and the cell walls were immediately broken in liquid nitrogen using a mortar and pestle. Genomic DNA was extracted and purified from the samples using a DNeasy plant maxi kit (Qiagen). For Illumina sequencing, a paired‐end sequencing library was prepared with the KAPA LTP library preparation kit (Roche). The concentrations of the libraries were determined with the Qubit Fluorometer (Invitrogen), and their profiles were determined using a Fragment Analyzer High Sensitivity NGS fragment kit (Agilent, Santa Clara, CA, USA). The molar concentrations of the libraries were derived using qPCR normalisation with a KAPA NGS Library qPCR kit. The prepared libraries were pooled for paired‐end sequencing on the Illumina HiSeq 2500 platform at Yourgene Bioscience Co. (New Taipei City, Taiwan) with 125‐bp paired‐end sequencing reads. DNA‐seq reads were mapped to the Synechocystis genome using Bowtie2 and Blat. Whole‐genome shotgun sequencing involved the PacBio Sequel platform (Pacific Biosciences, Menlo Park, CA, USA). One gel‐plus (20 kb) library and one SMRT cell were used. The average read length was 7.1–7.6 kb and the coverage was 43–95%. De novo genome assembly involved using the Hierarchical Genome Assembly Process (HGAP) assembler v.4 (Chin et al., 2013).

Southern blot analysis

Southern blot analysis was performed with a DIG High Prime DNA Labeling and Detection Starter Kit I (Roche) according to the manufacturer's instructions. Briefly, genomic DNA (2 μg) was digested with KpnI and BsgI enzymes. pAC559EM (5 ng), used as a positive control, was digested with ScaI. The digested DNA was subjected to agarose gel electrophoresis and transferred onto a nylon membrane (Hybond‐N+; GE Healthcare, Waukesha, WI, USA). The EFLJ probe was prepared from the pAC559EM plasmid by PCR amplification with primers for B559F (TAGGCGGCTCACAAAATAGT) and for B559R (GGTTTGGGTTTGAATTTCTG).

Droplet digital PCR to detect variation in copy number

Droplet digital PCR (ddPCR) reactions were prepared to contain 10 pg Alu‐digested genomic DNA and 10 μl of 2× ddPCR Supermix for Probes (Bio‐Rad), 900 nM of each primer pair (for the endogenous reference gene and target gene), and 125 nM of each probe; the final volume was adjusted to 20 μl with sterile ultrapure water. The primer pairs and probes are listed in Table S1. The cpcD DNA fragment was used as an endogenous control. 20 μl of each sample and 70 μl of Bio‐Rad Droplet Generation Oil were transferred to the respective wells in the cartridges and attached to gaskets. Droplets were generated in a QX200 droplet generator (Bio‐Rad). After thermal cycling, the plates were transferred to a QX200 droplet reader (Bio‐Rad). The software package provided with the ddPCR system (QuantaSoft 1.7.4.0917; Bio‐Rad) was used for data acquisition.

RNA‐seq library preparation and next‐generation sequencing

Total RNA was isolated from the samples using a miRNeasy mini kit (Qiagen). The quantity and quality of the extracted RNA were determined using a NanoDrop spectrophotometer and an Agilent 2100 Bioanalyzer (Agilent). To prepare strand‐specific RNA‐seq libraries, 5 µg purified RNA was depleted of ribosomal RNA using the Ribo‐Zero magnetic kit (Bacteria; Epicentre, Madison, WI, USA) following the manufacturer's instructions. RNA‐seq libraries were prepared with the Illumina TruSeq stranded mRNA sample preparation kit (Illumina). The libraries were pooled for paired‐end sequencing on the Illumina HiSeq 2500 platform with 125‐bp paired‐end sequencing reads. Sequencing reads were preprocessed for transcriptomic data analysis using the Fastx Toolkit (v.0.0.13) to remove low‐quality bases and reads < 20 bp. After removing reads that mapped to noncoding RNA sequences with Bowtie v.2.0.0 with default settings, the remaining qualified reads were mapped to the Synechocystis genome using Bowtie with default parameters. Gene expression levels were calculated with the standard reads per kilobase of gene per million mapped reads (RPKM) method. Genes with two‐fold or greater changes in expression vs the control in both biological replicates were considered differentially regulated genes.

cDNA synthesis and quantitative real‐time PCR

First‐strand cDNA was synthesised from 1.5 μg purified RNA using the SuperScript III First‐Strand Synthesis System (Invitrogen) with random hexamers. The constitutively expressed gene rnpB was used as a reference. The resulting cDNA was amplified with gene‐specific forward and reverse DNA oligonucleotides, as listed in Table S2. Quantitative RT‐PCR was performed on an ABI 7500 Real‐Time PCR System (Applied Biosystems, Waltham, MA, USA), and the amount of cDNA was detected with SYBR Green dye.

Radioactive labelling of proteins in vivo

Radioactive pulse labelling of cells was performed at 30°C for 30 min under 500 µmol photons m−2 s−1 of illumination using a mixture of [35S]Met and [35S]Cys (Hartmann Analytic Gmbh, Braunschweig, Germany) as described (Dobáková et al., 2007). For each strain, labelling was performed using the same number of cells corresponding to 75 µg Chl in the WT, which were re‐suspended in 250 µl BG‐11.

Preparation of cellular membranes

Cells in the exponential growth phase (100 ml) were pelleted, washed, and re‐suspended in buffer B (25 mM MES/NaOH, pH 6.5, 10 mM CaCl2, 10 mM MgCl2, 25% glycerol (v/v)). The cells were broken mechanically in a Mini‐Beadbeater (BioSpec, Bartlesville, OK, USA) using balotina beads. The membrane and soluble protein fractions were separated by high‐speed centrifugation (47 000  , 20 min). The pelleted membranes were washed and re‐suspended in buffer B.

Two‐dimensional PAGE analysis of membrane proteins and immunoblotting

Two‐dimensional analysis of membrane proteins was performed using a combination of clear native (CN) PAGE in a 4% to 12% gradient gel and SDS‐PAGE in a 12% to 20% gradient gel containing 7 M urea (2D CN/SDS‐PAGE) (Kiss et al., 2019). For immunoblotting, the gels were stained with SYPRO Orange (Sigma‐Aldrich); proteins from the gel were transferred onto a PVDF membrane that was incubated with a specific primary antibody, followed by a secondary antibody conjugated with horseradish peroxidase (Sigma‐Aldrich). The primary antibodies against D1, D2, CP43, PsbE and PsbF used in this study, which were prepared in rabbits, were described previously (Komenda et al., 2008). Two‐dimensional gels with radioactively labelled proteins were stained with Coomassie Brilliant Blue, photographed, dried, exposed to a phosphorimager plate (GE Healthcare) overnight, and scanned using a Storm 860 phosphorimager (GE Healthcare).

Results

Generation of a new type of autotrophic Cyt b 559 transformant by the antenna attenuation method

To characterise the changes responsible for the conversion of the nonautotrophic axial histidine Cyt b 559 mutants into autotrophic strains, we analysed two different types of these autotrophic transformants. Type A autotrophic transformants, in which the His22 residues in PsbE and PsbF were replaced by Lys and Tyr, were designated H22KαPS+ and H22YβPS+, respectively; these transformants were previously isolated by transforming a strain lacking psbEFLJ with the corresponding plasmid containing the mutated gene, following by segregation on glucose‐free agar plates, as described previously (Hung et al., 2007) (Fig. 1). Type B autotrophic transformants were prepared with the antenna attenuation method applied to the original nonautotrophic Cyt b 559 histidine mutants H22Yα and H22Cβ (Hung et al., 2007). This method is based on the parallel induction of autotrophic transformants and the site‐directed removal of phycobilisome rod linker proteins LR33 and LR30 (ΔcpcC1C2, abbreviated CB), resulting in a strain lacking two of three phycocyanin (PC) hexamers per rod (Ughy & Ajlani, 2004; Hung et al., 2010); or (2) all PC rod‐forming subunits, resulting in mutant cells lacking PC and containing only the allophycocyanin (APC) core (Ughy & Ajlani, 2004; Thomas et al., 2006) (Fig. 1). During this process, the decrease in antenna size may provide photoprotection or better resource allocation to autotrophic transformants, therefore leading to their more efficient generation and better survival.

Without this antenna attenuation step, no autotrophic transformant/pseudorevertants were obtained. However, once autotrophic transformants were generated, the reduced antenna size became deleterious for autotrophic proliferation and, in the absence of selection antibiotics (Km), they recovered the capacity to assemble a full phycobilisome (Table S3). When the autotrophic transformants were further propagated on agar plates containing Km, they lost their ability to undergo photosynthetic growth and died. We generated a number of these autotrophic transformants but, for the purpose of this study, as representatives of Type B autotrophic transformants, we used only two strains with the His22 residues of PsbE and PsbF replaced by Tyr and Cys residues, from this point forwards designated as H22YαPS+ and H22CβPS+, respectively.

Growth, oxygen evolution activity and PSII contents of mutant cells

The growth characteristics, light‐saturated oxygen evolution activity, and PSII content of the representative mutant strains are listed in Table 1. All autotrophic transformants were able to grow photoautotrophically and assemble appreciable amounts of PSII complexes, as judged from their oxygen‐evolving activities and quantification of PSII levels via measurement of variable fluorescence in the presence of DCMU and hydroxylamine (Chu et al., 1994). Among the autotrophic transformants, H22KαPS+ cells showed the highest oxygen evolution activity (c. 84% compared with WT), whereas H22CβPS+ showed the lowest oxygen evolution activity (c. 23% compared with WT). By contrast, all corresponding Cyt b 559 mutant strains (H22Kα, H22Yα, H22Yβ, and H22Cβ) failed to grow photoautotrophically, showed little or no oxygen evolution activity, and could not accumulate appreciable amounts of PSII core complexes (Table 1). Under our experimental growth conditions for photoautotrophic growth (light intensity of 30 μmol photons m−2 s−1), the doubling times for WT, H22KαPS+, H22YβPS+, H22YαPS+ and H22CβPS+ cells were c. 13, 14, 26 (Hung et al., 2007), 23 and 53 h, respectively (Fig. S1).

Maximum quantum yield of PSII photochemistry (F v/F m) was recovered in Cyt b 559 autotrophic transformants

Compared with WT and Cyt b 559 mutant cells, autotrophic transformant cells showed distinct time‐dependent, flash‐induced transient fluorescence yields in the presence and absence of actinic light (Fig. 2a). WT cells had a lower F 0 value (c. 0.13) and higher F v/F m value (c. 0.43) (Fig. 2a, panel i) than the autotrophic transformants, whose F 0 values ranged from 0.15 to 0.24 and F v/F m values from 0.41 to 0.15 (Fig. 2a, panels ii–v). By contrast, the corresponding Cyt b 559 mutant H22Kα and H22Yα strains failed to grow photoautotrophically and had higher F 0 values (0.29 and 0.33, respectively) and much lower F v/F m values (c. 0.01) (Fig. 2a, panels vi–vii) than the autotrophic transformants. In addition, the maximum fluorescence () in Cyt b 559 autotrophic transformant cells significantly increased during illumination with actinic light and decreased after this treatment. However, for WT, the values were approximately the same before, during and after actinic light illumination. The changes in maximal fluorescence yield elicited by actinic light in the autotrophic transformants can be attributed to state transitions (Huang et al., 2018; Calzadilla & Kirilovsky, 2020; and references therein) and to the fluorescence emissions from uncoupled phycobilisomes (Chiu et al., 2013).

Fig. 2

Photosynthetic characterisation of WT and Cyt b 559 mutant cells. (a) Time‐dependent flash‐induced transients of PSII fluorescence yield of WT and Cyt b 559 mutant cells with and without red actinic light. The light intensity of red actinic light was c. 40 μmol photons m–2 s–1. Up arrow (↑) indicates actinic light on. Down arrow (↓) indicates actinic light off. (b) 77 K fluorescence emission spectra from WT and Cyt b 559 mutant cells. Spectra were normalised to emission of PS I. au, arbitrary units.

Finally, most autotrophic transformants and especially the H22CβPS+ mutant showed decreased steady‐state fluorescence below the F 0 level that gradually increased after actinic light illumination was switched off (Fig. 2a, panel v). This result suggests that a fluorescence quencher was induced by the actinic light in a significant fraction of PSII in the autotrophic transformants, especially the H22CβPS+ mutant. This phenomenon has been attributed to back electron flow from dark reduction of the plastoquinone pool to PSII and is typical of acceptor‐side photoinhibition (Bondarava et al., 2003; Ohad et al., 2004; Chiu et al., 2013).

77 K fluorescence emission spectra

The top panel of Fig. 2(b) shows the 77 K fluorescence emission spectra for cells excited at 435 nm, a wavelength that Chl molecules preferentially absorb. In the spectra for WT and Cyt b 559 autotrophic transformant cells, preferential excitation of Chl at 435 nm resulted in three emission peaks: at c. 686, c. 695, and c. 724 nm. The emission peak at c. 686 nm originates from CP43 and the peaks at c. 695 and c. 724 nm originate from functional PSII and PSI reaction centres, respectively (Andrizhiyevskaya et al., 2005; Boehm et al., 2011; Lamb et al., 2018; Calzadilla & Kirilovsky, 2020; and references therein). Therefore, the increased emission peaks at c. 695 and c. 686 nm in the spectra for H22YαPS+, H22KαPS+, and H22YβPS+ may indicate that the ratio of PSII to PSI and the level of unassembled CP43 in these autotrophic transformants were higher than in the WT. In addition, the low emission peaks at c. 695 nm for the H22CβPS+, H22Kα and H22Yα mutants were associated with lower PSII content in these mutants in comparison with WT and the other mutants (Fig. 2a).

The bottom panel of Fig. 2(b) shows the 77 K fluorescence emission spectra recorded from cells excited at 580 nm, a wavelength that is preferentially absorbed by phycobilin pigments. The WT strain showed emission peaks with maxima at c. 664 nm (from APC), c. 690 nm (from PSII), and c. 723 nm (from PSI) (Ughy & Ajlani, 2004; Hung et al., 2007, 2010). In the spectra for the Cyt b 559 autotrophic transformants as well as H22Kα and H22Yα mutants, the emission peaks were strongly enhanced and shifted to c. 686 nm. The emission peaks at c. 686 nm mainly originates from terminal phycobilin emitters (Lamb et al., 2018; Calzadilla & Kirilovsky, 2020). Therefore, the enhanced emission peaks at 686 nm in these mutant strains indicate that the energy transfer from phycobilisomes to PSII reaction centres was significantly inhibited or uncoupled in these cells.

The genomes of the Cyt b 559 autotrophic transformants carry tandem repeats containing the psbEFLJ operon

To identify the possible second‐site mutation(s) in the genomes of the Cyt b 559 autotrophic transformants, we sequenced their entire genomes and mapped the next‐generation sequencing data to the Synechocystis genomic sequence. These autotrophic transformants did not carry significant second‐site mutation(s), but all had > 5–15 times increased read coverage in the regions containing psbEFLJ operons in their genomes (Fig. 3a). These results suggest that the autotrophic transformants carry high copy numbers of DNA segments containing the psbEFLJ operon. The multiplication units in the H22YαPS+ and H22CβPS+ (Type B) strains were c. 24 and 96 kb, respectively.

Fig. 3

Genomic analysis of autotrophic transformant cells. (a) Copy number, size and location of repeat elements in autotrophic transformant cells. (b) Copy number variation of psbEFLJ and pUC119 backbone in WT and autotrophic transformants by droplet digital PCR (ddPCR) (n = 3). (c) ddPCR analysis of the time course of copy number variation of psbEFLJ in autotrophic transformants grown under photoheterotrophic conditions and then returned to photoautotrophic conditions (n = 2). (d) ddPCR analysis of the time course of copy number variation of psbEFLJ in autotrophic transformants grown under photomixotrophic conditions (n = 2). Data are the mean ± SD.

The copy number of the psbEFLJ operon in the autotrophic transformants varies under different growth conditions

We performed ddPCR to determine the copy number variation of psbEFLJ in the transformants by comparing the ratios of the psbEFLJ gene products with the reference cpcD products. The high copy number of the psbEFLJ genes was verified in all autotrophic transformants (Fig. 3b). By contrast, all corresponding Cyt b 559 mutant strains (H22Kα, H22Yβ, H22Yα, and H22Cβ) failed to grow photoautotrophically, showed a single copy number of psbEFLJ genes and did not carry the pUC119 plasmid sequence in their genomes. In addition, we used ddPCR analysis to characterise the stability of the number of psbEFLJ operon repeats in these strains under photoheterotrophic conditions (plus the PSII inhibitor DCMU and glucose). Under these conditions, the number of repeats in the Type A transformants gradually decreased from c. 20 to 3 copies within 8 wk of photoheterotrophic growth (Fig. 3c). However, when the cultures were shifted back to photoautotrophic growth conditions, the copy numbers of the psbEFLJ genes gradually increased and returned to the original levels. Furthermore, under photomixotrophic conditions (with glucose), the number of repeats in the Type A transformants gradually decreased from c. 20 to c. 8 copies within 8 wk (Fig. 3d). The results suggest that the high copy number of the psbEFLJ genes is required for photoautotrophic growth of the autotrophic transformants.

Repeats of the psbEFLJ operon are located on chromosomes of the autotrophic transformants

To examine whether the psbEFLJ operon repeats were located on chromosomes or plasmids of the autotrophic transformants, we performed Southern blot analysis. We used the restriction enzymes KpnI and BsgI to digest the chromosome because their restriction sites are located outside of the psbEFLJ operon (Fig. 4a). When we used a psbEFLJ probe to identify fragments containing the psbEFLJ operon, WT and the Type B transformants showed one hybridisation band of c. 5.3 kb (Fig. 4b). By contrast, the hybridisation bands in the Type A autotrophic transformants were much larger (> 23 kb). These results confirm that the mutant plasmids were integrated into (and amplified in) the chromosomes of the Type A transformants. In addition, although the hybridisation bands in the Type B transformants were the same size as for the WT, the bands were more intense in the transformants compared with WT, which is consistent with the amplification of the psbEFLJ operon.

Fig. 4

Schematic representation of chromosome structure (a) and Southern blot analysis (b) of WT and Cyt b 559 mutants. The mutation site of H22Hα or H22β is indicated as a star (*) in (a). Restriction enzyme site: H, HindIII; E, EcoRI; K, KpnI; B, BsgI. Genomic DNA was digested with KpnI and BsgI. pAC559Em plasmids were a positive control. The probe used in (b) was the EFLJ probe.

Sequence analysis of the amplified chromosome segments in autotrophic transformants

All autotrophic transformants examined retained the heme‐ligand mutations, as revealed by whole‐genome sequencing and PCR Sanger sequencing (Table S4). In addition, the multiplication units in all autotrophic transformants were direct tandem repeats. The repeats in H22KαPS+ transformants exactly matched the size of the pAC559Em plasmid (c. 7 kb). In H22YβPS+ transformants, about half of repeats had the same size as the pAC559Em plasmid and the others had a reduced size due to deletion of EmR (coding for Em resistance gene) (Fig. S2). We also analysed the border and junction sequences flanking the multiplication units. No sequences were inserted or deleted in junctions between the repeats. However, the border sequences of the multiplication units in the Type B transformants all differed (Fig. 5a). A hexanucleotide sequence (5′‐TCCATT‐3′) was present in border sequences of the multiplication units in H22YαPS+ and H22CβPS+. In addition, the sequence 5′‐TTCGTT‐3′ was present in both the 5′‐end and 3′‐end border sequences as well as in new junction sequences between the individual repeats within the multiplication units in H22CβPS+ but was absent in all other autotrophic transformants. Furthermore, the Type B autotrophic transformants all had a very short (1–5 nucleotides) homologous junction sequence flanking the multiplication units (Fig. 5b). The mechanism of tandem gene multiplication in the Type B transformant is likely to be initiated by RecA‐independent illegitimate recombination that utilises very short homologous sequences (Fig. 5c) (Andersson & Hughes, 2009; Reams & Roth, 2015).

Fig. 5

Analysis of the sequences at the borders and junctions of the amplified chromosome fragments in autotrophic transformants. (a) Border sequences: two short regions of homologous sequences (5′‐TCCATT‐3′ and 5′‐TTCGTT‐3′) are shown as underline and bold type, respectively. (b) Novel junction sequences: the presence of very short (1–5 nucleotides) homologous junction sequences in Type B autotrophic transformants are indicated in boxes and red colour. (c) A model for tandem gene amplification of chromosome segments containing the psbEFLJ operon (marked with an asterisk) in Type B autotrophic transformants.

The psbEFLJ operon and several photosynthesis‐related genes were differentially expressed in the autotrophic transformants

The transcript levels of psbEFLJ genes were 16‐fold to 27‐fold and 6‐fold to 23‐fold higher in the Type A and Type B autotrophic transformants, respectively, compared with WT, as revealed by RNA‐seq (Table 2) and quantitative RT‐PCR (Fig. 6). The other photosynthesis‐related genes that were significantly upregulated (2‐fold to 7‐fold) in most autotrophic transformants vs WT included psbD2 (slr0927, the gene for D2), hliB (ssr2595, the gene for high light‐inducible HliB stress protein (He et al., 2001)), ftsH3 (slr1604, the gene for FtsH3 protease (Komenda et al., 2006)) and bicA (sll0834, the gene for a low‐affinity, high‐flux Na+‐dependent SulP type bicarbonate transporter (Wang et al., 2019)). The significantly downregulated photosynthesis‐related genes in these transformants included sll1471 (cpcG2, the gene for a phycobilisome linker (Kondo et al., 2007)), cmpA/B (slr0040/slr0041), sbtA/B (slr1512/1513), cupA (sll1734) and ndhD3 (sll1733) (involved in inorganic carbon transport systems (Omata et al., 1999; Zhang et al., 2004)), sigE (sll1689, encodes an RNA polymerase sigma factor involved in regulating sugar catabolism and photosystem‐related genes (Tokumaru et al., 2018)) and sll0450 (norB, the gene for a quinol‐dependent single‐subunit nitric oxide reductase (Busch et al., 2002)) (Table 3). Furthermore, genes within the tandemly amplified segments in the Type B transformants (e.g. from region SGL_RS04470 to SGL_RS04590 in H22YαPS+) were all upregulated. Therefore, the high copy numbers of these genes resulted in high transcript levels (Table S5).

Table 2

A list of common upregulated photosynthesis‐related genes in autotrophic transformant cells.

Gene ID	Symbol	Fold change related to the level in WT				Description
		H22KαPS+	H22YβPS+	H22YαPS+	H22CβPS+
Photosystem II
slr2034	ycf48	2.2	2.7	12.8	5.9	PS II assembly factor Ycf48
ssr3451	psbE	16.4	25.0	21.3	5.5	Cyt b 559 subunit alpha
smr0006	psbF	15.6	23.3	21.0	5.6	Cyt b 559 subunit beta
smr0007	psbL	16.0	25.9	22.8	7.9	PsbL
smr0008	psbJ	15.2	26.5	21.7	7.0	PsbJ
sll0851	psbC	1.5	2.5	2.7	2.6	CP43
sll0849	psbD1	1.7	2.8	2.9	2.4	D2 protein
slr0927	psbD2	2.3	5.9	5.0	4.9	D2 protein
Photoprotection
slr1604	ftsH3	1.7	3.1	2.9	5.1	ATP‐dependent metallopeptidase
ssr2595	hliB	2.7	3.3	2.5	3.2	High light‐inducible protein
slr0228	ftsH2	1.4	2.0	1.9	2.7	ATP‐dependent metalloprotease
Transporters
sll0834	bicA	1.7	2.8	6.4	3.0	Bicarbonate transporter BicA

Fig. 6

Changes in transcript abundance in WT and autotrophic transformants under photoautotrophic growth conditions using real‐time PCR. The primer pairs are described in Table S5. The transcript level of rnpB was used as a reference. Data are the mean ± SD from three independent cultivation experiments.

Table 3

A list of common downregulated photosynthesis‐related genes in autotrophic transformant cells.

Gene ID	Symbol	Fold change related to the level in WT				Description
		H22KαPS+	H22YβPS+	H22YαPS+	H22CβPS+
Photosynthesis
sll0219	flv2	−1.2	−2.3	−4.0	−3.1	Flavoprotein
sll0218		−1.2	−1.9	−4.6	−2.6	Hypothetical protein
sll0217	flv4	1.1	−1.1	−1.5	−2.6	Flavoprotein
sll1471	cpcG2	−4.5	−3.8	−40.3	−42.9	Phycobilisome rod‐core linker
Inorganic carbon transports
sll1734	cupA	−1.3	−2.4	−2.6	−4.0	CO2 hydration protein
sll1733	ndhD3	−1.4	−2.7	−2.4	−4.2	NADH dehydrogenase subunit
sll1732	ndhF3	1.0	−2.6	−1.8	−4.5	NADH dehydrogenase subunit
slr1512	sbtA	−1.3	−2.5	−9.6	−9.6	Sodium dependent bicarbonate transporter
slr1513	sbtB	−1.1	−1.8	−5.9	−4.8	Hypothetical protein
slr0040	cmpA	−1.7	−8.5	−16.9	−19.9	Transporter substrate‐binding protein
slr0041	cmpB	−1.6	−8.5	−25.5	−13.9	Bicarbonate transporter permease
slr0043	cmpC	−1.3	−2.5	−4.5	−3.2	Transporter ATP‐binding protein
slr0044	cmpD	−1.2	−2.3	−5.7	−5.0	Transporter ATP‐binding protein
Metabolism
sll1689	sigE	−1.7	−2.8	−4.1	−4.3	RNA polymerase sigma factor, sigE
sll0450	norB	−6.1	−12.1	−25.4	−18.4	Nitric oxide reductase

Analysis of membrane protein complexes reveals differences among WT, Cyt b 559 mutants and the corresponding autotrophic transformants

To evaluate the impact of the Cyt b 559 mutations and both types of pseudoreversions on the membrane composition and (especially) content and assembly state of PSII complexes, we analysed membranes of WT and the H22Kα, H22Yβ and H22Yα mutants and the corresponding autotrophic transformants H22KαPS+ and H22YβPS+ (Type A) and H22YαPS+ (Type B) using a combination of CN and SDS gel electrophoresis followed by protein immunodetection. CN‐PAGE analysis (Fig. 7) clearly documented the differences in PSI and PSII complex contents between the Cyt b 559 mutants and the corresponding autotrophic transformants. As judged from the Chl fluorescence scan of the gel, all Cyt b 559 mutants were strongly depleted in the assembled dimeric and monomeric PSII core complexes (PSII(2) and PSII(1)), but they contained a highly fluorescent band in a region containing small proteins most probably belonging to the unassembled PSII antennae (see results of 2D immunoblotting described below). In the autotrophic transformants, the levels of PSII(2) and PSII(1) markedly increased and were higher in the H22KαPS+ strain vs the two other autotrophic transformants; these findings are in agreement with the PSII contents determined by variable fluorescence measurements (Table 1). Furthermore, the increased PSII levels in the autotrophic transformants were accompanied by the depletion of unassembled antenna proteins. The contents of PSI complexes in the strains, especially the trimer, were proportional to the levels of PSII complexes, confirming the notion that PSI levels depend on the activity and accumulation of PSII complexes (Kiss et al., 2019).

Fig. 7

Analysis of photosynthetic complexes in site‐directed mutants of cytochrome b 559 and corresponding autotrophic transformants of Synechocystis. The strains were grown in BG11 containing 5 mM glucose at 40 µmol photons m−2 s−1. Designation of complexes: FP, free pigments; PSI(3) and PSI(1), trimeric and monomeric Photosystem I; PSII(2) and PSII(1), dimeric and monomeric PSII core complexes; RCCS, supercomplex of PSI and PSII; RC47, the monomeric PSII core complex lacking CP43; u.CP43 and u.CP47, unassembled PSII antennae. Loading was performed based on the same OD750 nm; in WT, this OD750 nm corresponded to 4 μg Chl.

To confirm the identity of the PSII complexes observed in the CN gel and to obtain a more complete information about their composition (focusing on the presence of mutated PsbE and PsbF subunits of Cyt b 559), we analysed the complexes that were separated by CN‐PAGE by SDS‐PAGE and immunoblotting in the second dimension (Figs 8, 9, S3). SYPRO staining and immunodetection with antibodies that specifically recognise D1, D2 and CP43 confirmed both the identity of the separated fluorescent complexes as PSII(2) and PSII(1) and the differences in their levels between the Cyt b 559 mutants and the corresponding autotrophic transformants. The PSII complexes contained all large proteins examined as well as both the PsbE and PsbF subunits. SYPRO staining and immunodetection confirmed the presence of the CP47 and CP43 antennae in the fluorescence band in the region containing unassembled proteins in the mutants (on 2D stained gels, indicated by arrows).

Fig. 8

Two‐dimensional PAGE analysis of photosynthetic complexes in (a) WT, (b) the site‐directed mutant H22Kα and (c) its corresponding autotrophic transformant H22KαPS+ of Synechocystis sp. PCC6803. The strains were grown in BG11 containing 5 mM glucose at 40 µmol photons m−2 s−1. Designation of complexes as in Fig. 7. Arrows in the blot designate unassembled D1. Loading was based on the same OD750 nm; in WT, this OD750 nm corresponded to 4 μg Chl.

Fig. 9

Two‐dimensional PAGE analysis of photosynthetic complexes in (a) the site‐directed mutant H22Yα and (b) its corresponding autotrophic transformant H22YαPS+ of Synechocystis sp. PCC6803. The experimental conditions were the same as in Fig. 8, designation of complexes as in Fig. 7. Arrows in the blot designate unassembled D1.

A more detailed look at individual protein contents showed that the Cyt b 559 mutants contained low levels of unassembled D1 protein (indicated by an arrow) but lacked any unassembled D2. Because we previously showed that the absence of PSII in the strain lacking the psbEFLJ operon is related to the inhibition of D2 synthesis (Komenda et al., 2004), the depletion of PSII in the mutant is most likely to be due to the lack of functional PsbE and PsbF. To verify this hypothesis, we radioactively labelled H22K mutant cells and the autotrophic transformant H22KαPS+ and compared the patterns of their labelled membrane proteins using 2D CN/SDS‐PAGE combined with autoradiography (Fig. 10). Notably, in both strains, most D1 protein (in all three forms: pD1 precursor, iD1 intermediate, and D1 mature protein (indicated by a triple arrow)) was identified in the region of unassembled proteins together with labelled CP43 and CP47. However, unlike the Cyt b 559 mutant, the autotrophic transformant contained some labelled D2 protein in PSII(1) and PSII(2). These results support the notion that insufficient biosynthesis of D2 due to a deficiency of PsbE and PsbF is the reason for the depletion of PSII and the lack of autotrophic growth of the mutants with defects in the axial ligands of Cyt b 559.

Fig. 10

Two‐dimensional PAGE analysis of radioactively labelled photosynthetic complexes in (a) WT, (b) the site‐directed mutant H22Kα and (c) its corresponding autotrophic transformant H22KαPS+ of Synechocystis. The strains were grown in BG11 containing 5 mM glucose at 40 µmol photons m−2 s−1 and radioactively labelled with a mixture of [35S] methionine and cysteine. Designations of complexes are as in Fig. 7. Triple arrow indicates labelled unassembled pD1, iD1 and D1 (from top to bottom). Loading was based on the same OD750 nm; in WT, this OD750 nm corresponded to 4 μg Chl.

Discussion

In the current study, regardless of how the autotrophic transformants were obtained, whole‐genome sequencing revealed that all four strains lacked true second‐site mutations, but all contained 5‐fold to 15‐fold tandem repeats of the chromosomal regions containing the psbEFLJ operon with the original Cyt b 559 mutation (Fig. 3a). The sizes of the tandem repeat units depended on the procedure used to obtain these autotrophic transformants. Type A transformants such as H22KαPS+ and H22YβPS+, containing c. 7‐kb repeat units, were generated during the selection of the psbEFLJ‐less strain for autotrophy using a plasmid containing the mutated psbE/F gene. The size of the repeat units present in these strains corresponded well to the size of the psbEFLJ operon, resistance cassette, flanking regions comprising parts of the ycf48 and proB genes, and the rest of the plasmid. There were no additional nucleotide sequences at the border of the whole multiplication unit or at the internal junction points between the individual repeats. This type of tandem multiplication probably involved integration of the entire mutant plasmid into the genome, followed by its tandem multiplication via rolling cycle replication or other mechanisms (Andersson & Hughes, 2009; Reams & Roth, 2015). In H22YβPS+ transformants, about half of the repeats had the reduced size due to deletion of EmR (Fig. S2). This could be a secondary event occurring during tandem multiplication. Notably, an ISY203a transposase gene (slr2036) is located downstream of the psbEFLJ operon, but whether this transposase plays a significant role in the tandem multiplication process is unknown.

By contrast, Type B transformants such as H22CβPS+ contained much larger repeat units (up to 96 kb). The presence of these repeat units points to the internal spontaneous amplification of a large DNA segment. Type B transformants do not contain a common repetitive sequence flanking multiplication units, but all have a very short (1–5 nucleotides) homologous junction sequence flanking its multiplication units. Because the RecA‐mediated homologous recombination generally requires lengths of repetitive sequences of at least 20–40 nucleotides (Andersson & Hughes, 2009), the mechanism of tandem multiplications in the Type B transformant may be initiated by illegitimate recombination that utilises a very short homologous sequence and could be further amplified by RecA‐mediated homologous recombination (Fig. 5c) (Andersson & Hughes, 2009; Reams & Roth, 2015). Such gene amplifications occur in many organisms, from bacteria to higher eukaryotes, and may play important roles in evolutionary adaptation to selection pressure and stressful environmental conditions (Andersson & Hughes, 2009; Elliott et al., 2013; Reams & Roth, 2015). To our knowledge, this is the first report of targeted tandem gene amplification in cyanobacteria.

One recent study reported the tandem duplication of a large (110‐kb) chromosomal segment in a glucose‐tolerant substrain (GT‐W) of Synechocystis (Tichý et al., 2016). This tandemly duplicated large chromosomal segment harbours c. 100 genes, including several PSII‐related genes (such as psbA2 and psb28) and the gene for Mg‐protoporphyrin methylestercyclase. In addition, this chromosomal segment is bordered by two ISY100 transposases, sll0431 and sll1397, which might mediate this tandem multiplication process. The GT‐W strain showed aberrant PSII assembly and deficient Chl biosynthesis compared with the PCC strain of Synechocystis (Tichý et al., 2016). Notably, tandem multiplication in the genomes of GT‐W cells was selected during prolonged growth on glucose and disappeared in cells grown on agar plates under photoautotrophic conditions (Tichý et al., 2016). Similarly, we detected changes in tandem repeat numbers in the autotrophic transformants depending on growth conditions (photoautotrophic vs photoheterotrophic growth).

As revealed by RNA‐seq and quantitative RT‐PCR, the formation of multiple tandem repeats containing the psbEFLJ operon led to a 6‐fold to 27‐fold increase in psbEFLJ transcript level compared with the WT (Table 2). In these transformants, psbD2 (slr0927) was upregulated 2‐ to 6‐fold, pointing to a relationship between the expression of Cyt b 559 and the D2 protein, as psbD2 encodes this PSII reaction centre protein. Cyt b 559 is required for the synthesis of D2, and both proteins interact to form a D2‐Cyt b 559 precomplex, an initial PSII assembly intermediate (Müller & Eichacker, 1999; Komenda et al., 2008; Kiss et al., 2019). Our RNA‐seq and quantitative RT‐PCR analysis also revealed the downregulation of several photosynthesis‐related genes, such as cpcG2 and the flv4‐sll0218‐flv2 operon in most or all autotrophic transformant strains. The CpcG2 linker is involved in transferring energy from specific PC rods to PSI (Kondo et al., 2007). Products of the flv4‐sll0218‐flv2 operon are involved in dissipating excess energy from PSII (Bersanini et al., 2014; Shimakawa et al., 2015). The function of PSII was impaired in most autotrophic transformant strains, so the decreased transcript levels of cpcG2 and flv4‐sll0218‐flv2 may serve to balance the energy flow between PSII and PSI and optimise the overall photosynthesis efficiency of the autotrophic transformants. The above results are consistent with 77 K fluorescence data, which revealed an increase in the PSII to PSI ratio in most autotrophic transformant strains compared with WT cells (Fig. 2b), and native gel electrophoresis data (Fig. 7), which revealed a lower PSI complex levels in the transformants compared with WT.

In addition, many genes involved in high‐affinity inorganic carbon transport systems, such as cmpA/B (slr0040/slr0041), sbtA/B (slr1512/1513), cupA (sll1734) and ndhD3 (sll1733), were downregulated in all autotrophic transformants except H22KαPS+, which contained almost WT levels of PSII and oxygen‐evolving activity. The expression of the above genes involved in inorganic carbon transport systems in cyanobacteria was regulated by NADPH and α‐ketoglutarate concentrations (Daley et al., 2012). Due to impaired photosynthetic activity, these autotrophic transformant strains may be unable to generate sufficient levels of NADPH and ATP synthesis for active inorganic carbon transport.

The results of the analysis of membrane protein complexes in the autotrophic transformants compared with WT and Cyt b 559 mutants were quite consistent with the negative effect of the axial ligand mutations in each Cyt b 559 subunit on the accumulation of PsbE and PsbF Cyt b 559 subunits and the assembly of PSII in the mutant cells. The H22Kα, H22Yβ, and H22Yα strains contained only small amounts of PSII, but transformants with high copy numbers of psbEFLJ genes accumulated sufficient amounts of the proteins for the assembly of photochemically active PSII and for autotrophic growth. The transformants still differed in their levels of PSII and in their PSII dimer‐to‐monomer ratios. H22KαPS+, the strain with the highest oxygen‐evolving activity, also contained the highest amount of the PSII dimer, which confirms the crucial function of the dimer in cyanobacterial oxygen evolution. Notably, radioactive labelling of H22KαPS+ revealed that most newly synthesised D1 still remained unassembled. Therefore, compared with the WT, the autotrophic transformants still showed deficient synthesis of D2, the assembly partner of D1.

The results also confirmed that proper coordination of the heme cofactor in Cyt b 559 is important to the stability of Cyt b 559 apoproteins and therefore for the assembly of PSII in Synechocystis (Pakrasi et al., 1991; Hung et al., 2007, 2010). Surprisingly, in the thermophilic cyanobacterium Thermosynechococcus elongatus the constructed H23Aα and H23Mα Cyt b 559 mutants accumulated the PsbE–PsbF pair in the absence of heme and the accumulation of active PSII was comparable with that in WT cells (Sugiura et al., 2015). The results suggested that, by contrast with Synechocystis, in T. elongatus, the heme does not affect the stability of Cyt b 559 apoproteins. Alternatively, it cannot be excluded that the studied T. elongatus mutants were in fact ‘pseudorevertants’ similar to those described in our study, which originated during the process of transformation and autotrophic selection.

Taken together, we revealed a unique mechanism used by cyanobacteria to cope with mutation‐induced instability of photosynthetic protein subunits. The mutations in the heme axial ligands of Cyt b 559 destabilised the PsbE/F apoproteins (possibly due to their altered conformation) and they are quickly degraded. Since the synthesis of D2 is dependent on the presence of the PsbE/F pair, the essential D2 module cannot be formed and PSII assembly is inhibited preventing autotrophic growth of the Cyt b 559 mutants. The autotrophic transformants compensated for mutation‐destabilised Cyt b 559 subunits via multiple tandem amplifications of chromosomal segments containing the mutated psbEFLJ operon. This led to an increase in the transcript level of the mutated Cyt b 559 gene. The resulting overproduction of the mutation‐destabilised Cyt b 559 subunits allows the D2 synthesis and enables PSII accumulation and photoautotrophic growth of the strains.

Author contributions

Y‐FC, JK and H‐AC designed the study. Y‐FC constructed the Type B strains used in this study. Y‐FC, H‐YF, PS, K‐ML and JK performed the research. Y‐FC, H‐YF, PS, K‐ML, JK and H‐AC analysed the data. Y‐FC, J‐MK and H‐AC wrote the article. J‐MK and H‐AC acquired the funding and supervised the project. All authors discussed the results and commented on the article.

Fig. S1 Photoautotrophic growth curves of wild‐type and autotrophic transformant cells.

Fig. S2 Mapping of Illumina sequencing reads of autotrophic transformants to the reference sequences of the pAC559Em plasmid.

Fig. S3 Two‐dimensional PAGE analysis of photosynthetic complexes in the site‐directed mutant H22Yβ and its corresponding autotrophic transformant H22YβPS+ of Synechocystis.

Table S1 Primers used for droplet digital PCR analysis.

Table S2 Sequences of the gene‐specific primers used in real‐time‐PCR assays.

Table S3 Verification of copy number variation of cpcC1C2 in the wild‐type and autotrophic transformants by droplet digital PCR.

Table S4 The reads/coverage of mutation sites of H22α or H22β in autotrophic transformants obtained by whole‐genome sequencing.

Click here for additional data file.

Table S5 Common differentially expressed genes in autotrophic transformant cells.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Click here for additional data file.