Effects of polyploidy on the coordination of gene expression between organellar and nuclear genomes in Leucanthemum Mill. (Compositae, Anthemideae)

Abstract Whole‐genome duplications (WGD) through polyploid speciation are associated with disruptions of well‐tuned relationships among the three plant cell genomes. Key metabolic processes comprising multi‐subunit enzyme complexes, for which partner proteins are both nuclear‐ and plastid‐encoded, are dependent on maintenance of stoichiometric ratios among the subunits to avoid cytonuclear imbalances after WGDs. By using qPCR for gene copy and transcript number quantification, we have studied the relationship of subunit expression in the two gene pairs rbcL/rbcS (the two subunits of RuBisCO) and psbA/psbO (two members of photosystem II) in closely related members of Leucanthemum (Compositae, Anthemideae), comprising a diploid, a tetraploid, and a hexaploid species. While gene copy numbers exhibit the expected pattern of an increase in the nuclear‐encoded partner gene relative to the plastid‐encoded one, we find that the two partner gene systems behave differently after WGD: While in the psbA/psbO partner gene system, shifts in the gene copy balance caused by polyploidization are not accommodated for through changes in transcription intensities of the two partner genes, the rbcL/rbcS system even shows an unexpected reversed dosage effect with up‐regulated transcription intensities on both the nuclear and the plastidal side. We interpret the behavior of the psbA/psbO partner gene system as being due to the stoichiometrically relaxed relationship between the two gene products caused by a fast, damage‐provoked combustion of the psbA gene product (the D1 core protein of PSII). Conversely, the finely tuned expression dependencies of the rbcL/rbcS system may be the reason for the observed positive feedback runaway signal as reaction to gene copy imbalances caused by a polyploidization shock.


| INTRODUC TI ON
Considerable progress has been made in the attempt to identify effects of polyploidy on photosynthesis-related anatomical, morphological, physiological, and biochemical traits in plants Warner & Edwards, 1993). However, most of the research has been focused on autopolyploids, and synthetic or young allopolyploids . Information obtained from natural and old allopolyploids is still rather sparse, partly due to difficulties in disentangling the contribution of genome duplication, hybridization, and subsequent natural selection (Hegarty et al., 2013).
During allopolyploidization, doubling of the nuclear genome occurs, yet only one set of organelles is present in the cell. Since genes involved in photosynthesis are often organized in gene networks comprising both genes located in the nucleus and genes located in the chloroplast genome, coordinated expression to produce the correct amounts of nuclear and chloroplast proteins and to maintain a proper function is required. Such coordination can occur at the level of transcription, RNA stability/degradation, translation, import, protein turnover, or metabolite flux being controlled by plastid and nuclear factors (Pogson & Albrecht, 2011). Polyploidization may thus represent a challenge with regard to maintaining the proper stoichiometry among the products of interrelated genes coming from different cellular compartments (Sharbrough, Conover, Tate, Wendel, & Sloan, 2017). This question was experimentally addressed in  and Coate, Schlueter, Whaley, and Doyle (2011), who studied the evolutionary impact of polyploid and nonpolyploid duplications on genes encoding the major functional groups in photosynthesis (i.e., photosystem I, photosystem II, the light-harvesting complex, and the Calvin cycle) in soybean (Glycine max), barrel medic (Medicago truncatula), and Arabidopis thaliana. In their work, the authors could observe distinct evolutionary patterns and duplicate retention in gene network evolution after duplication events: photosystem I and photosystem II gene families seem to be governed largely by dosage sensitivity, and Calvin cycle gene families seem to be less dosage sensitive and thus have larger capacities for functional divergence.
In the present study, we would like to address the question whether and how polyploidy affects the functionally tightly related gene pairs from photosystem II (PSII) and Calvin cycle (CC) in more detail. The first gene pair surveyed comprises genes encoding the large subunit (LSU) and the small subunit (SSU) of the RuBisCO (ribulose-1,6-bisphosphate carboxylase/oxygenase) enzyme (rbcL and rbcS, respectively). For plants, production of the right amount of RuBisCO is essential to correctly carry out photosynthesis. It is a multimeric enzyme located in the stroma of the chloroplasts (Coen, Bedbrook, Bogorad, & Rich, 1977) consisting of eight SSUs and eight LSUs. While the SSU polypeptide is encoded on nuclear genes and translated on cytoplasmic ribosomes into precursor peptides that are post-translationally imported and processed in the chloroplast (Highfield & Ellis, 1978), the LSU is encoded by a single rbcL gene located on the plastidic genome. The SSU is encoded by the nucleus-located rbcS gene family containing 4-10 or more genes in higher plants (Caffarri, Kouřil, Kereïche, Boekema, & Croce, 2009).
The second gene pair studied in the present contribution comprises the psbA and psbO genes from photosystem II (PSII), which is a large multi-subunit pigment-protein complex localized in the thylakoid membrane of plants, algae, and cyanobacteria. PSII splits water into oxygen, protons, and electrons during the photosynthetic process and thus provides energy and oxygen (Barber, 2003). The core protein D1 of the reaction center is encoded by the psbA gene which in higher plants usually exists as a single-copy gene located in the plastome (Singh, 2000). The psbO gene encodes for the manganesestabilizing protein, which is in higher plants located on the lumenal side of the membrane and together with other proteins compose the oxygen-evolving complex participating in the stabilization of the Mn-cluster required for an efficient oxygen evolution (Popelkova & Yocum, 2011). It plays a role in the stabilization and turnover of the D1 during the PSII damage-repair cycle (Komenda, Knoppová, Krynická, Nixon, & Tichý, 2010;Yamamoto et al., 1998). In higher plants, psbO is encoded by a small nuclear gene family (Pérez-Bueno, Barón, & García-Luque, 2011), with one to four gene members (according to the KEGG Pathway database). A precursor protein of psbO is synthesized in the cytoplasm and imported into the chloroplasts (Keegstra, Olsen, & Theg, 1989;Popelkova & Yocum, 2011).
Our present contribution is based on representatives from the genus Leucanthemum (Compositae, Anthemideae), which is a polyploid complex from southern and central Europe, comprising 41 species with ploidy levels ranging from 2x to 22x (Greiner, Vogt, & Oberprieler, 2013). In the work of Greiner, Vogt, and Oberprieler (2012), Greiner et al. (2013) and Oberprieler, Greiner, Konowalik, and Vogt (2014), a closely related and geographically isolated species group from the northwest Iberian Peninsula was identified. This group was denoted as the Leucanthemum pluriflorum clan, and it comprises the diploid L. pluriflorum Pau, the tetraploid  on the other hand. Under both scenarios, the ratio between chloroplast-encoded gene copies and nuclear-encoded partner gene copies (a 1 and b 1 in Figure 1) is expected to decrease with increasing ploidy level if not the number of the latter is balanced by increased numbers of chloroplasts per cell and/or plastid genomes per chloroplast. Conversely, calculation of ratios of transcript copy numbers from chloroplast-and nucleus-encoded partner genes (a 3 and b 3 in Figure 1) and of grand indices (a 4 and b 4 in Figure 1) measuring the discrepancy between gene copy ratios (a 1 and b 1 , respectively) and transcript copy ratios (a 3 and b 3 , respectively) would make both scenarios distinguishable.

| Plant material
Seeds of two geographically distinct populations for each of the three species under study (i.e., L. pluriflorum [2x], L. pseudosylvaticum [4x], and L. sylvaticum [6x]; see Table S1 for further information on the location of accessions), were grown in the greenhouse of the Institute of Plant Sciences of Regensburg University (Regensburg, Germany) under the constant light and temperature conditions. Two fully developed young rosette leaves from each plant were collected at approximately the same time (11 a.m. to 1 p.m.) during the same day (12.09.2012) and at the same developmental stage (rosette stadium), immediately put into liquid nitrogen (one leaf per Eppendorf tube) and stored at −80°C. Ten accessions for each species (five individuals for each of the two populations) were selected for further analyses.

| DNA and RNA extraction and cDNA synthesis
Nucleic acids (RNAs and gDNA) were isolated from a single rosette leaf of the selected individuals. The extraction of total RNA was performed using a TriFast reagent (PEQLAB Biotechnologie) according to the TRIzol extraction protocol (Invitrogen). To obtain genomic DNA, DNA extraction from TRIzol using a back extraction buffer (4 M guanidine thiocyanate, 50 mM sodium citrate, 1 M Tris) was performed according to the TRIzol manufacturer's instructions (Invitrogen). Integrity of both RNA and gDNA was checked on 1.5% agarose gels. The concentration and purity of the nucleic acids were measured spectrophotometrically using the Nanodrop ND-1000 system (Thermo Fisher Scientific). To eliminate the interference of genomic DNA in the cDNA synthesis step, a DNaseI (Fermentas) treatment of RNA was performed according to the manufacturer's recommendations followed by precipitation using 2.5 volumes of ice-cold 96% EtOH and 0.1 volume of 3 M sodium acetate at −70°C for at least 1 hr. Conversely, the DNA extracts were treated with 1 μg/μl RNase A (Fermentas) at 37°C for 30 min, followed by ethanol precipitation. After these treatments, the quality of both, RNA and gDNA, was checked on 1.5% agarose gels and the final concentrations and purities were again measured spectrophotometrically.
F I G U R E 1 A comparison of hypothetical outcomes for gene copy (colored boxes) and transcript numbers (colored lines) of the four candidate genes under study (the two chloroplast-encoded genes psbA and rbcL and the two nucleus-encoded partner genes psbO and rbcS) normalized against actin copy numbers under the assumption of 1:1 dosage effect (left, a) and dosage compensation (right, b) after tetra-and hexaploidization in the Leucanthemum pluriflorum group. Under both scenarios, the ratio between chloroplast-encoded gene copies (Gene cp ) and nucleus-encoded partner gene copies (Gene nuc ) is expected to decrease with increasing ploidy level (a 1 , b 1 ), if the doubled and tripled gene copy numbers on the nucleus side are not balanced by increased numbers of chloroplasts or plastid genomes per chloroplast. (a) Under a 1:1 dosage effect with no differences of transcription intensity (mRNA/Gene) in both genomes concerned (a 2 ), the ratio of transcript copy numbers from chloroplast-and nucleus-encoded partner genes a 3 (mRNA cp /mRNA nuc ) should follow the ratio between gene copy numbers (a 1 ), and the grand index a 4 measuring the discrepancy between gene copy ratio (a 1 ) on the one hand and transcript copy ratio (a 3 ) on the other should level out across ploidy levels. (b) Under the assumption of dosage compensation-either by an increase in transcription intensity with ploidy on the chloroplast-gene side or by a decrease on the nucleus-gene side (b 2 )-the ratio of transcript copy numbers from chloroplast-and nucleus-encoded partner genes b 3 (mRNA cp /mRNA nuc ) should be stoichiometric, and the grand index b 4 measuring the discrepancy between gene copy ratio (b 1 ) on the one hand and transcript copy ratio (b 3 ) on the other should increase with increasing ploidy 1 : 1 dosage effect Dosage compensation The amount of 1 μg of DNaseI-treated mRNA was used for the first-strand cDNA synthesis using the i-Script kit (Bio-Rad Laboratories) according to the manufacturer's instructions. The efficiency of the reverse transcription was proved by PCR with primers specific for actin (see below for primer sequences). mRNA extracts without reverse transcriptase treatment (RT-) were used as negative controls to check for any gDNA contamination.

| Characterization of Leucanthemum-specific PCR primers
Since the exact nucleotide sequences of the psbA, psbO, rbcL, and rbcS genes were not known for the Leucanthemum species under study, nucleotide sequence resources from GenBank (NCBI) had to be used to design specific primers: For the chloroplast genes psbA and rbcL, corresponding nucleotide sequences (from some Asterid representatives for psbA and from some Anthemideae representatives for rbcL) obtained from GenBank (NBCI) were aligned. Due to a high degree of conservation, specific primers could be designed in the regions of highest sequence similarity. For the nuclear rbcS gene, a sequence of the closely related Chrysanthemum ×morifolium was available. The C. ×morifolium rbcS sequence was aligned with available sequences from other Asterids (Solanaceae representatives and Helianthus).

Nondegenerated primers were then designed according to the
Chrysanthemum rbcS sequence matching the most conserved regions among the aligned sequences. For the second nucleus-encoded psbO gene, degenerated primers were designed based on the alignment of sequences available in GenBank. Gene-specific primers were designed using Primer3 (Rozen & Skaletsky, 1999), and degenerated primers were designed manually. All primers are listed in Table S2. Plasmid DNA was isolated with either the NucleoSpin Plasmid miniprep kit (Macherey-Nagel) or the GeneJET Plasmid kit (Thermo Fisher Scientific). Inserts were then cycle-sequenced with the DTCS kit (Beckman Coulter) using either sequence-specific primers or cloning vector-matching primers jetF and jetR available in the CloneJET PCR Cloning kit.

| Quantitative real-time PCR
To enable an absolute quantification of PCR product amounts and to test the efficiency of the PCRs, a calibration vector was constructed by ligating one copy of each gene fragment (psba, psbO, rbcL, rbcS, actin) into a pBluescript vector (Stratagene

| Flow cytometry
For the determination of DNA content and ploidy level, we used a two-step flow cytometry protocol (Doležel, Greilhuber, & Suda, 2007) with Petunia hybrida (PxPc6) or diploid L. pluriflorum as an internal standard. The amount of leaf probe material (about 30 mm 2 ) was approximately twice that of the internal standard material.
Fragments of leaves were chopped with a razor blade in Otto I buffer (Otto, Oldiges, Göhde, & Jain, 1981). The suspension of nuclei was filtered through a mesh with a pore size of 50 μm and kept on ice, followed by the centrifugation for 5 min at 150 g in 4°C. The isolation buffer was removed leaving c. 50 μl above the pellet which was then dissolved in ice-cold LB01 buffer (Doležel, Binarová, & Lcretti, 1989) with 4 mg/L of DAPI. Excitation of the sample was done using UV laser 365 nm (16 mW) with an accompanying band-pass filter 455/50 nm on a CyFlow Space cytometer (Partec). Acquisition was automatically stopped at 8,000 nuclei. The relative genome size was calculated by multiplying the known genome size of internal standards (P. hybrida or L. pluriflorum) by the quotient between the 2C peak positions of the target probe and the internal standard in the histogram of fluorescence. The ploidy level was inferred from the relative genome sizes based on previous measurements made on plants with known chromosome numbers.

| Data analysis
To test for a 1:1 dosage effect scenario versus a dosage compensation scenario for the studied gene pairs (Figure 1), statistical analyses were performed with the IBM SPSS Package, version 20.0 (IBM Corporation). When necessary, a transformation of the data set was performed in order to receive normally distributed variables.
Analysis of variance was computed with either parametric ANOVA and post hoc tests or nonparametric Kruskal-Wallis test with subsequent Bonferroni-corrected Mann-Whitney U tests.

| RE SULTS
To

| Chloroplast number per cell
We estimated the mean number of chloroplasts in guard cells and cells of the palisade and spongy mesophyll parenchyma, respectively, throughout the three different ploidy levels (

| Transcript copy numbers per cell
When compared to the expression level of the actin locus/loci, all photosynthesis-related loci-irrespective of their location on the nuclear or the plastid genome-showed higher expression rates, ranging in average from around 10-fold in psbO over 50-to 600-fold in rbcL and rbcS to 1,600-to 4,800-fold rates in psbA (Figure 2b; Table S4).
When referenced with the expression levels of actin, no significant differences in expression levels of our candidate photosynthesis gene pairs were observed across ploidy levels (psbA: h = 1.30, df = 2, p = .523, 2x ≤ 4x ≥ 6x; rbcL: h = 3.63, df = 2, p = .163, 2x ≤ 4x ≥ 6x; psbO: h = 5.35, df = 2, p = .069, 2x ≤ 4x ≥ 6x; Table S5). An exception was the nucleus-encoded rbcS gene, in which the trend of tetraploids showing higher expression rates than diploids and hexaploids gained significance (F = 11.98, df = 2, p < .001, 2x < 4x > 6x). Interestingly, a similar trend was also seen in the other genes, however, with insignificant statistical support (see above). When referenced to their nucleus-encoded partner genes, the expression of the plastid-encoded rbcL gene leveled off with its nuclear counterpart rbcS, while psbA (plastid) showed a 163-to 666-fold higher expression than psbO (nucleus). No significant differences in these plastid versus nucleus expression ratios were found among the ploidy levels in the psbA/psbO gene pair (h = 2.30, df = 2, p = .317, 2x ≤ 4x ≥ 6x). In the rbcL/rbcS system, however, we encountered a significant higher ratio toward the plastid partner gene in diploids with only insignificant trends in tetra-and hexaploids (F = 8.72, df = 2, p = .001, 2x > 4x ≤ 6x). In the rbcL/rbcS system, therefore, the increased expression level of the nuclear gene partner (rbcS) caused by the higher number of gene copies through polyploidization is not fully compensated by higher expression levels (or higher copy numbers) of the plastid gene partner (rbcL).

| Expression intensity relative to genomic copy numbers
Expression intensity measured as the ratio of actin-normalized transcript number (cDNA) to actin-normalized genomic gene copy number (gDNA) followed a trend with higher values for the nucleusencoded genes rbcS and psbO and lower ones for their plastid-encoded gene partners rbcL and psbA (Figure 2c; Table S5). In all genes surveyed, there was a trend observable toward a higher expression intensity in tetraploids than in diploids and hexaploids; however, this trend became only significant in the nuclear gene rbcS (F = 13.02, df = 2, p < .001, 2x < 4x > 6x) and only unilaterally so in its plastidic counterpart rbcL (F = 4.13, df = 2, p = .028, 2x < 4x ≥ 6x). Therefore, the trend of higher expression levels of all four genes in tetraploids

| Expression intensity relative to ploidy level
With the calculation of the "grand indices" I (cDNA psbA/psbO)/(gDNA psbA/ psbO) and I (cDNA rbcL/rbcS)/(gDNA rbcL/rbcS) for each ploidy level, we tested whether the ploidy-dependent imbalances (plastidic to nuclear copy number ratios) observed on the gDNA level (i.e., 2x > 4x > 6x in both gene pairs) is paralleled by corresponding trends on the cDNA level (i.e., 6x < 2x < 4x in psbA/psbO; 4x < 6x < 2x in rbcL/rbcS) across ploidy levels (Figure 2d; Table S4). For the psbA/psbO gene pair, we found no significant differences in this "grand index" among ploidy levels (h = 0.06, df = 2, p = .972; 2x ≤ 4x ≥ 6x), indicating that observed imbalances on the mRNA level are not significantly different from copy number differences observed on the genomic level. As a consequence, no changes in the transcriptional activities of the nuclear and plastid genes involved need to be assumed to explain the mRNA imbalances. The psbA/psbO system, therefore, corresponds to the hypothetical pattern expected from a 1:1 dosage effect ( Figure 1, a 4 ).
A different situation was observed in the rbcL/rbcS system, where the discrepancy between imbalances on the gDNA and those on the cDNA level measured by the "grand index" gained significance between the diploid and tetraploid level (h = 6.30, df = 2, p = .043; 2x > 4x ≤ 6x). This indicates that the polyploidy-induced shift in genomic copy number relationships (with a stronger increase F I G U R E 2 Boxplot diagrams for diploid (red), tetraploid (green), and hexaploid (blue) representatives of the Leucanthemum pluriflorum group studied for gene copy and transcript numbers of the two chloroplast-encoded (cpDNA) genes psbA and rbcL and its nucleus-encoded (nDNA) partner genes psbO and rbcS. All diagrams show a logarithmic scale on the y-axis; letters above boxplots indicate statistically significant differences among the three ploidy levels (see also Table S5). (a) Results for the qPCR quantification of genomic (gDNA) copy numbers of the four genes under study; either referenced with nucleus-encoded actin housekeeping gene copies (left four columns) or partner gene-wise copy number ratio (right two columns). (b) Results for the qPCR quantification of transcript (cDNA) copy numbers for the four genes under study; either referenced with transcript numbers of the nucleus-encoded actin housekeeping gene (left four columns) or partner gene-wise transcript number ratio (right two columns). (c) Results for the qPCR-based quantification of expression intensity calculated as the ration of transcript copy numbers (b) per gene copy numbers (a) for the four genes under study. (d) Results for the comparisons between ploidy-dependent cpDNA-nDNA gene copy number imbalance on the one and cpDNA-nDNA transcript copy number imbalance on the other hand in the two partner gene systems psbA/psbO (left) and rbcL/rbcS (right)  (Table 1).
This result is best interpreted by assuming that the number of chloroplast per cell (and the often observed-albeit here not measured-correlated cell size) is not influenced by nucleotypic effects (effects derived from the amount of DNA per cell independent of that DNA's information content; Bennett, 1987) alone, but also by genetic effects (effects driven by changes in the information content of the genome and epigenome; Coate et al., 2012). This interpretation is all the more reasonable since we still lack a satisfying hypothesis on the genealogical relationships among the three Leucanthemum species involved in the present study. While sharing of the same chloroplast haplotypes across the three ploidy levels indicates that L. pluriflorum may have been the maternal partner in the formation of the polyploids (Greiner et al., 2013), it remained unclear whether the polyploids were formed through auto-or allopolyploidy and (in the latter case) which diploid could have acted as the paternal diploid (Oberprieler et al., 2014).  (Vogt, 1991), these differences could also have an ecologically adaptive connotation. For the following analyses and interpretations, however, we feel justified to disregard these aspects, because all plants analyzed were cultivated under the same conditions and all quantifications of (gene and transcript) copy numbers were controlled for by internal standardizations as suggested by Coate and Doyle (2010), who also had to deal with natural diploids and polyploids with a long-lasting history of independent evolution in Glycine.
Copy numbers of nuclear partner genes (i.e., psbO, rbcS) were found showing no significant increase or decrease across ploidy levels when compared to the equally nucleus-encoded (and therefore equally duplicated) household gene actin (Figure 2a; Table S3). This indicates that duplication and triplication of the original (diploid) suite of copy numbers of these genes through polyploidization have not been changed dramatically after the polyploidization events.
This is contrary to observations by De Smet et al. (2013) who investigated the existence of "duplication-resistant" genes in the genomes of 20 flowering plants and found that "there is a large set of genes that is convergently restored to single-copy status following multiple genome-wide and smaller scale duplication events" (De Smet et al., 2013Smet et al., : 2898. However, in contrast to our present study group of Leucanthemum, for which WGD events are assumed to have occurred during the Pliocene or Pleistocene (neopolyploidization), those studied by De Smet et al. (2013) rather date back dozens to hundreds of million years (meso-to paleopolyploidization). Hence, stoichiometric consequences of polyploidy for cytonuclear interactions might become counterbalanced through selection for gene loss of nuclear gene families; but this seems to happen on other timescales than the one studied here in the L. pluriflorum group and may be connected to more drastic genome rearrangements observed in meso-and paleopolyploids (diploidization).
A further potential mechanism to maintain cytonuclear stoichiometry after WGD could be the increase in genome copies per organelle (here: chloroplast) to balance the increase in copy numbers of the nuclear partner genes (Sharbrough et al., 2017). If this was realized in the present study group, one would expect no significant differences in plastid-to-nucleus copy number ratios among ploidy levels. However, we observed trends in both partner gene systems that the increase in nuclear gene copies through WGD is not counterbalanced by a proportional increase in plastidic gene copies (Table   S3). Together with the above-mentioned results on chloroplast numbers per cell in our study group, this indicates that, if regulation for maintenance of cytonuclear stoichiometry after WGD takes place, this is not accommodated (exclusively) on the genomic but rather on the transcriptional level.  Table S5).
In respect to the observed absolute transcript number differences between the nuclear and the plastidic partner genes, our findings are somewhat in line with a study of Recuenco-Muñoz et al. (2015) who analyzed the RuBisCO system in the green algae Chlamydomonas reinhardtii. It is astonishing to learn that even in this ecologically and physiologically extremely important holoenzyme, the actual quantities of the two subunit proteins "have never been determined in any photosynthetic organism" (Recuenco-Muñoz et al., 2015) until the mentioned study. Here, the authors demonstrated that rbcL:rbcS transcript and protein ratios deviated from the expected 1:1 stoichiometry and were shifted to 5:1 ratios, indicating that the small subunit (the rbcS protein) is the limiting factor for the assembly of the holoenzyme (Recuenco-Muñoz et al., 2015). A possible reason for this shift away from the expected 1:1 ratio could be an increased transcript degradation of rbcL upon light as observed by Salvador, Klein, and Bogorad (1993), which would make necessary the overexpression of the plasidic partner gene to result in an equimolar ratio of gene products.
In the RuBisCO system of Leucanthemum, we observed a (compared with Chlamydomonas) less pronounced 2:1 transcript ratio in the diploids and an even more balanced 1:1 transcript ratio in the two polyploid taxa. This is in stark contrast to the 160-670:1 ratios measured in the psbA/psbO system of diploids and polyploids.
As a consequence, compared with the RuBisCO system, the role of the nuclear-encoded psbO protein as limiting factor for the assembly of the multi-subunit pigment-protein complex of photosystem II (PSII), in which both proteins co-operate, appears considerably more dramatic. However, because the design of PSII allows protection for most of its protein components and damage mainly occurs at the re- in the molecular laboratory of the working group at the University of Regensburg is also gratefully acknowledged. Additionally, we would like to express our sincerest thanks to two anonymous reviewers, whose comments improved the present contribution considerably.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interests.

AUTH O R CO NTR I B UTI O N S
CO and JG designed the present study; MT and JG performed the qPCR experiments and analyzed the data; CO and MT wrote the manuscript, which was improved with regard to the description of qPCR methods and in linguistic respects by JG.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data of the present study are made available through the Tables S1-S5.