Introduction

Plant cells contain three genomes being located in the nucleus, the mitochondria, and the plastids. The organellar genomes are remnants from the endosymbiotic past that gave rise to mitochondria and plastids, which have retained only a small proportion of the original coding capacity of the respective endosymbiont. In fact, the vast majority of the original genes from the endosymbiont were transferred to the nucleus by horizontal gene transfer (Timmis et al. 2004). Nevertheless, both plastids and mitochondria kept a conserved set of genes. In vascular plants, the plastid genome is roughly 150 kb in size and encodes around 120 genes mainly for components of the photosynthetic apparatus, tRNA, rRNAs, and for protein subunits of the plastid-own translation and transcription machinery (Sugiura 1992; Bendich 2013). As an evolutionary consequence of the gene transfer from the endosymbiont to the host, the plastid transcription machinery became heavily re-structured resulting in a complex system of phage-, prokaryotic-, and eukaryotic-type components (Börner et al. 2015; Pfannschmidt et al. 2015). Plastids of vascular plants contain a nuclear-encoded, phage-type single-subunit RNA polymerase called NEP (for nuclear-encoded polymerase) and a plastid-encoded, prokaryotic-type multi-subunit RNA polymerase called PEP (for plastid-encoded polymerase). While NEP transcribes mainly plastid-encoded housekeeping genes (referred to as class-III genes), PEP transcribes mostly photosynthesis and tRNA genes, which were designated as class-I genes. Class-II genes are transcribed by both NEP and PEP (Hajdukiewicz et al. 1997; Liere et al. 2011).

The PEP enzyme is encoded by four plastid-encoded genes; rpoA, rpoB, rpoC1, and rpoC2. The latter three form a conserved operon, while rpoA is part of a mixed operon together with rsp11 and petD (Igloi et al. 1990). These rpo genes exhibit many homologies to the orthologous rpo genes in bacteria and encode a core enzyme that resembles that of E. coli. For promoter recognition, this core enzyme interacts transiently with sigma factors that are encoded in the nucleus (Lerbs-Mache 2011). In Arabidopsis, six different sigma factors are known from which sigma 2 and sigma 5 are the most important. This PEP-core enzyme has been biochemically purified only from etioplasts (see below for definition) or intermediate greening chloroplasts (Hu and Bogorad 1990; Pfannschmidt and Link 1994). In fully developed chloroplasts, the PEP-core enzyme appears to be coated with additional subunits of eukaryotic origin. Biochemical purification of chloroplast PEP enzymes followed by mass spectrometry identified those additional subunits (Pfannschmidt et al. 2000; Suzuki et al. 2004; Pfalz et al. 2006; Steiner et al. 2011; Melonek et al. 2012) and finally led to the identification of 12 PEP-associated proteins (PAPs) that exhibit strong and reproducible interaction with the core complex (Pfalz and Pfannschmidt 2013). These observations strongly suggest that the plastid PEP enzyme is structurally re-organized during the initial steps of photomorphogenesis and chloroplast development. Until now, it is not known whether this is cause or consequence of chloroplast development.

In angiosperms, seeds germinating in the dark follow a developmental programme called skotomorphogenesis. It comprises the formation of an apical hook, hypocotyl elongation, a strong inhibition of leaf development, and a yellow pigmentation. Plastids develop from the proplastid stage in the embryonic cells of the seed into a special dark-related form called etioplast. These are plastids without thylakoid membranes or chlorophylls. Instead, they contain a para-crystalline prolamellar body comprised of NADPH, protochlorophyllide, the enzyme NADPH:protochlorophyllide:oxido-reductase (POR), and the thylakoid membrane lipids digalactosyldiacylglycerol (DGDG) and monogalactosyldiacylglycerol (MGDG) (Solymosi and Schoefs 2010; Pribil et al. 2014; Bastien et al. 2016; Liebers et al. 2017). Etioplasts represent a developmentally arrested stage, as its further development is strictly dependent on light. Upon illumination of the seedling the photomorphogenetic programme is initiated that rapidly results in chloroplast biogenesis. In this process, etioplasts turn into chloroplast within a few hours (Pogson et al. 2015) including the formation of thylakoid membranes, chlorophylls, and the massive expression of photosynthesis proteins.

In parallel, the PEP-core enzyme becomes decorated with the mentioned 12 PAPs. These proteins are highly diverse in their function and structure. According to their predicted functional domains PAPs can be divided into three major groups. Group 1 comprises PAPs involved in DNA/RNA metabolism (PAP1, 2, 3, 5, and 7), group 2 includes proteins involved in redox regulation (PAP6 and 10) and ROS protection (PAP4 and 9) and group 3 includes proteins with unknown function (PAP8, 11, and 12) (Steiner et al. 2011; Kindgren and Strand 2015; Pfannschmidt et al. 2015). The previous analyses using reverse genetics demonstrated that genetic inactivation of any of the PAP genes, regardless of the potential protein function, resulted in an albino phenotype strongly suggesting that the re-organisation of the PEP complex represents a bottleneck for chloroplast biogenesis (Pfalz et al. 2006; Myouga et al. 2008; Arsova et al. 2010; Schröter et al. 2010; Steiner et al. 2011; Gao et al. 2011; Gilkerson et al. 2012). An appealing interpretation for this observation is that each PAP is essential for the proper assembly and/or stability of the PEP complex, so that the absence of only one of them impedes the function of the whole. As a consequence, this inactivation of the PEP complex blocks chloroplast biogenesis. Therefore, as a corollary, we proposed that PAP genes must be coordinately expressed during the early steps of chloroplast biogenesis. Here, we report a detailed spatio-temporal expression analysis of PAP genes throughout the life cycle of Arabidopsis. Using promoter–reporter constructs, we could identify a canonical expression pattern with tissue-specific features, providing a comprehensive expression catalogue of PAP genes. Our data reveal unexpected cell-type-specific expression dynamics especially during seedling development and indicate that the expression of PAP genes represents a trustful marker for chloroplast biogenesis.

Materials and methods

Plant material and growth conditions

The wild-type Arabidopsis thaliana, ecotype Columbia-0 (Col-0, accession SALK_6000) was used for stable transformations using electro-competent Agrobacterium tumefaciens strain C58C1 pMP90. For selection of promoter lines, developmental series and histological cuts plants were grown on ½ strength MS media, 1% sucrose, and 0.8% agar. Seeds were imbibed and stratified for 2 days at 4 °C, before growth at 21 °C for 3 days in darkness. Afterwards, plants were transferred to continuous white light (~ 30 µmol m−2 s−1) for the time indicated in the according experiment. For the GUS assay of rosette plants, flowers, and embryos, seeds were sown on soil and grown at 21 °C, ~ 70 µmol m−2 s−1, and 60% humidity in long days (8 h dark/16 h light).

Cloning of the promoter regions

The promoter regions were amplified from Col-0 gDNA, using primers described in Table 1.

Table 1 List of primer sequences (5′–3′) used for amplification and subsequent cloning of promoter regions
Full size table

The PCR was carried out with the Phusion® High-Fidelity DNA Polymerase (New England Biolabs), followed by A-tailing with Taq-Polymerase for 5 min at 72 °C (Promega). PCR products were purified according to the manufacturer’s instructions (Geneclean® III kit), followed by cloning into the pGemT-vector system (Promega) then into a binary vector, deriving from pArt27. All promoters include the 5′-UTR and were fused to the open reading frame of the β-glucuronidase. The used vector contained a resistance against spectinomycin for the selection in bacteria and a resistance against hygromycin for the selection in plants.

Transient transformation of onion cells

Gold carrier particles (Seashell technology) were coated with 1 µg of the binary vector and 1 µg of an internal control. Gold particles were delivered into onion cells using a particle gun (Bio-Rad). The transformed cells were allowed to express the construct for 24 h before GUS staining.

Generation of homozygous pPAP::GUS lines

T1 seeds (primary transformants) were harvested and grown on ½ MS media, containing 1% sucrose, 0.8% agar, and 30 µg ml−1 hygromycin. Seeds were imbibed and stratified for 2 days at 4 °C, before growth at 21 °C in far-red enriched light (Wagner et al. 2008) for 7 days, followed by a transition for 3 weeks to WL (30 µmol m−2 s−1). Resistant plants were transferred on soil and grew at 21 °C, ~ 70 µmol m−2 s−1 and 60% humidity under short-day conditions (8 h dark/16 h light). After 3 weeks of growth on soil, hygromycin-resistant plants were genotyped and one leaf of the rosette plant was used for the GUS assay. T2 seeds were then harvested and grown on hygromycin for the segregation record. Lines that displayed ~ 75% hygromycin resistance (single locus insertion) were transferred on soil and genotyped. Furthermore, plants of these lines were applied for GUS assay (Fig. 2d). T3 seeds were then selected and checked for a 100% hygromycin resistance. These lines were homozygous for the insertion and thus used for the expression analyses.

Genotyping of the promoter::GUS constructs

Genotyping was carried out using the forward primer that was used for the cloning and the reverse primer oGUS-J. Due to the size of the promoter region of PAP5, additional primers within the amplicon were used to assure a strong amplification of the product. The promoter region of PAP9 was only partially amplified (Table 2).

Table 2 List of primer sequences (5′–3′) used for genotyping of transformants
Full size table

RNA isolation and reverse transcription PCR

For RNA isolation, 1 g of Col-0 seeds were imbibed and stratified on ½ MS media. Plants were grown as described above. Samples were collected at the different time-points, ground in liquid nitrogen, and resuspended in 500 µl of trizol™. After ~ 2 min incubation, 100 µl chloroform were added, and the mixture was vigorously mixed, incubated for 2 min, and centrifuged for 15 min at 4 °C maximum speed (16,000g). The aqueous phase was mixed with equal volume of isopropanol and RNA was precipitated for 10 min at room temperature, before centrifugation for 15 min at 4 °C maximum speed (16,000g). The sediment was rinsed with 80% ethanol, vacuum dried, and resuspended in 100 µl of water. The RNA was treated with the Roche-DNase (Promega) according to the manufacturer’s instructions. The DNase was removed by phenol–chloroform extraction and the pure RNA was reverse transcribed into cDNA using the SuperScript® III Reverse Transcriptase (Invitrogen), following the manufacturer’s instructions. cDNA was the amplified using the primers given in Table 3. The absence of DNA in the samples was tested by PCR of RNA samples without performing reverse transcription. Amplification products were separated on agarose gels; the obtained bands were quantified using ImageJ. Relative transcript levels are PAP-expression values divided by those of EF1α.

Table 3 List of primer sequences (5′–3′) used for determination of PAP gene expression by RT-PCR
Full size table

GUS staining

Samples were treated with acetone for 15 min. After a wash with rinse buffer (25 mM phosphate buffer pH 7, 0.25% Triton X 100, and 0.25 mM EDTA) samples were transferred for 10 min into the GUS staining solution (25 mM phosphate buffer, 0.25% Triton X 100, 0.25 mM EDTA, 1.25 mM K4[Fe(CN)6], and 1.25 mM K3[Fe(CN)6]) before vacuum infiltration in the GUS staining solution containing 2 mM X-gluc. The staining was carried out at 37 °C and stopped with 70% EtOH.

Seed staining and clearing

Siliques of different developmental stages were selected, seeds harvested, and GUS stained. The reaction was stopped with 70% EtOH; the seeds were incubated for 1 h with 100% EtOH/acetic acid (1:1, v/v). The pre-treated seeds were transferred on a microscopic slide and incubated overnight with HOYER’s solution (7.5 g arabic gum, 100 g chloral hydrate, 5 ml glycerol, and 60 ml water) before imaging.

Histological investigations

Histological sections were performed according to the published methods (Ambrose et al. 2000) with minor modifications. Samples were harvested and subjected to GUS staining. Afterwards, the samples were dehydrated using an ethanol series (70, 85, 95, 100, and 100%) and stained in 0.1% eosin/100% EtOH for 20 min. The EtOH:histoclear steps were carried out for 20 min. The histoclear:paraplast mix was exchanged six times (2× per day) with 100% molten paraplast before pouring paraplast solution containing the tissues into weighing boats. After cooling and hardening of the paraplast cubes, 7-µm-thickness sections were sliced on a microtome, transferred on glass slides, and dried overnight, before adding 50% glycerol and a cover slip. Subsequently they were analysed under a Nikon AxioScope light microscope and pictures were taken.

Results

PAP genes exhibit comparable transcript accumulation during the transition from skoto- to photomorphogenesis

The previous in silico analyses revealed similar patterns of expression among the PAP genes (Steiner et al. 2011; Pfannschmidt et al. 2015). Here, we intended to compare the steady-state transcript levels of PAP1–PAP10 using RT-PCR (Fig. 1). Due to the assembly of the PAP proteins to the PEP-core complex after illumination of dark-grown seedlings (Pfannschmidt and Link 1994; Pfannschmidt et al. 2000), a peak of PAP gene expression was expected to occur shortly after the onset of light. To test this, stratified seeds were grown in darkness (72 h at 21 °C) to promote skotomorphogenesis, followed by a light treatment (Fig. 1a). The plant material was harvested at the end of the dark phase (etiolated plants), as well as after 15 min, 1 h, 24 h, and 3 days of illumination.

Fig. 1
figure 1

PAP transcript accumulation during the transition from skotomorphogenesis to photomorphogenesis. a Experimental set-up in which plants were imbibed and stratified for 48 h (I + S), before 72 h growth in the dark (D). Afterwards, plants were shifted into continuous white light for the indicated times. Pictures depict phenotypes of representative seedlings at the respective time-points. b RT-PCR of PAP gene transcript accumulation. Total RNA was isolated from Col-0 at the indicated time-points. The HY5 gene was used as control for light induction. The gene for EF1α served as loading control. Top label indicates growth condition of the plants: D dark; numbers indicate hours (h) of growth in the light; g represents genomic DNA. c Quantification of the steady-state levels of the transcripts displayed in b. Mean value of PAP1, 2, 4, 6, 7, 8, and 9 as well as PAP3, 5, and PAP10 are plotted against the values of HY5. Values of the dark-grown samples were set as 1 and used for normalization for values from the light-grown samples. Error bars represent the standard error. x-axis represents the time-points of harvesting

Full size image

All PAP genes exhibited a significant transcript accumulation in the dark prior to illumination (Fig. 1b, c). The subsequent light exposure was found to trigger comparable expression patterns for PAP1, 2, 4, 6, 7, 8, and 9 indicating a potential co-expression of these genes. While low transcript abundance was found in etiolated seedlings, a transient peak, at 15 min of illumination, was followed by a reduction of the transcripts within the first 3 days reaching expression values below the control values at the end of the dark period (Fig. 1b, c). The same tendency, but with no or a less apparent transient peak, was observed for PAP3, 5, and 10 (Fig. 1c). As control, we compared the PAP gene expression pattern to that of the gene ELONGATED HYPOCOTYL 5 (HY5) encoding a basic leucine zipper domain (bZIP) transcription factor known to promote photomorphogenesis. In our system, HY5 showed transcript accumulation in the dark, with a transient increase from 15 min to 1 h of light exposure. Afterwards, the HY5 transcript pool declined to a level comparable to basal dark expression (Fig. 1b, c) following a pattern of transcript accumulation comparable to that of the PAP genes. This suggests a potential involvement of, or a correlation with, the photomorphogenesis programme in the transcriptional activation of the PAP genes upon illumination.

Putative PAP promoter regions display transcriptional activity

Since our RT-PCR analysis was performed with whole-plant RNA extracts tissue-specific responses of PAP gene expression could be masked. To obtain a more detailed view of PAP gene expression at the cellular level during seedling development, we decided to test the promoter activity of PAP genes using a GUS-reporter system.

We generated GUS-reporter gene fusions with the genomic untranslated regions of PAP1 to PAP10. All chosen regions included sequences upstream of the ATG and were restricted to a maximum size of 2 kbp (Fig. 2a). The promoter regions displayed high diversity in their annotation and their cis-element composition. The putative promoter regions of PAP2, 5, and 9 were found to correspond to rather large intergenic regions, while the other promoters were in regions of high gene density: PAP4, 6, and 8 promoters were identified close to the 5′-UTR of an upstream gene, while PAP 1, 3, 7, and 10 were found close to the 3′-ends of the upstream gene. All genes except PAP1 and 2 included a predicted transcription initiation start (TIS) (Fig. 2a). We, therefore, fused the chosen promoter sequences at the respective predicted translational start codon of each PAP gene with the open reading frame of the β-glucuronidase reporter gene. Hence, the constructs included all potential 5′-UTR regulatory sequences in addition to potential transcriptional cis-elements. In addition, we designed a positive control for the GUS assays using the promoter region of the ubiquitously expressed EXPORTIN1B (XPO1B) gene (Blanvillain et al. 2008), chosen for its low activity (Blanvillain et al. 2011) to avoid saturated GUS signals in the controls such as those commonly observed with the highly active pCaMV35S promoter.

Fig. 2
figure 2

PAP-promoter structures and functional test of the corresponding promoter::GUS constructs. a Representation of the chosen DNA regions (large white boxes) for the analysis of PAP1–PAP10 promoters. The annotations of elements within these regions are according to TAIR. Dark grey symbols represent upstream genes with introns (flexed lines) and exons (boxes). The arrowheads indicate the orientation of the upstream gene. Light grey symbols are untranslated regions (UTRs; bars) with intronic regions (flexed lines) or intergenic not annotated regions (gaps). + transcription initiation start (TIS). The black bar at the top represents 100 bp. pXPO1b is ubiquitously expressed and serves as a positive control for strong expression. The promoter regions are displayed all in 5′–3′ orientation. The promoter regions were directly fused to the β-glucuronidase (uidA) gene, without the open reading frame of the PAP gene. b, c Transcriptional activity of the promoter::GUS constructs in transient assays. Onion cells were co-transformed with one promoter::GUS construct and an internal control expressing GFP (pKar6). 24 h after the bombardment, the region of the onion peel with cells expressing GFP was identified and a GUS assay was performed. Onion epidermal cells transformed with pPAP::GUS were stained for 72 h; peels expressing pXPO1b::GUS were stained for 3 h. b Representatives of two independent transformation experiments are depicted. c Quantity of transformed cells was calculated by the numbers of cells showing GUS expression, divided by the number of cells showing GFP (NGUS/NGFP) within a defined region of the onion peel. d GUS assay with stable Arabidopsis transformants of the T2 generation using the same promoter constructs as in b. Seedlings were grown according to the experimental set-up (see Fig. 1a) and illuminated for 4 h with white light before the GUS assay. The duration of the GUS staining varied depending on the strength of the construct (PAP1: 22 h, PAP2: 17 h, PAP3: 42 h, PAP4: 42 h, PAP5: 17 h and 15 min, PAP6: 47 h, PAP7: 18 h and 45 min, PAP8: 17 h and 30 min, PAP9: 49 h, and PAP10: 49 h). Two representative seedlings per promoter::GUS construct (PAP1–10) are depicted

Full size image

Transcriptional activity of the promoter::GUS constructs was first tested by transient expression in the epidermal layer of onion cells. Therefore, onion peels were co-transfected with one PAP promoter construct and pKar6, carrying a constitutively expressed enhanced GFP targeted to the endoplasmic reticulum (pCaMV35S::tevL:eGFPer) that is used as an internal transfection control (Blanvillain 2000). After 24 h of expression, each onion peel was searched for a region displaying GFP signals, which was subsequently used for the GUS assay (Fig. 2b). To obtain an additional quantitative assessment, the ratio of cells expressing GUS divided by cells expressing GFP within the same area of transfection was calculated (Fig. 2c). All PAP promoter constructs displayed expression. However, PAP gene expression of PAP2–10 was with NGUS/NGFP ratios between 0.01 and 0.1, roughly 18–180-fold lower when compared to the control pXPO1B (Fig. 2b, c). Only pPAP1::GUS displayed a slightly increased NGUS/NGFP ratio that equals to 0.39 (when compared to the other PAP genes) which was around 4.6 times lower than the control (Fig. 2c). The comparable GUS-expression levels of the ten PAP genes support the hypothesis of co-regulation of the PAP genes and, in addition, validate the functionality of these constructs.

To investigate the PAP gene expression in their target cells, we stably transformed A. thaliana. Primary transformants were selected for hygromycin resistance, then genotyped by PCR, and grown to set seeds (Suppl. Fig. S1). In the second generation, at least 12 different lines per construct were tested for their GUS expression during the transition from skotomorphogenesis to photomorphogenesis (Fig. 2d, Suppl. Fig. S2). The expression pattern of each construct was considered reliable when several distinct transformation events of the same construct yielded similar results. We grew the plants according to the experimental set-up for 72 h in darkness (Fig. 1a) and transferred them for 4 h to white light before GUS staining. Out of the ten constructs, PAP4, 9, and 10 did not yield any detectable GUS expression (Fig. 2d), while all other PAP promoters led to largely cotyledon-specific GUS expression, which is in agreement with the previous descriptions of some pPAP::GUS expression patterns (Myouga et al. 2008; Chen et al. 2010; Wimmelbacher and Börnke 2014). However, some partial blue staining of the hypocotyl proximal to the cotyledons could be observed in the very strong expressing lines of PAP2 and PAP5. This coloration could be due to leakage of the blue product, but could also be the result of residual expression in the apical part of the hypocotyl where low formation of chloroplasts can be observed (Jin et al. 2001). The GUS activities of the pPAP1, pPAP3, pPAP6, and pPAP7 lines were consistently low but with very similar expression patterns specific to the cotyledons. The promoter constructs corresponding to PAP2, 5, and 8, however, yielded a robust and strong signal allowing for detailed in vivo transcriptional analysis of these PAP genes. Lack of any or weak GUS expression of the other pPAP::GUS constructs is likely caused by a low promoter usage and does not need necessarily to correspond to the level of transcript accumulation as detected in the RT-PCR (see “Discussion”).

Spatio-temporal expression patterns of pPAP::GUS constructs

To add cell- and tissue-specific information to the time-course results of the RT-PCR, strongly expressed pPAP::GUS lines were grown according to our experimental set-up (Fig. 1a), harvested at the indicated time-points (Fig. 3) and immediately transferred to GUS staining. The GUS activity was just under the threshold of detection in dry and imbibed seeds as well as after 20 h of dark growth, so that, using longer staining times, some weak signals could be detected. Then, the activity of PAP promoters increased with time during the whole period of skotomorphogenesis to reach a maximum level at 72 h of dark growth, with a similar pattern of GUS expression as seen for the 4-h illuminated seedlings (Figs. 2d, 3), showing GUS activity restricted to the cotyledons and some weak signal in the root apex (Supplemental Fig. S3). As the organ-specific expression of the PAPs could be already observed in the dark, this suggests a developmental control of PAP gene expression during the early phases of germination and seedling development rather than an environmental control only.

Fig. 3
figure 3

Promoter activities of PAP genes during skoto- and photomorphogenesis. Transcriptional activity during germination in homozygous plants carrying the strongly expressed pPAP8::GUS, pPAP5::GUS, and pPAP2::GUS constructs. Imbibed and stratified seeds were placed in darkness at 21 °C for 72 h to induce skotomorphogenesis (left panels). Afterwards, plants were either kept in darkness (lower right panels) or shifted to continuous white light (upper right panels). Samples were collected at the given time-points and a GUS staining was performed (17 h). Growth conditions and time-points of sample collection are indicated on top of each panel. Pictures show representative individuals. n.a not applicable. Scale bar indicates 1 mm

Full size image

The GUS activity in the cotyledons slightly increased within the first 15 min of illumination, while the staining after 1 h of light returned to the initial value found in 72 h dark-grown plants (Fig. 3, right top panel). These data are in agreement with the peak of transcript accumulation found by RT-PCR (Fig. 1c). The GUS activity further declined during illumination until the cotyledons of plants at the rosette stage did not express the pPAP::GUS constructs anymore. At later stages, strongest expression was found in young, newly emerging leaves, while older leaves displayed a progressive loss of expression in a centrifugal manner, from the inner part to the margins of the leaf blade (Fig. S4). Control plants kept in darkness, however (Fig. 3, lower right panel), exhibited steady high levels of GUS expression. Together, the data suggest a dual control mode of PAP gene expression including a first light-independent developmental control followed by a second level involving a light-dependent transient activation of the PAP genes. This dual mode could support (1) high PEP activity during the initial steps of chloroplast biogenesis and (2) maintain a lower, basal PEP activity in later stages when photosynthesis is established within the fully developed chloroplast.

Activity of PAP2, PAP5, and PAP8 promoters correlate with the presence of photosynthetic tissues

To obtain more insights into the putative dual control mode, we analysed the expression patterns of pPAP2::GUS, pPAP5::GUS, and pPAP8::GUS at various developmental stages. As indicated by the previous in silico analyses (Steiner et al. 2011; Pfannschmidt et al. 2015), the three PAP promoters displayed similar expression patterns, although pPAP5::GUS exhibited the strongest activity. We observed GUS expression in green leaf buds of all three GUS lines (Fig. 4a, e, i; black arrows). However, fully developed cauline leaves were almost free of GUS expression and only, in some cases, a blue coloration of the leaf tip was detectable. This is reminiscent of the results described above for rosette leaves (Fig. 3, upper right panel). Hence, the transcriptional activity of PAP gene promoters is highest in leaf primordia and diminishes with leaf development and ageing.

Fig. 4
figure 4

PAP gene expression in flower buds, flowers, and siliques. Tissue was subjected to GUS staining for ~ 21 h. Pictures were taken on the Nikon AxioScope light microscope. a–d pPAP8::GUS. e–h pPAP5::GUS. i–l pPAP2::GUS. a, e, i Expression of pPAP::GUS constructs in cauline leaf and flower buds. Black arrows indicate the leaf buds. b, f, j Expression of pPAP::GUS constructs in flowers of increasing age. c, g, k Expression of pPAP::GUS constructs in siliques. d, h, l Higher magnification of flowers from b, f and j displaying style, stigma, and anthers. m Cross section of a closed flower (stage 9) expressing pPAP8::GUS. n Cross section of an open flower (stage 15) expressing pPAP8::GUS. Symbols in the upper right corner of m and n correspond to the icons for flower developmental stage according to the eFP browser. Asterisks indicate the carpels

Full size image

To build an expression map of PAP genes, we performed GUS assays on flowers of different developmental stages as well as on siliques. Late flower buds (stage 9–12) did express the PAP promoter constructs (Fig. 4a, e, i). Histological cross sections of these buds revealed pPAP8::GUS expression both in sepals and petals (Fig. 4m). In open flowers, however, GUS expression was not present anymore in petals (Fig. 4b, f, j, n). This correlates to observations reporting that petals of unopened flower buds contain chloroplasts that turn later into leucoplasts in fully developed and expanded petal blades (Pyke and Page 1998). In hypocotyl and roots (Figs. 2d, 3), we observed similar PAP-expression patterns with none or low PAP gene expression in tissues in which plastids differentiate into leucoplasts. Anthers and pollen grains remained unstained in all stages of flower development (Fig. 4d, h, l–n). Starting from the microspore at meiosis, pollen grains contain only amyloplasts and proplastids, and anthers lose chloroplasts that differentiate into chromoplasts during the early stages of microsporogenesis (Clément and Pacini 2001).

From these observations, we conclude that activity of PAP gene promoters positively correlates with tissues that generate specifically chloroplasts, while PAP expression vanishes upon differentiation of plastids into other forms, such as leucoplasts or amyloplasts. In agreement with that conclusion, we found that green flower tissues such as sepals, carpels, and silique valves at later stages, but also, to lesser extent, stamen filaments displayed GUS expression with some variation in intensity. While pPAP2::GUS and pPAP5::GUS lines displayed strong expression, fainter expression was detected for pPAP8::GUS (Fig. 4b–d, f–h, j–n).

The Arabidopsis embryo is known to be photosynthetically active from the globular to the bent cotyledon stage generating bona fide chloroplasts during this phase (Tejos et al. 2010; Le et al. 2010; Allorent et al. 2013, 2015; Liebers et al. 2017). Hence, we surmised that, if PAP gene promoters are always active in tissues forming chloroplasts, they should be active also during embryogenesis. Indeed, a blue coloration corresponding to pPAP8::GUS expression was clearly detected in the central cell of the ovule, whereas GUS activity was never detected in the maternal tissue at any stages of embryogenesis (Fig. 5a). At later stages in the embryo, GUS activity was low at the globular and the heart stage (Fig. 5b, c). Starting with the torpedo stage, the embryo was the only tissue that exhibited GUS staining, although at very low levels (Fig. 5d–f). pPAP2::GUS showed similar results with a more intense blue coloration (Fig. 5g–l) that was additionally visible in the maternal tissue of the ovule (Fig. 5g) and at the globular stage (Fig. 5h). pPAP5::GUS was expressed in the maternal tissue, the endosperm and the embryo proper throughout embryogenesis (Fig. 5m–r).

Fig. 5
figure 5

PAP gene expression during embryogenesis. Siliques along the main stem were collected, seeds separated, and GUS stained for ~ 18 h. After clearing of the seed coat with Hoyer’s solution pictures were taken on the Nikon AxioScope light microscope. a–f Expression of pPAP8::GUS. g–l Expression of pPAP2::GUS. m–r Expression of pPAP5::GUS. a, g, m Ovule. b, h, n Globular stage. c, i, o Heart stage. d, j, p Torpedo stage. e, k, q Walking stick. f, l, r Bent cotyledon. Black arrows indicate the embryo. Scale bar = 20 µm (a, g, m), 50 µm (b–f, h–l, n–r)

Full size image

The observed PAP gene expression in endosperm and seed coat coincides with the observation of photosynthetic activity in the outermost parts of pea and barley seeds (Tschiersch et al. 2012; Radchuk and Borisjuk 2014). It was hypothesized that photosynthesis provides oxygen to the otherwise hypoxic tissues and, therefore, enhances the formation of starch, which is partially dependent on the supply of oxygen (Gifford and Bremner 1981; Radchuk and Borisjuk 2014). The importance of chloroplast formation during embryogenesis was recently demonstrated, showing that inhibition of chloroplast biogenesis causes strongly reduced accumulation of storage reserves (Liu et al. 2017), which negatively influence the fitness of the seed as seen by delayed germination and shortening of seed storability (Allorent et al. 2015). The differences observed in the pattern of the three pPAP::GUS lines, however, may reflect some gene-specific features such as different strength in expression or additional functions unrelated to our focus on the PEP activity. Nonetheless, these data confirm PAP promoter activity prior and during the greening phase of embryogenesis. In sum, the data show that the expression of PAP genes occurs specifically in young tissues that will develop functional chloroplasts.

PAP promoter activity in cotyledons occurs in a cell-specific pattern

Based on the results above, we expected that PAP expression should occur in photosynthetic tissue and, therefore, in the mesophyll cells of the leaf blade. We intended to analyse the cellular specificity of PAP expression in 4 h illuminated cotyledons. To this end, plants were grown as described above (Fig. 2d), followed by GUS assay (Fig. 6a) and eosin staining. The samples were embedded in wax, cut into 7 µm sections, and visualized under a light microscope (Fig. 6b–f). Longitudinal and cross sections of cotyledons carrying pPAP8::GUS revealed that the GUS expression occurred mainly in the epidermal layer, while faint GUS expression was detectable in the mesophyll tissue (Fig. 6b, c, f). The same expression patterns were detected in pPAP2::GUS and pPAP5::GUS lines, displaying strong GUS expression in the upper and lower epidermis and very low expression in the spongy and the palisade mesophyll (Fig. 6d, e). Hence, the strong promoter activity in the epidermal layer appears to be a common feature of PAP genes.

Fig. 6
figure 6

Histological sections of 4 h illuminated cotyledons determine cell-type specificity of PAP expression. Seedlings were grown according to the experimental set-up (see Fig. 1a) After 4 h of illumination and subsequent GUS staining, seedlings were used for preparation of histological sections. Imaging of cell-specific GUS staining was done under a Nikon AxioScope light microscope. a Expression of pPAP8::GUS in cotyledons after GUS staining. b Longitudinal section of a seedling expressing pPAP8::GUS. c Higher magnification of the cotyledons from b. The adaxial and abaxial leaf sides are indicated. d–f Cross sections of leaves from lines expressing pPAP2::GUS, pPAP5::GUS and pPAP8::GUS constructs. g Illumination-dependent expression of pPAP8::GUS constructs in histological cross sections of Arabidopsis leaves of different age. h 3-week-old rosette leaf

Full size image

To investigate whether the observed epidermal expression pattern changes in a light-dependent manner cross sections of dark-grown pPAP8::GUS lines and samples illuminated for 4 h, 24 h (Fig. 6g) or 3 weeks were prepared (Fig. 6h). Dark-grown seedlings displayed expression patterns comparable to that of 4-h illuminated cotyledons (Fig. 6g). However, 24-h illuminated cotyledons and 3-week-old rosette leaves displayed an almost equalized GUS expression both in epidermal and mesophyll cells, though the overall staining was weaker, when compared to the epidermal layer of 4-h illuminated seedlings (Fig. 6g, h). Taken together, the light-independent expression of the PAP genes (Figs. 1b, c, 3) is specific to the cotyledons and, in particular to the epidermal layer. Thus, tissue identity might play a crucial role in PAP promoter activity. Longer light exposure, however, leads to a reduction of PAP gene expression in the epidermis (Figs. 1b, c, 3) leading to equal expression in epidermis and mesophyll cells (Fig. 6g, h). These data strongly suggest that PAP gene expression is regulated by a tight interplay of developmental stage, light exposure, and tissue identity.

Discussion

PAP genes form a regulon

PEP-associated proteins were originally identified in biochemical studies as stably associated components to the PEP-A complex of mustard (Steiner et al. 2011). Lack of any PAP by genetic inactivation was found to result in albino/pale-green phenotypes of corresponding mutants (Pfalz and Pfannschmidt 2013). The best explanation for this effect so far is that the missing protein subunit either de-stabilizes the PEP-A complex or hampers its assembly, resulting in a severely disturbed but not blocked expression of plastid photosynthesis genes (Grübler et al. 2017). Based on the finding that a functional PEP-A complex was found only in illuminated seedlings (Pfannschmidt and Link 1994; Pfannschmidt et al. 2000), it was proposed that the complex is generated in a light-dependent manner. Indeed, expression profiles of some individual PAP genes were reported (Gao et al. 2011; Yagi et al. 2012; Yu et al. 2013) pointing to a potential light-dependent co-regulation of them. However, these results are difficult to compare, since growth conditions of seedlings varied significantly between the different analyses. To strengthen the co-expression hypothesis that is mainly based on in silico analyses (Steiner et al. 2011; Pfannschmidt et al. 2015), we compared expression profiles of all ten PAP genes in one experimental approach (Fig. 1). The pattern of expression during the skoto- to photomorphogenesis transition was found to be consistent among all PAPs supporting the idea of co-expression (Fig. 1). Interestingly, using RT-PCR, we detected significant transcript accumulations of all PAP genes already in the dark while illumination induced, when it applies a relatively weak and transient peak. As known so far, the corresponding proteins do accumulate swiftly after light exposure (Pfannschmidt and Link 1994; Yagi et al. 2012). Likely, transcript accumulation in the dark allows for a rapid translation of PAPs upon illumination, as it is known for transcripts of photosynthesis genes. The observed accumulation patterns, however, have to be considered with care as the RT-PCR cannot distinguish between the different transcriptional activities of PAP genes in epidermis and mesophyll cells (see further below) and, therefore, do not represent an accurate quantitative measure. Nevertheless, the detection of PAP transcripts by RT-PCR can be taken as marker that in the investigated organs chloroplasts will develop upon illumination.

Further positive proof for the co-expression hypothesis came from the analysis of the three promoter::GUS constructs pPAP2::GUS; pPAP5::GUS and pPAP8::GUS (Fig. 3). Although differences in strength of promoter activity were observed in these lines, spatial and temporal expression patterns were comparable for all of them throughout the life cycle of Arabidopsis. This is true at the level of organs in flowers, siliques and embryos, and at the level of cell identity in cotyledons and leaves. The differences in expression strengths between individual PAPs, therefore, might reflect potential compensation mechanism for lower stability of the corresponding proteins. The PAP5 gene promoter, for instance, displayed highest activity among all PAP promoters in our hands, but the corresponding protein is known to contain a PEST domain, which is a signature of short-lived proteins (Chen et al. 2010). Future investigations will address questions on relative protein stabilities among the different PAPs. The observation that only three pPAP::GUS constructs yielded a degree of expression that was sufficient for observation, while all PAP genes were detected in the RT-PCR could argue against the co-expression hypothesis. However, the two technologies used here focus on very different steps in gene expression (i.e., promoter usage versus transcript accumulation) and differences between them can be easily explained by, e.g., selective transcript stabilization or reduced transcript degradation. Therefore, the data obtained by the two techniques are rather complementary than contradictory.

Although, in general, PAP genes displayed a high degree of co-regulation, we detected also some differences in the spatial expression of PAP2, PAP5 and PAP8 during embryogenesis (Fig. 5). While the pPAP5::GUS construct was expressed in all tissues throughout the embryo development, expression of pPAP2::GUS and pPAP8::GUS in the endosperm occurred only in the ovule and at the globular stage. In subsequent developmental stages, expression was restricted to the embryo itself. These differences, however, might be caused simply by the detection threshold of the method, since the GUS staining appeared to be around three-times higher in pPAP5::GUS lines compared to PAP2 and PAP8. Hence, it is likely that pPAP2::GUS and pPAP8::GUS levels simply did not reach a level of detection in the endosperm of older embryos. This assumption is supported by available microarray data uncovering a decrease of general PAP expression in the endosperm starting from the heart stage (Belmonte et al. 2013). Nonetheless, it cannot be fully excluded to date that differences in specific spatial PAP gene expression are due to gene-dependent effects. PAPs encompass multiple different functions that may require individual regulation at certain localizations or conditions that are unknown to date.

Overall, our results support the hypothesis that PAP genes belong to a regulon and, thus, are likely controlled by a common regulator. One potential candidate for this is HY5, a bZIP transcription factor that is crucial for the induction of photomorphogenesis (Oyama et al. 1997; Quail 2002; Eckardt 2007; Lee et al. 2007; Li et al. 2011). The transcript accumulation profile of HY5 parallels the transcript accumulation of PAP genes during the skoto-to-photomorphogenesis transition (Fig. 1), suggesting that HY5 may act as an activator of PAP gene expression. On the other hand, HY5 becomes degraded in darkness due to its direct interaction with the ubiquitin E3 ligase CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) (Ang et al. 1998; Osterlund et al. 2000; Saijo et al. 2003). This could disqualify HY5 as the responsible regulator of basal PAP gene expression in the dark unless HY5 follows similar cell-specific regulations. Should HY5 be involved in the light-induced peak of PAP gene expression, it would likely not be the only one, since hy5 null mutant plants display a green phenotype with fully developed chloroplasts. We concluded that other factors of the complex and functionally redundant light-signalling network in Arabidopsis are involved. Furthermore, we could recently demonstrate that albino pap7-1 mutants grown on sucrose-containing medium do develop a largely normal phenotype despite their albinism, indicating that the photomorphogenic programme and the photoreceptor-mediated signalling network are functional in this mutant background (Grübler et al. 2017). This indicates that chloroplast biogenesis is not a strict requirement for photomorphogenesis; however, proper expression of all PAP genes is essential for chloroplast formation. This suggests a complex regulatory network that connects PAP gene expression, chloroplast biogenesis, and the photomorphogenic programme. More work will be necessary to understand the molecular nature of these relationships.

PAP genes display a specific spatio-temporal expression pattern

It was demonstrated that inactivation of any PAP gene results in pale-green/albino phenotypes due to an arrest in the build-up of functional chloroplasts (Pfalz et al. 2006; Myouga et al. 2008; Arsova et al. 2010; Chen et al. 2010; Steiner et al. 2011; Gao et al. 2011; Yagi et al. 2012; Gilkerson et al. 2012; Yu et al. 2013; Wimmelbacher and Börnke 2014; Pfalz et al. 2015). This renders PAPs an essential prerequisite for proper chloroplast formation. However, so far, it was not investigated whether PAP genes are also expressed in non-photosynthetic tissue. Here, we provide a comprehensive expression catalogue for PAP2, 5, and 8 in various tissues throughout the whole life cycle of Arabidopsis plants, demonstrating that the expression of these genes occurs in a specific spatio-temporal manner.

PAP expression in non-photosynthetic tissues is actively repressed

Expression of pPAP2::GUS, pPAP5::GU, and pPAP8::GUS was not detectable in roots and hypocotyls (Figs. 2, 3), which is consistent with our in silico analyses and the previous studies that reported little or no PAP gene expression in such tissues (Pfalz et al. 2006; Myouga et al. 2008; Chen et al. 2010; Gao et al. 2011; Yu et al. 2013; Wimmelbacher and Börnke 2014; Pfannschmidt et al. 2015). However, although roots contain mainly leucoplasts, they have the capability to form chloroplasts, as demonstrated recently (Jin et al. 2001; Usami et al. 2004; Hermkes et al. 2011; Kobayashi et al. 2012). Therefore, roots and hypocotyl might be able to express PAP genes under some circumstances, which could explain the partial blue coloration of the hypocotyl in the strong expressing lines of PAP2 and PAP5 (Fig. 2). The same is true also for the staining in the root apex of PAP5, which appears to be restricted to the undeveloped meristematic cells (Fig. 2; Fig. S3). Basal expression of PAPs in young cells with low degrees of differentiation is also observed in various embryonic tissues (Fig. 5) and may indicate the postulated developmental control of PAPs. Interestingly, this may explain why, in some cases, roots can green. For example, the lack of COP1 and DE-ETIOLATED 1 (DET1) leads to strongly enhanced chloroplast formation and greening in roots and hypocotyls (Chory and Peto 1990; Deng et al. 1992). COP1 and DET1 promote the targeted degradation of proteins involved in photomorphogenesis, and hence, their absence leads to de-repression of chloroplast biogenesis. Consequently, COP1 and DET1 are not only involved in the repression of photomorphogenesis in the dark, but also in illuminated roots (Quail 2002; Usami et al. 2004; Lopez-Juez and Pyke 2005; Kobayashi et al. 2012) and one could postulate that also PAPs are actively degraded by them. Furthermore, cytokinin can promote root greening, while auxin works antagonistically indicating an involvement also of the hormonal network (Kobayashi et al. 2012; Chiang et al. 2012). In sum, chloroplast formation appears to be controlled by organ identity and is either actively repressed (roots and hypocotyl) or the repression is released (cotyledons and leaves) due to the interplay between light- and hormone-dependent signalling pathways (Kobayashi et al. 2012; Liebers et al. 2017). This suggests that PAP gene expression is controlled by these networks.

PAP expression is regulated by light and internal developmental cues

We could demonstrate that PAP2, PAP5, and PAP8 are transcribed in seeds and dark-grown seedlings, which corroborate the previous studies that examined PAP1, PAP5, PAP7, and PAP12 transcripts within dark-grown tissue (Chen et al. 2010; Gao et al. 2011; Yagi et al. 2012; Yu et al. 2013). Moreover, GUS expression of dark-grown seedlings occurred exclusively in cotyledons; hence, expression is restricted to tissues that are prone to become photosynthetic (Fig. 3). Therefore, we conclude that basal transcription and transcript accumulation of PAPs is controlled by internal developmental signals in a light-independent manner. However, although, in Arabidopsis, PAP gene expression occurs during skotomorphogenesis, an active PEP-A complex could be purified only from de-etiolated or fully green seedlings in mustard (Pfannschmidt and Link 1994). A species-specific difference cannot be fully excluded, but is unlikely, since mustard and Arabidopsis belong to the same family of Brassicaceae, and has been a useful substitute model for Arabidopsis in biochemical studies (Schröter et al. 2010, 2014). We assume that either translation or the association of PAPs to the PEP-core enzyme is regulated by light. PAP1 was reported to accumulate in dark-grown Arabidopsis seedlings to only trace amounts but exhibited massive accumulation within 3 h of illumination (Yagi et al. 2012). This favours the scenario of a light induction of PAP translation and complex assembly. In addition, also the transcription of PAP genes exhibited some light-induced regulation. We observed that most PAP transcripts transiently increase in response to the onset of light (Figs. 1, 3), which is consistent with other reports (Gao et al. 2011; Yu et al. 2013; Wimmelbacher and Börnke 2014). A subsequent decrease in PAP gene expression within 24 h of illumination was detectable that continued until expression vanished in older rosette (Fig. S4) and cauline leaves (Gao et al. 2011; Yu et al. 2013; Wimmelbacher and Börnke 2014). This observation might be explained best by the fact that, after ~ 24 h of illumination, the majority of chloroplasts are formed and the organelle biogenesis is completed (Liebers et al. 2017). Thus, the demand for chloroplast transcription is reduced. This assumption correlates with the early observations on photosynthesis-associated gene transcripts of maize, which displayed a rapid increase upon illumination, followed by a decrease to pre-illumination levels after 20–44 h (Rodermel and Bogorad 1985). Furthermore, also older barley and Arabidopsis leaves displayed a reduced plastid gene transcription (Mullet and Klein 1987; Baumgartner et al. 1989; Zoschke et al. 2007) and the translation of most maize plastid transcripts declines, once the photosynthetic apparatus is established (Chotewutmontri and Barkan 2016). We conclude that the PEP-A complex functions mostly in the maintenance of the chloroplast, once the initial biogenesis is completed. For that, only a relatively small number of complexes might be sufficient, which would explain the difficulties to purify it from mature leaf material. Interestingly, mass spectrometry data suggest that the overall-function of the nucleoid changes from RNA metabolism and ROS quenching in proplastids towards homeostasis and DNA repair in chloroplasts (Majeran et al. 2012). This view supports the hypothesis that the transcription machinery of fully developed chloroplasts is mainly involved in maintenance of the status quo. It should be, however, noted that nothing is known about the stability of the PEP-A complex. Therefore, its turnover rate remains to be elucidated.

PAP expression is dependent on organ and tissue identity

During skotomorphogenesis, transcription of PAP genes occurred only in cotyledons, an organ that will become photosynthetic, while non-photosynthetic organs, such as roots and hypocotyls did not express PAP genes. Thus, PAP gene expression appears to be strongly regulated by organ identity. This assumption is supported by the results from the pPAP::GUS constructs throughout different stages of plant development. During embryogenesis, PAP expression occurs in the seed coat, the endosperm, and the embryo itself, already prior to the greening phase of the seed (Fig. 5). However, during developmental phases in the light, only chloroplast containing tissues do express PAP genes. Expression in the flower was specific to sepals, the carpel, and partially the stamen filaments, while non-photosynthetic tissues like anthers and pollen grains did not express pPAP::GUS constructs. Upon chloroplast-to-leucoplast transition in petals (Pyke and Page 1998), the pPAP::GUS expression vanished (Fig. 4). Hence, PAP gene expression strongly correlates with organs or tissues that will or do contain chloroplasts. Therefore, expression of PAP genes can be seen as a biomarker for the biogenesis of chloroplasts in a given tissue.

Most interestingly, PAP gene expression is even cell-type specific, since the expression is restricted to the epidermal layer of the cotyledons during the early seedling development (Fig. 6). This epidermal expression is not yet understood but may support recent publications demonstrating that the epidermal layer does contain chloroplasts. These plastids are different from “normal” chloroplasts as they are low abundant, 50–70% smaller than typical chloroplasts from mesophyll tissue, and are proposed to possess a sensor function for environmental signals striking the epidermis (Pyke and Leech 1994; Virdi et al. 2016; Barton et al. 2016). The strict epidermal expression was, however, only present during skotomorphogenesis and the first hours after the onset of light, while, after 24 h of illumination, GUS expression was present in epidermis and mesophyll cells (Fig. 6). Why expression occurs first in the epidermal layer and only later in the mesophyll tissue will be a matter of future research. This observation, however, indicates that even cell-type-specific regulation events must be considered in the early steps of photomorphogenesis and chloroplast build-up. Interestingly, the formation of chloroplasts during embryogenesis and development of leaf primordia in the shoot apical meristem occurs first in the epidermal layer (Tejos et al. 2010; Charuvi et al. 2012). Chlorophyll fluorescence at the globular and the heart stage was detected specifically in the protoderm or in the epidermis, respectively, while the ground tissue started to exhibit chlorophyll fluorescence at the torpedo stage. In the walking stick stage, however, chlorophyll fluorescence was detectable throughout the whole embryo (Tejos et al. 2010). Correspondingly, in silico analysis of microarray data revealed increased PAP transcripts at the heart stage and in the linear cotyledon (Belmonte et al. 2013; Kremnev and Strand 2014). In the shoot apical meristem, the L1 layer, which generates the epidermis, does contain chloroplasts, while the central zone of the L2 layer that gives rise to the mesophyll tissue contains only proplastids. Young leaf primordia that developed from the L1 and L2 layers, however, contain chloroplasts, both, in the epidermis and the mesophyll cells (Charuvi et al. 2012). In addition, mature leaves contain chloroplasts in epidermis and mesophyll tissue, although the majority of epidermal chloroplasts have re-differentiated into leucoplasts (Pyke and Leech 1994; Charuvi et al. 2012; Virdi et al. 2016; Barton et al. 2016). In sum, these data suggest that PAP gene expression in the epidermal cell layer has a specific role in the early seedling development and concomitantly in primary chloroplast formation. Interestingly, it has been reported that phytochrome B also displays an epidermal expression in cotyledons of dark-grown seedlings (Somers and Quail 1995) pointing to a potential connection between the epidermal PAP expression and the light-signalling network that controls photomorphogenesis. Future research on the epidermal function of the PAPs during the early stages of chloroplast biogenesis may uncover novel features of organelle development.

Based on the expression data presented here, we conclude that PAP genes are already expressed prior to the initiation of chloroplast biogenesis, but only in tissues that will develop chloroplasts upon illumination (except for the expression in the root apical meristem which will require separate investigation). It is unknown yet whether these PAP transcripts are used for translation in a light-dependent manner. If so, the resulting PAPs may translocate into proplastids or etioplasts and trigger the transformation of the PEP-B into the PEP-A complex by assembly of the PAPs to the PEP core. This process would likely operate within minutes and could specifically initiate the expression of plastid-encoded photosystem II genes providing essential building blocks for subsequent chloroplast biogenesis. PAP genes, therefore, represent genuine biomarkers for chloroplast biogenesis in plant tissues and might be used as analytical tools helping to understand the initiation of chloroplast formation also in other species than Arabidopsis.

Author contribution statement

ML and RB conceived and designed research. ML, FC, and RB conducted experiments. ML, RB, and TP analysed data. ML, RB, and TP wrote the manuscript. All authors read and approved the manuscript.