Nucleo-plastidic PAP8/pTAC6 couples chloroplast formation with photomorphogenesis

The initial greening of angiosperm occurs upon light-activation of photoreceptors that trigger photomorphogenesis followed with the development of chloroplasts. In these semi-autonomous organelles, the construction of the photosynthetic apparatus depends on the coordination of nuclear and plastid gene expression. Here we show that PAP8, as an essential subunit of the plastid-encoded RNA polymerase, is under the control of a regulatory element recognized by the photomorphogenic factor HY5. PAP8 is localized and active in both plastids and the nucleus and particularly essential for the formation of late photobodies. In the albino pap8 mutant, phytochrome-mediated signalling is altered, PIFs are maintained, HY5 is not stabilized, and GLK1 expression is impaired. PAP8 translocates into plastids losing its pre-sequence, interacts with the PEP, and using an unknown route or a retrograde transport, reaches the nucleus where it has the ability to interact with pTAC12/HMR/PAP5. Since PAP8 is required for the phytochrome-B-mediated signalling cascade and the reshaping of the PEP, it may coordinate nuclear gene expression with the PEP-driven chloroplastic gene expression during chloroplast biogenesis.


Introduction
Chloroplasts are the organelles conducting photosynthesis in plants and green algae (Jarvis & Lopez-Juez, 2013. In angiosperms, the photosynthetic organelles differentiate from proplastid in a light-dependent manner involving photoreceptors. Seedlings sheltered from light, launch a chloroplast-free developmental programme, named skotomorphogenesis, leading to cell elongation of specific organs (e.g. the hypocotyl). The inhibition of chloroplast development in the dark can be regarded as a way to optimize the use of limited resources stored in the seed to efficiently reach the surface. Then illumination causes the conversion of photoreceptors such as the phytochromes into an active state launching a developmental programme named photomorphogenesis (Solymosi & Schoefs, 2010). This programme involves the repression of hypocotyl elongation and the opening of the cotyledons that are rapidly engaged in chloroplast biogenesis, a tightly regulated process usually regarded as a part of the photomorphogenic programme (Pogson, Ganguly et al., 2015).
As remnant of their endosymbiotic origin, plastids possess their own genetic system, which contributes to the construction of the photosynthetic apparatus after illumination (Jarvis & Lopez-Juez, 2013). A plastid-encoded RNA polymerase (PEP) is required for proper transcription of photosynthesis genes encoded by the plastid genome. The PEP complex is composed of a prokaryotic core of 4 distinct bacterial-like subunits (α 2 , β, β´, β´´) surrounded by (at least) 12 additional nuclear-encoded subunits of eukaryotic origin (Pfannschmidt, Blanvillain et al., 2015) known as PEP-associated proteins (PAPs). The association of PAPs to the prokaryotic core is strictly light induced and phytochrome-mediated (Pfannschmidt & Link, 1994, Yang, Yoo et al., 2019, Yoo, Pasoreck et al., 2019. Importantly, the PAP association to the core of the PEP appears to be one important bottleneck of chloroplast formation since genetic inactivation of any PAP results in albinism (Pfalz & Pfannschmidt, 2013). The genes for PAPs appear to represent a regulatory unit that exhibits very similar coexpression profiles albeit the involved genes encode proteins that belong to very different functional classes that could not be predictably united before. They all exhibit a basal expression in the dark followed by a rapid and transient peak after light exposure strongly suggesting a connection of their expression to the light regulation network .
Stabilized HY5 can then initiate expression of photomorphogenic factors (Lee, He et al., 2007). Meanwhile, light exposure triggers the transcriptional activation of GOLDEN2-LIKE 1 and 2 (GLK1 and 2) transcription factors that are responsible for the proper expression of nuclear photosynthesis genes , Waters, Wang et al., 2009). The action of phytochrome within the nucleus was visualized using a GFP-tag (PHYB-GFP or PBG) revealing that phytochrome B translocates into the nucleus, and then aggregates into specific speckles within the nuclear matrix (Yamaguchi et al., 1999). Early speckles are numerous and small, while later speckles become larger and less abundant. Late speckle formation, also designated "late photobodies", requires the presence of HEMERA (HMR), a bi-localized protein present in the nucleus and in plastids (Chen, Galvao et al., 2010, Nevarez, Qiu et al., 2017. In plastids HMR is known as pTAC12/PAP5 representing a member of the PAP family that is essential for chloroplast biogenesis since genetic inactivation of the protein blocks plastid differentiation leading to albinism (Pfalz, Holtzegel et al., 2015, Pfalz, Liere et al., 2006. For PAPs, different functions can be predicted from their amino-acid sequences, but their precise roles are not yet understood. PAP8 is one of the most enigmatic members among the PAPs, as its deduced amino-acid sequence does not harbour any known functional domain. Here we show that PAP8 is a bi-localized nucleoplastidic protein with a nuclear pool important for the proper timing of chloroplast biogenesis.
In particular PAP8 is essential for phytochrome-mediated signal transduction, PIF1 and PIF3 degradation, HY5 stabilization and GLKs transcript accumulation indicating that it represents a novel key regulator of the light-signalling network.

Results
PAP8 plays an essential role in chloroplast biogenesis. PAP8 was originally identified by targeted proteomics as pTAC6, a component of the transcriptionally active chromosome of plastids. The T-DNA insertion line 'SALK_024431' of Arabidopsis, referred as the pap8-1 allele in this study, displayed an albino phenotype (Fig. S1) with a strong depletion of photosynthesis transcripts and pigments accumulation as well as developmentally arrested plastids (Pfalz et al., 2006). An orthologous protein of pTAC6 was then isolated from a purified Sinapis alba PEP complex and subsequently renamed PAP8 as being a bona fide component of this PEP complex (Steiner, Schroter et al., 2011). The pap8-1 allele corresponds to the insertion of an inverted repeat of the T-DNA into the first intron of the gene (Fig. 1A). Amplicon sequencing, after PCR-based genotyping (Fig. 1B) 1C), indicating that pap8-1 is a genuine null allele. The conceptually translated protein sequence is found in the terrestrial green lineage starting from mosses to Eudicots (Fig. 1D), though absent in ferns, gymnosperms and a few basal angiosperms. A predicted N-terminal chloroplastic transit peptide (cTP) rapidly diverged while a highly conserved region (HCR) of unknown function seems to be under a strong selection pressure, as it is almost unchanged since the last common ancestor of all terrestrial plants. Hence the sporophytic lethality of the pap8 mutant triggers the assumption that the protein had brought an important function to the green lineage in its way to conquer dry lands, and then became essential to Eudicots.
In all orthologous proteins a nuclear localization signal (NLS) could be predicted, pointing to a possible localization of the protein inside the nucleus. Functional complementation (Fig.   S2) using a full length coding sequence of PAP8 driven by 1.1 kb of its own promoter (pPAP8::P8; Table S1) could fully restore the greening of the mutant with a chlorophyll content undistinguishable from that of the wild type ( Fig. 1G-I; Fig. S1D-E). Heterozygotes were phenotypically undistinguishable from wild type except within the developing silique where one quarter of the embryos were unable to green ( Fig. 1E) following, without gametic distortion, the Mendel's ratio for the segregation of recessive alleles (Fig. S1B). Mutant homozygotes, however, were albino and sporophytic lethal, with a strong reduction of cotyledon size (Fig. S1C). pap8-1 dies quickly after light exposure unless grown in vitro on a carbon source in dimmed light (Fig. 1F). Albeit their heterotrophic growth, plants pursued a rather normal development until reproduction. PAP8 is therefore a specific factor essential for chloroplast biogenesis without noticeably affecting plastid functions or the apparent photomorphogenic programme that is associated with de-etiolated plants.
The pap8-1 promoter involves typical light-responsive cis-elements. PAP genes are transcriptionally co-regulated ; as a canonical example, the promoter activity of PAP8 is transitorily specific to tissues with photosynthetic potential such as the cotyledons and leaf primordia. It is first restricted to the epidermis during skotomorphogenesis, induced in the palisade after light exposure and then slowly diminished . Searching for cis-regulatory elements by a deletion series of the PAP8 promoter, a short sequence starting at -97 from the transcriptional initiation start (tis) was found to be sufficient to retain cotyledon specificity while a construct starting at position +1 completely lost its reporter activity ( Fig. 2A, B). The two short versions of the promoter (-257 and -97) driving PAP8 expression were able to complement pap8-1 (Table   S1). Within the 97-bp region (Fig. 2C), a nearly palindromic element (GAcGCTC) was predicted to be a putative non-symmetrical element recognized by proteins with basic leucine zipper domains (bZIP). Site-directed mutagenesis of this element resulted in a disturbed GUS expression (Fig. 2B). Using PlantPAN3 (Chow et al., 2018) three bona fide elements for bZIP transcription factors (TF) were predicted in both strands of the DNA ( Fig. S3; Table S2).
Interestingly the two bZIP TFs, HY5 and HYH are known to be involved in the early steps of photomorphogenesis (Holm et al., 2002, Li, Zheng et al., 2017. Hence a few bZIP TFs, TGA2 as the best prediction according to the two elements found on the plus strand, HY5 and HYH as educated guesses, and bZIP60 as an out-group related to stress response (Iwata, Fedoroff et al., 2008) were tested in a dual luciferase reporter assay (Fig. S4). HY5 proved to be the most efficient, enhancing transcriptional activity of the long (-1133 bp) PAP8 promoter region by more than 5 fold over the control (Fig. 2D). For the shorter though functional -97-bp-promoter, HY5 promoted transcriptional activity with a 2-fold increase while a 3-bp replacement in the core of the element yielded significantly reduced activation.
Moreover recombinant HY5 was able to specifically bind the cis-regulatory element in vitro (Fig. 2E) in strength comparable to that of the canonical G-box element used as competitor (Yoon, Shin et al., 2006). In addition, the release of Chromatin-Immuno-Precipitation (ChIP) sequencing data using "GFP" antibody on a hy5/HY5::HY5-YFP genetic background (Hajdu, Dobos et al., 2018) allowed the detection of HY5 on the 5'-region containing the identified regulatory element and the 3'-region of PAP8 after blue light or red light exposure (Fig. 2F).
While the expression of PAPs is essential for greening, hy5 mutants display slight greening defects indicating that functional redundancies and compensations occur in the regulation of its target genes (Gangappa & Botto, 2016). For example, the paralogous transcription factor HYH (Holm et al., 2002) is also active on the PAP8 promoter (Fig. S4). In conclusion ChIP and EMSA indicate that HY5 can bind the PAP8 promoter and that it can activate the promoter in a heterologous system, but given that no expression changes were seen in a hy5-1 mutant, possibly due to functional redundancy, the ChIPseq/EMSA/transactivation data remains to be challenged in more sophisticated genetic backgrounds. Moreover, the epidermal specificity of the PAP-promoter activity during skotomorphogenesis may result from a separate pathway linked to cell identity in relation to development. In this context though, it is of interest to note that PHYB promoter activity in the dark shows a pattern similar to that of the PAP8 promoter (Somers & Quail, 1995).
PAP8 functions in plastids and in the nucleus. PAP8 displays a predicted chloroplastic transit peptide (cTP) and a predicted nuclear localization signal (NLS) , and therefore may belong to a group of bi-localized proteins (Krause, Oetke et al., 2012). Both signals are simultaneously functional since a translational fusion of PAP8-GFP ( Fig. 3A) displayed a bi-localized pattern in nucleus and plastids of transiently transfected onion cells (Fig. 3B). A polyclonal serum was raised against a recombinant protein PAP8 conceptually processed, starting after the predicted cleavage site. The specificity of the serum was validated in planta using the mutant pap8-1 as well as the recombinant protein (rP8, Fig. S5A). PAP8 is largely enriched in the sub-cellular fraction corresponding to sedimented organelles (mostly nuclei and plastids) obtained from 5-day-old Arabidopsis seedlings (Fig. S5B). PAP8 was then detected both in the nucleus and the plastid fractions obtained from seedlings either grown in the dark or under white light ( Fig. 3C) confirming the dual localization of PAP8. The distribution of PAP8 between the nucleus and the corresponding plastid-type fraction is changed after light exposure. In etiolated seedlings PAP8 was found mainly in the nucleus with traces in etioplasts (EP) while in photomorphogenic seedlings PAP8 was strongly enriched in chloroplasts (CP). Notably, both fractions (nuclei and plastids) displayed a signal of the same apparent molecular weight and similar to that of the designed ∆cTP recombinant protein suggesting that the nuclear fraction contains the processed version of the protein originating from plastids where the cleavage of the pre-sequence occurs during import.
To investigate this in more detail PAP8 localization was artificially uncoupled using a mutation strategy. Variants of PAP8-GFP lacking the cTP (∆cTP), the NLS (∆NLS), both signals (∆∆), or containing a mutated NLS with five neutral substitutions of the positively charged amino acids within the NLS (NLS m5 ) were cloned. In transiently transfected onion cells as well as in Arabidopsis thaliana lines with stable expression, PAP8 ∆cTP -GFP displayed nuclear accumulation (Fig. 3E, H, S6A, B) whereas the ∆NLS and the NLS m5 variants were strictly restricted to plastids ( Fig. 3J-K, 4A, S6A, C) indicating that the cTP supports chloroplast import and that nuclear localization depends on its NLS. Thus the PAP8 subcellular localization can be controlled in transgenic plants using the different targeting signals, and the corresponding transgene can therefore be assessed for functionality in pap8-1. Hence PAP8 variants fused or not to GFP, as indicated, were expressed under the constitutive promoter CaMV35S or its own promoter pP8 (pPAP8 -1133 ), in wild type or in pap8-1. In contrast to pP8::PAP8, all genetic constructions with the GFP tag were unable to yield functional complementation (Fig. S7). Since GFP may very likely impose a steric hindrance to the function of PAP8, only protein accumulation and subcellular localization were tested using the fluorescent marker. In addition, the greening of plants expressing part or full-length sequence of PAP8 under 35S promoter was strongly altered with no regard to its functionality or its proper localization (Fig. S7b), suggesting that over-expression or missexpression of the transgene with part of the PAP8 sequence might titter a component of unknown nature (protein, RNA, or else) that affects the biogenesis or stability of the chloroplast within the cell.
In wild type, the GFP signal of pP8::PAP8 ΔNLS -GFP increased in chloroplast sub-domains during the transition from dark to light (Fig. 3J, K). Therefore protein accumulation follows the promoter PAP8 induction in the palisade cells , and is consistent with the immune-detection of the native PAP8 in subcellular fractions. In contrast to the fluorescently tagged PAP10 that does not contain a predicted NLS and show a sharp and distinct localization (Fig. S6D), a wider signal of PAP8-GFP indicates that foci slowly appear after light exposure while part of the pool remains in the stroma. The foci, specifically marked with PAP10, may correspond to the assembly of the prokaryotic PEP core complexes with the eukaryotic PAPs. The stroma-localization of PAP8-GFP was confirmed by the PAP8-GFP signal transiently observed in stromules of onion cells while the PAP10-RFP signal is absent from these stromules ( Fig. S6E-H). Therefore, PAP8 may be set free from the PEP-PAP complex allowing for re-localization in the nucleus. Whether this release is allowed through saturation of the complex or a change in its affinity remains unknown.
Since GFP-tagged PAP8 could not rescue the mutant, untagged PAP8 variants were tested in functional hemi-complementation ( Fig. 4C-H). In contrast to the ΔcTP variant unable to cross the plastidial envelope and unable to rescue the albinism (Fig. 4C, H), the NLSm5 variant could restore the greening of the mutant albeit with strong delays in growth ( Fig. 4C- Fig. S8) suggesting that the chloroplast-localized PAP8 NLSm5 carries its chloroplast function for the greening but that in absence of the nucleus-localized pool the timing of chloroplast biogenesis is altered, with substantial consequences on the timing of lightcontrolled development. Therefore, PAP8, through its nuclear pool, may carry a function related to the light signalling response.
PAP8 mediates phytochrome signalling. To test this assumption, different light qualities were applied to the plants. Although, in our in vitro growing conditions, pap8-1 responded normally to red and white lights with proper de-etiolation with cotyledon and apical hook opening, far-red light treatment yielded a significantly reduced repression of hypocotyl length, a phenotype similar to that of hmr-2/pap5-2 (Chen et al., 2010) (Fig. S9).
Stable over-expression of a phytochrome PHYB-GFP (PBG) is known to mediate hypersensitivity of Arabidopsis seedlings to red light (8 to 30 µmol.m -2 .s -1 , Fig. 5A, B) leading to a significant inhibition of hypocotyl elongation when compared to WT (Yamaguchi et al., 1999). After introducing PBG into the pap8-1 mutant background, however, this PBG effect was largely lost, indicating that PAP8 plays a role in the PHYB-mediated light response. This lack of physiological response correlates with the retention of small PBG speckles in pap8-1 corresponding to the absence of late photobodies in comparison to wild type ( Fig. 5C-G).
The change in the photobodies patterning is not due to a change in PBG accumulation as tested by immune-detection of the GFP tag in the different genetic backgrounds (Fig. 5H).
Late photobodies are known to be associated with the targeted degradation of PIFs ( The light-induced destabilization of PIF1 and PIF3 is altered in pap8-1 and PBG/pap8-1, conversely the light-induced stabilization of HY5 does not occur in pap8-1 (Fig. 5J).
Interestingly, these molecular phenotypes in pap8-1 are very similar to those observed in pap5-2 used as control. Therefore, PAP8 supports the degradation of PIF1 and PIF3 and stabilizes HY5 with no effect caused by the presence of PBG. In the light signalling cascade PIFs are known to act upstream of HY5 and in a reciprocal negative feedback loop with PHYB (Leivar & Monte, 2014). This indicates that the alteration of the signalling in pap8-1 (the albino block depicted in Fig. 7) acts upstream of PIF1 and PIF3 by specifically blocking the HY5 to GLK pathway without altering the de-etiolating pathway: the apical hook and the cotyledons can open. The nature of the block remains unknown; it could be due to direct functional alteration of the PAP nuclear sub-complex in which PAP8 and PAP5 may act coordinately and dependently, or due to an upstream retrograde signal coming from the challenged pap8-deficient chloroplast (Martin, Leivar et al., 2016). Concerning the growth of the hypocotyl, the situation remains complex whether PBG is considered or not.
Therefore, PAP8 is important for the proper expression of GLKs. Should this occur through the nuclear function of PAP8, directly or through a PAP8-containing complex, this would simply explain the delayed greening and growth observed in the partially rescued phenotype of the PAP8 NLSm5 variant, in which nuclear PAP8 is absent. Should the expression of GLK1 be controlled by the state of the plastids through a distinct molecular pathway, this would then be an indirect consequence of the pap8-1 phenotype and more generally of the pap albino syndrome. Future research will probably help solving this conundrum.

PAP8 physically interacts with HMR/PAP5. The cellular distribution of PBG and other
defects in pap8-1, are highly similar to those of hmr-2/pap5-2. HMR/PAP5 is a nucleoplastidic protein identified to be (1) important for the initiation of photomorphogenesis (Chen et al., 2010, Qiu, Li et al., 2015, and (2) a component of the chloroplast PEP complex (Nevarez et al., 2017, Steiner et al., 2011. Although yeast-two-hybrid studies did not report any interaction between the two proteins (Arsova, Hoja et al., 2010, Gao, Yu et al., 2011, Yu, Lu et al., 2013, bimolecular fluorescence complementation technology (BiFC, Fig. 6A, S10 for control experiments) revealed that PAP8 ∆cTP and HMR/PAP5 ∆cTP could together restore split YFP fluorescence indicating that they get in close proximity within the nucleoplasm. HMR/PAP5 physically interacts with PAP8 through a well-structured region. Hence, these physical properties reinforce the assumption that PAP8 and PAP5 might form a nuclear complex. It is very likely that additional components could stabilize the unstructured region of PAP8 and enhance its affinity to PAP5 allowing BiFC detection in vivo. Should such a nuclear complex exist, the proteins could work cooperatively in an interdependent fashion.
Consequently, individual mutant phenotypes would resemble each other as it is observed for pap8-1 and pap5-2 reminding genetic epistasis where the lack of one gene is equivalent to the lack of the second gene.
Concluding remarks. This study revealed that PAP8 represents a novel regulatory component that links photomorphogenesis and chloroplast biogenesis through its dual localization. PAP8, therefore, is a novel member of the nucleo-plastidic protein family involved in chloroplast biogenesis (Yang et al., 2019. It is proposed that the nuclear fraction of PAP8 is essential to properly transduce the light signal from photoactivated PHYB to the expression of GLK1, one of the master regulators of nuclear photosynthesis genes. The amount of immunodetected PAP8 and PAP5 in dark-grown seedlings rises to nearly their maximum amount within 5 minutes following light exposure was also found on the chromatin associated with both GLK1 and GLK2 (Hajdu et al., 2018) and could therefore activate them directly or indirectly through other light responsive factors.
In turn GLKs, under GUN1-mediated retrograde signalling (Tokumaru, Adachi et al., 2017), activate the photosynthesis associated nuclear genes (PhANGs  (Porra, Thompson et al., 1989).   At 2 OD 600 (16h), the culture was centrifuged at 4000 g and the pellet was used to inoculate 1L of M9 + antibiotics. Culture, induction and purification were done as described for PAP8. Organelles fractionation. 5-day-old Arabidopsis seedlings, exposed to light or dark, were homogenized in liquid N 2 . The powder was dissolved in a cold native extraction buffer (NEB:

MBP
Tris HCl 100 mM pH 7.4, glycerol 25%, KCl 20 mM, EDTA 2 mM, MgCl 2 2.5 mM, Sucrose 250 mM, DTT 5 mM, protease inhibitor Roche TM , 1 tablet/ 50 mL) at a ratio of 1:3 (w/v). The extract was filtered through 3 layers of miracloth and one layer of nylon (100 µm) centrifuged (10 min at 1500 g, 4°C). The supernatant was deposited on percoll 80% and centrifuged (swinging rotor, 5 min at 2,300 g, 4°C) to remove the pellet of starch. The supernatant was loaded on 35% percoll and centrifuged (swinging rotor, 5 min at 2,300 g, 4°C) to separate swimming plastids from the pellet of nuclei. The nuclei were washed two times with a plastidlysis buffer (NEB + 2% triton) followed with centrifugation (5 min at 1500 g, 4°C). Fractions, corresponding to plastids or nuclei, were suspended in DEB shaken on a vortex (10 min at 4°C) before centrifugation (10 min at 9,300 g, 4°C). The TSP samples were subjected to western blot analysis as above. the bench. We thank E Thevenon in the Parcy lab for advice on fluorescent labelling EMSA.
We thank S Lerbs-Mache for critical reading. We express our gratitude in memory of D The authors declare that they have no conflict of interest. I Content of total chlorophylls (Chl(a+b)) normalized to fresh weight and relative to wild type in the given genotypes grown in the dark (D) or grown in the dark followed with 30 hours of white light treatment (+L); n.a. not applicable. sequencing data 30 at the PAP8 locus; TAIR10, annotation according to the Arabidopsis thaliana information resource. ChIP on hy5-ks50; 35S:HY5-YFP exposed to blue light or red light using GFP antibody and compared to mock corresponding to ChIP control experiment done without antibody. Each treatment is presented as track overlay of triplicates: the read count is given within the "group autoscale" range in brackets. Close up on the 5'-UTR region centred on the -95-promoter element in yellow.   for PBG/pap8-1 were tested. Coomassie blue staining presented as loading; signals were quantified using ImageJ. I RT-qPCR analysis on wild type, pap8-1, PBG and PBG pap8-1.

Figure legends
Seedlings were grown in the dark (D) or under white light (L, 30 µmol.m -2 .s -1 ); levels of transcripts are given relative to EF1α; error bars correspond to standard errors on technical triplicates and the dark sample is the wild-type PBG line. J Immuno-blots showing the levels of PIF1, PIF3, HY5 in given genotypes: p5/+, mix of an heterozygous pap5-2 siblings progeny undistinguishable from wild type; p5-2, pap5-2 and pifq, quadruple pif1-1 pif3-3 pif4-2 pif5-3 mutant; Histone H3 (H3) RbcL and PAP8 were used as controls; n.a., not applicable.   Col-0, wild type; p8/p8, homozygous mutant pap8-1; L35 and L49, two lines of pap8-1/pP8::PAP8; n.a., not applicable. Absorbance was normalized to fresh weight (FW); Chla, chlorophyll a; Chlb, chlorophyll b; Car, carotenoids. Fig. 1 and 3) Flow chart of the pap8-1 functional complementation test. p8-1, pap8-1 allele associated with nptII, neomycin phosphotransferase II marker used to select plants resistant to kanamycin (Kan R ) although this resistance is partially lost in pap8-1. A Heterozygosity test on the progeny of one plant, if 1/4 of albino plants appear then the tested seeds were used for floral dip with an allelic frequency of 1/3 for pap8-1, Agrobacterium-mediated transformation; hptII, hygromycin phosphotransferase II gene to select resistant plants (Hyg R ) from those that are untransformed and sensitive (Hyg S ). Selection under low red light (660 nm at 8 µmol.m -2 .s -1 ) allowing for rapid elongation of Hyg R plants. The etiolating response cannot be used with PBG that causes a strong de-etiolated phenotype. Primary transformants (T1) were PCR-selected for the presence of the tested goi (gene of interest), presence of the pap8-1 allele, and the wild-type allele. The yellow scenario represents a successful complementation test in T1 (FC); the albino plant in brackets may be retrieved in some tests at a low ratio (1/6 of all T1) as negative for the complementation test (noC). The white box represents a common event of interest (1/3 of all T1 carrying the goi and one allele pap8-1). Conclusion is made after testing the genetics in the T2 generation. B, C Heterozygosity test to retrieve doubly homozygous (C) complemented T3 line with the pPAP8::PAP8 transgene (p8P8).  Table of the occurrences (occ.) of binding sites detected for the different transcription factors (TF) families using PlantPan3 26 . The promoter was broken down to two regions (-497 to -97) and (-97 to +1); C Selected binding sites represented on the (-497 to +63) PAP8 promoter sequence, >, plus strand; <, minus strand; <-->, element detected on both strands; the yellow box depicts the near palindromic bZIP element found at -97; annotations for individual elements are given according to Table S2. Fig. 2) Transactivation test in onion epidermal cells using the dual luciferase reporter assay. TF, transcription factor; Luc, luciferase; glow is the emission of photons after the protein samples have been supplemented with the luciferase-type specific substrate. Activity of the tested transcription factor calculated as the relative value of Firefly glow / Renilla glow to that of the reference sample. Fig. 3) A Immunoblots showing the levels of PAP8 and PAP5 in Col0, pap8-1 (p8-1) or pap5-2 (p5-2) respectively, using the recombinant proteins PAP8 (rP8) and PAP5 (rP5) as controls. Proteins on PAGE were detected by instant blue to display equal loading. B PAP8 immunodetection in fractions obtained during organelles enrichment; rP8, as above; S1, soluble fraction in the supernatant after centrifugation of the blender-disrupted cellular sample; Tot, total plant extract; C+N, pellet containing organelles, mostly chloroplasts and nuclei separated from S1; H3 antibody used to evaluate nuclear enrichment and a Coomassie staining presented as loading control.   (see table S1). B Transgenic lines obtained with the constructions described above. Three phenotypic classes have been recorded corresponding to albino (white squares) pale green (light green squares) and green plants (green squares). Col-0, wild type; pap8-1/+, heterozygous mixture; n, total number of recorded plants. C Functional complementation output. Hygromycin resistant plants were transferred on soil and grown under long day conditions (16 h light / 8 h dark; ~70 µmol.m -2 .s -1 ) at 21°C and 60 % humidity. Genomic DNA was isolated from true leaves and used for genotyping. The presence of the pap8-1 allele was confirmed using the primer ortpF/oLBb1.3. PAP8 wild-type allele tested with ortpF/op8i2_R. The insertion of the transgene of interest was tested with ortpF/oE3R. The number of the tested plants (Hyg R #T1), the number of double heterozygous plants (p8-1/+;TG/-) and the number of sesqui-mutant plants (p8-1/p8-1; TG/-) are depicted for each construction; p8P8 presented as positive control. None of the 39 T1 plants with pP8::PAP8-GFP, were photosynthetic and homozygous pap8-1. Therefore, two doubly heterozygous (pap8-1/+; TG/-) expressing GFP were tested for their segregation pattern (Line 1: 34% albino, n=169 and Line 2: 24% albino, n=199) and compared to that of pap8-1/+ (28% albinos, n=99). In absence of statistical difference between the samples (ε = 0.102<<1.96 for α=0.05; ε-test, Fisher Yates), PAP8-GFP was declared not functional as opposed to PAP8. Fig. 4) Hemi-complementation test in pap8-1. Phenotype of pap8-1 transformed with pPAP8::PAP8 NLSm5 . WT, 5-week-old Col-0 control; [DG], Delayed Greening phenotype observed for the partial rescue of pap8-1 mutant expressing PAP8 NLSm5 under its endogenous promoter; pictures depict three 15-weekold plants in which the alteration of the greening corresponds to the emergence of white leaves that slowly acquire the photosynthetic apparatus; plants #7, #31 and #57 are siblings of the same genotype (Ai10#34). Bars equal 10 mm.