Lil3 Assembles with Proteins Regulating Chlorophyll Synthesis in Barley

The light-harvesting-like (LIL) proteins are a family of membrane proteins that share a chlorophyll a/b-binding motif with the major light-harvesting antenna proteins of oxygenic photoautotrophs. LIL proteins have been associated with the regulation of tetrapyrrol biosynthesis, and plant responses to light-stress. Here, it was found in a native PAGE approach that chlorophyllide, and chlorophyllide plus geranylgeraniolpyrophosphate trigger assembly of Lil3 in three chlorine binding fluorescent protein bands, termed F1, F2, and F3. It is shown that light and chlorophyllide trigger accumulation of protochlorophyllide-oxidoreductase, and chlorophyll synthase in band F3. Chlorophyllide and chlorophyll esterified to geranylgeraniol were identified as basis of fluorescence recorded from band F3. A direct interaction between Lil3, CHS and POR was confirmed in a split ubiquitin assay. In the presence of light or chlorophyllide, geranylgeraniolpyrophosphate was shown to trigger a loss of the F3 band and accumulation of Lil3 and geranylgeranyl reductase in F1 and F2. No direct interaction between Lil3 and geranylgeraniolreductase was identified in a split ubiquitin assay; however, accumulation of chlorophyll esterified to phytol in F1 and F2 corroborated the enzymes assembly. Chlorophyll esterified to phytol and the reaction center protein psbD of photosystem II were identified to accumulate together with psb29, and APX in the fluorescent band F2. Data show that Lil3 assembles with proteins regulating chlorophyll synthesis in etioplasts from barley (Hordeum vulgare L.).

In photosynthesis, a family of light harvesting complex (LHC) proteins regulates harvesting of light energy and photoprotection. Members of the LHC protein superfamily include LHC proteins, LHC-like proteins, subunit S of photosystem II (PsbS), ferrochelatase II and red lineage chlorophyll a/b-binding (CAB)-like proteins (RedCAP) [9,10]. All LHC-proteins are nuclear-encoded, bind pigments, such as chlorophyll (Chl) and carotenoids, have one to four membrane-spanning helices, are involved in light absorption and are characterized by at least one LHC motif [9].
The two-helix light-harvesting-like protein 3 (Lil3) has been associated with Chl and tocopherol synthesis in thale cress (Arabidopsis thaliana) [11]. In a lil3:1 and lil3:2 double mutant, Chl yield was low and Chl was exclusively found esterified to geranylgeraniol (GG) (Chl GG ). Decrease in α-tocopherol, and loss of the enzyme geranylgeranyl reductase (GGR, EC1.3.1.83) led to the conclusion that Lil3 stabilizes GGR against degradation [11]. Recently, transmembrane amino acids in the LHC-motif of Lil3 were reported to anchor GGR to the membrane, and to be responsible for oligomerization of GGR [12]. However, recombinant plant GGR did not require Lil3 for reduction of GGPP to PYPP [12]. GGR catalyzes the NADPH dependent three-step reduction of the pyrophosphate (PP) form of GG or its esterified form in Chl GG . End products are phytyl pyrophosphate (PYPP) or Chl PY [7]. The enzyme operates at the branching point between Vitamin E and Chl synthesis. In vitamin E biosynthesis, PYPP and homogentisate are the substrates for homogentisate phytyltransferase (HPT1, AT2G18950) and hence for tocopherol biosynthesis [13][14][15]. In Chl biosynthesis, Chl PY provides the entry point for assembly of the Chl binding protein complexes of the photosynthetic machinery.
In Synechocystis PCC6803, a complex of Chl-synthase (CHS) and the high-light-inducible protein HliD, a one-helix Lil homolog, together with Ycf39 and YidC was characterized [16]. HliD was not found to be essential for the function of CHS. However, deletion of HliD increased the Chlorophyllide (Chlide) content in Synechocystis about 6-fold, and authors concluded that HliD could have a function in directing the assembly of other proteins like Ycf39 into a complex with CHS [16]. In barley, Lil3 was identified as the first protein to bind Chl during deetiolation [17].
Here, the Chlide and GGPP dependent assembly of Lil3 was investigated in etioplasts from barley. Protein and pigment analysis show that Chlide triggers Lil3 assembly with POR and CHS and that Chlide esterification initiates GGR and photosystem assembly.

Assembly of Lil3 correlates with the accumulation of Chlide and ChlGG
Separation of protein complexes from etioplast membranes by native LN-PAGE reveals two fluorescent protein bands and a high fluorescence emission at the gel front (Fig 1, Fluorescence). The two fluorescent protein bands had earlier been identified as the low molecular weight monomeric (Cyt b 6 f (1)) and the high-molecular weight dimeric (Cyt b 6 f (2)) form of the Cyt b 6 f complex [18]. In addition Protochlorophyll was identified in the dimeric Cyt b 6 f (2) band [18]. In 4.5 day-old dark-grown etiolated seedlings, a high concentration of POR, and 80-120 pmol of phototransformable Pchlide had been determined per 10 7 etioplasts [19]. Interestingly, no Pchlide could be extracted from the fluorescent bands in this work; however, the extraction of Pchlide at the gel front indicated that Pchlide was released from POR during solubilization of the membrane and native PAGE analysis. Illumination of etioplasts and photoreduction of Pchlide resulted in a marked increase in the fluorescence intensity at about the mobility of the Cyt b 6 f (1) band (Fig 1, Fluorescence, F3). Also, two bands of higher molecular weight, but very low fluorescence intensity were detectable (Fig 1, Fluorescence, F1 and F2).
Data indicated that the photoreduction of Pchlide resulted in a more stable interaction of Chlide with POR or that additional proteins assembled the newly synthesized Chlide.
In etioplasts isolated from etiolated barley seedlings, Chlide is only effectively esterified to yield Chl when GGPP is supplemented [19]. When GGPP was added, Chlide synthesis and esterification resulted in a strong fluorescent increase in the two higher molecular mass bands, while the Chlide specific fluorescent band was maintained albeit at a reduced intensity (Fig 1, Fluorescence, F1, F2 and F3). This indicated that Chlide triggers accumulation of the F3 band, and that GGPP dependent esterification of Chlide may trigger accumulation of F1 and F2. Whether light or Chlide triggered assembly of the F3 band was tested without GGPP supplementation by addition of Zn-pheophorbide (Zn-pheide) to etioplasts maintained in darkness (Fig 1, Fluorescence, lane 4 and 5). Accumulation of fluorescence in F3 was triggered by Znpheide alone, whereas accumulation of fluorescence in F1 and F2 required the presence of Znpheide and GGPP. Interestingly, buildup of the F1 and F2 bands correlated with the loss of F3, and most importantly, the correlated reactions did not require a preceding phototransformation. Instead, the presence of Chlide and the esterification of Chlide in darkness were sufficient to trigger the assembly state changes. Data therefore excluded that the increase of fluorescence in the F3 band was based on a binding of Chlide by POR. It was therefore investigated whether other proteins could bind Chlide in the F3 band.
A promising candidate for binding the de novo synthesized chlorine molecules was identified by comparison of fluorescence with gel-blot data of the native gel. A peptide specific antibody against the Lil3 protein (S1 Fig) revealed a positive correlation between the signal strength of the Lil3 antibody and the Chlide (F3) and GGPP dependent fluorescence (F1, and F2) (Fig 1, Fluorescence and Lil3 gel-blot). In addition, four Lil3 containing bands were identified in the gel-blot analysis that did not show a corresponding fluorescence in the native gel (Fig 1, Lil3 gel-blot, NF1, NF2, NF3 and NF4). Hence, Lil3 protein assembled in higher molecular weight assembly states already in the absence of chlorines and the assembly state of Lil3 containing bands was changed in the presence of Chlide and GGPP. Lil3 had earlier been shown to bind Chl during deetiolation [17]; however, the structural and functional context for binding Chlide and Chl was unclear. It was therefore investigated whether the enzymes POR, CHS, and GGR interact with Lil3 in the different fluorescent bands and how Lil3 may contribute to a binding of Chlide and Chl.

POR and CHS are potential interaction partners of Lil3
The protein composition of the fluorescent bands was investigated further by gel-blot analysis upon 2D LN-/SDS-PAGE. Etioplast membranes were isolated from etiolated plants after illumination for 10s. Protein complexes solubilized from the membranes were separated by LN-PAGE and protein subunits in native bands were resolved by SDS-PAGE. The gel-blot of the second dimension gel was sequentially probed with polyclonal antibodies against Lil3, POR, CHS, and Cyt b 6 (Fig 2 and S2 Fig). Lil3 was identified at about 25 kDa in spots corresponding to native bands F1, F2, F3. In addition, Lil3 was identified in two spots within the molecular weight range of micelles containing free pigment (Fig 2, Lil3). CHS was detected at about 30 kDa (Fig 2, CHS) with highest concentration of the protein corresponding to the two Lil3 spots at the edge of the free pigment and an extension of the signal especially towards the native F3 band (Fig 2, CHS). POR was identified at about 40 kDa (Fig 2, POR). The mobility of POR during native PAGE correlated well with CHS, and with the Lil3 subunits of the native Lil3 bands F1, F2, and F3 and with the two Lil3 bands that were identified at the pigment front of the native gel (Fig 2).
Analysis of the Cyt b 6 subunit identified the two mobility states of the Cyt b 6 f complex in native PAGE corroborating a dimeric and monomeric assembly state (Fig 2, Cyt b 6 f). Fluorescence analysis after LN-PAGE and gel-blot analysis of the Cyt b 6 subunit showed furthermore that the mobility of the Cytb 6 f dimer was distinct from the native F1 band (Fig 2, Cyt b 6 f). The Cytb 6 f monomer showed some mobility overlap with a Lil3 signal intermediate between the F2 and F3 bands (Fig 2, Cyt b 6 f). The overlapping mobility of protein subunits Lil3, CHS, and POR indicated an interaction of the three proteins in the fluorescent native PAGE bands; however, it remained unclear whether additional proteins were assembled in the fluorescent bands. Overlapping mobility of proteins in the 1D gel has been connected by a vertical line. A peptide specific antibody against the C-terminus of Lil3 was employed (S1 File). Therefore, the gel region of bands F1, F2, and F3 was analyzed by mass spectrometry before and after induction by Chlide and GGPP.

GGR accumulation in bands F1 and F2 is light induced
A comparative MS experiment was set up to reveal general changes in the protein composition of fluorescent bands between the dark, the Chlide, and the GGPP induced state. Membranes were solubilized and protein complexes separated by native PAGE before and after induction of Lil3 assembly. Fluorescent bands and the corresponding non-fluorescent gel regions in the non-induced states were identified and gel regions excised (S3 Fig). The total peptide count was determined for all identified proteins, and proteins with a peptide cut-off value higher than ten were selected for hierarchial cluster analysis (Fig 3 and S1 Table) [20,21]. Changes in the protein composition between the induced and the non-induced states and the ratio of protein changes were determined to selectively show which proteins were associated with an increase of fluorescence in the F1 and F2 band relative to a decrease in the F3 band (Fig 3). The band containing the dimeric Cyt b 6 f was used as a control. MS analysis revealed a selective 6-fold light-induced increase of GGR in the F1 and a 3-fold increase in the F2 band, relative to F3 (Fig 3). This showed that GGPP triggered the assembly of GGR in the F1 and F2 bands. In contrast, GGR levels were found constant in the gel region of the F3 band in the absence and presence of Chlide and the enzyme was not detected in the molecular weight region of the Cyt b 6 f complex.
Two protein subunits of photosystem II (PSII) also showed a remarkable increase in bands F1 and F2. A 3-fold induction was found for Psb29 in the F1 band and a 9-fold induction was found for both PsbD and Psb29 in the F2 band. Both proteins clustered with thylakoid based ascorbate peroxidase (APX) (Fig 3, Cluster 5).
In the dimeric Cyt b 6 f band, the comparative MS analysis revealed a 4-fold increase in the plastid encoded NAD(P)H plastoquinone oxidoreductase subunit 5 (NDHF) [22] (Fig 3). However, protein clustering was less stringent for the NDH subunit as compared to the identification of Cyt b 6 f subunits PetA, PetB and PetC in cluster 11 (Fig 3). The protein clusters identified in the native Lil3 and Cyt b 6 f bands reflected complexes for Chl synthesis (Fig 3, clusters  1-3), photosystem assembly (clusters 4-8), ATP synthesis (cluster 9), and electron transport via the Cyt b 6 f (cluster 11) and the NDH complex (cluster 12). Proteins FTSH8, PSBQlike, NDHF, RPL9, PRQX, and GGR did not cluster stringently (Fig 3). The experimentally induced changes in the GGR levels of fluorescent bands F1 and F2 and the gel-blot based indications for interaction between Lil3 and the pigment binding enzymes CHS and POR encouraged us to investigate a direct interaction between the Lil3 protein and CHS, POR, and GGR in a split ubiquitin system and to investigate the pigments bound to the Lil3 protein bands.

Lil3 and CHS interact in a split ubiquitin Y2H screen
The direct interaction of Lil3 with the potential binding partners CHS, POR, and GGR was investigated by the split ubiquitin system in yeast. Expression of protein pairs resulting in growth of yeast (+) under suppressing growth conditions was plotted (Fig 4 and S2 Table). Lil3:1 and Lil3:2 were fused to the N-terminus of the Cub moiety protein and to the C-terminus of the NubG moiety protein to generate BTC-lil3:1, BTC-lil3:2, PRN-lil3:1 and PRN-lil3:2, respectively. CHS, PORA, PORB, PORC and GGR were fused to the C-terminus of the NubG moiety protein to generate PRN-CHS, PRN-PORA, PRN-PORB, PRN-PORC and PRN-GGR. Each of the PRN constructs was co-expressed in yeast NMY51 cells with BTC-lil3:1 and BTC-lil3:2. The ability of yeast NMY51 cells to only grow in the absence of histidine by direct protein-protein interaction between the two proteins encoded by the candidate genes was investigated. Among all tested combinations, Lil3:1 and Lil3:2 were found to directly interact with CHS ( Fig 4A). Also, a direct interaction between Lil3 and POR was identified (Fig 4A). No interaction was identified between Lil3 and GGR. The lack of interaction between Lil3 and GGR could be due to the orientation of the genes in the expression vector. Therefore, GGR and CHS were fused to the N-terminus of the Cub moiety protein to generate BTC-GGR and BTC-CHS, respectively. Interactions between CHS and Lil3:1 and Lil3:2 remained positive. CHS also showed an interaction with itself ( Fig 4B). GGR was found to interact with POR, CHS, and with itself ( Fig 4B). However, CHS did not directly interact with any of the POR proteins or GGR. No direct interaction between GGR and Lil3 could be identified (Fig 4B). An interaction of Lil3 with the two enzymes CHS and POR indicated that Lil3 participates in the regulation of Chlide esterification. It was therefore investigated next whether Lil3 binds Chlide.
In fluorescent band F3, Chlide, Chl GG and Chl PY accumulate In order to determine whether Lil3 binds Chlide in the F3 band, the changes in pigment composition were investigated in the plastid membrane and in the fluorescent F1, F2, and F3 bands. In order to study Chlide induced changes in the absence of downstream enzymatic processes, etioplasts were frozen on dry ice before illumination. Under these conditions phototransformation of Pchlide is possible, but the reaction speed of subsequent enzymatic steps is decreased [23] (Fig 5). Esterification was controlled by supplementation of etioplasts with GGPP before or after phototransformation. For analysis of pigment accumulation in plastids, frozen plastids were melted into acetone. For analysis of pigments bound to the fluorescent protein bands, membranes were solubilized in darkness by thawing the frozen etioplasts in a solubilization buffer on ice. Pigments were extracted from the gel-bands after native PAGE and separated by HPTLC. Tetrapyrrol derivatives were analyzed by fluorescence scanning (Em 670 BP30, Ex 633 nm) (Fig 5).
The fluorescence emission of Pchl associated with the Cyt b 6 f complex was found unchanged under all conditions (Fig 5A). In addition, no changes in the relative amount of Pchl-GG,-DHGG,-THGG and Pchl-PY were identified after separation of etioplast membrane extracts (Fig 5B, lanes 1-3, Pchl). Illumination of the frozen plastids in the absence and presence of exogenously added GGPP clearly showed that the Lil3 band F3 increased selectively ( Fig 5A, lane 3, F3). The corresponding pigment extract from the membranes showed the selective increase in the Chlide signal and indicated that the fluorescence in the native band F3 originated from Chlide binding (Fig 5B, lanes 2 and 3). In the presence of GGPP supplementation the amount of Chlide also increased and in addition, a band corresponding to Chl GG was identified ( Fig 5B, lane 3). Supplementation of the etioplasts with GGPP and NADPH increased the yield of the Chl GG band and in addition, a band corresponding to the mobility of Chl PY was detected (Fig 5B, lane 4). The identity of Chl GG in the presence of GGPP supplementation was verified since two more GGPP dependent fluorescence bands were detected (Fig 5B, lanes 4 Chl GG Ã , and Ã ). The identity of the Chl GG and Chl PY bands was corroborated by exchange of the tetraterpenoid GGPP against the triterpenoid farnesyl-pyrophosphate (FPP). In the presence of FPP, Chl F was formed and the Chl GG signal and one of the non-polar GGPP dependent signals (Chl GG Ã ) were lost (Fig 5B, lanes 4 and 5, Chl F , Chl GG Ã , and Ã ). Finally, the position of Chl PY on the HPTLC plates was verified by separation of etioplast extracts isolated from etiolated plants illuminated for 10s in vivo (Fig 5, lane 6, Chl PY ). Data corroborated that in isolated etioplasts no GGPP was accessible for esterification of the de novo synthesized Chlide (Fig 5,  lane 2). Also, hardly any Chl GG and GGPP dependent signals were identified when plants were illuminated in vivo (Fig 5, lane 6). This indicates that the concentration of GGPP available for esterification in etioplasts is low in vivo.
When pigments were extracted from the F3 band after illumination of etioplasts, Chlide and Chl GG were identified (Fig 5C, lane 1). This corroborated that POR and CHS were enzymatically active and indicated that pigments in F3 were bound to the enzymes or to Lil3. When pigments were extracted from the three fluorescent protein bands F1, F2, and F3 upon a 10s illumination of the etiolated plant, Chl PY was identified in all fluorescent bands ( Fig 5C, Lane  2, 3 and 4). This corroborated that all three Lil3 containing bands participated in the synthesis of Chl PY .

Discussion
Four levels of evidence indicate that Lil3 directly assembles with proteins regulating Chl biosynthesis (Fig 6). The electrophoretic mobility of Lil3 is decreased when etioplasts are supplemented with Chlide (Fig 1). In the absence of GGPP the electrophoretic mobility of Lil3 correlates with the mobility of POR and CHS in native PAGE (Fig 2). Chlide, Chl GG , and Chl PY are identified in the F3 band ( Fig 5) and Lil3 directly interacts with CHS and POR in a split ubiquitin screen (Fig 4). It is therefore concluded that Lil3 interacts with the two enzymes POR and CHS during synthesis of Chl PY in barley etioplasts.
In the presence of GGPP, Lil3 associates with the two fluorescent protein bands F1 and F2. In both bands, Chl is isolated in the phytylated form (Fig 5). This indicates that Lil3 participates in esterification of Chlide by CHS. It remained open, whether Chlide and Chl GG are bound to Lil3 or to CHS. Lil3 binding of Chlide in the fluorescent band F3 could mediate exposure of the carbonyl group thereby allowing esterification by CHS. The lack of Chl GG or Chl PY accumulation in isolated etioplasts after illumination, and the sole accumulation of phytylated Chl in the F1, F2, and F3 bands in vivo (Fig 5) indicated that GGR provides CHS with PYPP in etiolated plants in vivo. Lil3 binding of Chl GG after in vitro supplementation with GGPP and of Chl PY in vivo could have a function in exposing the esterified chlorine for photosystem assembly.
In Synechocystis PCC6803, the one-helix Lil-homolog HliD, Ycf39, YidC, and CHS were shown to interact [16], and a coregulation of Chl synthesis and membrane integration of photosystem proteins was concluded from a copurification of CHS and unassembled D1 with a Ycf39-Hlip complex [24]. Interestingly, GGR was not isolated with the complexes regulating Chl synthesis in Synechocystis PCC6803 [16,24]. The enzyme also did not cluster with any of the proteins identified in the fluorescent Lil3 bands in our bioinformatic analysis of the MS datasets (Fig 3). The GGPP dependent accumulation of GGR in the fluorescent bands F1 and F2 may result from assembly with proteins that participate in the regulation of GGPP reduction but that were not identified in the MS approach or were excluded from the hierarchial cluster analysis due to the selected cut-off value (Fig 3). Lil3 was identified in the F1 and F3 bands, but count numbers were below the cut-off value and data were excluded from further analysis. Loss of GGR has been reported as an explanation for the loss of Chl PY and tocopherol in a Lil3 double knockout line of Arabidopsis thaliana [11]. However, the low-growth phenotype and low yield of Chl GG remained unexplained [11]. GGR mutants (ΔchlP) in Synechocystis PCC6803 demonstrated that accumulation of Chl GG and growth of mutants were comparable to WT in low light [25]. According to our model, the absence of Lil3 may lead to a low yield of Chlide esterification and hereby limit autotrophic growth. The loss of GGR in the mutant lines may be explained by the loss of the direct interaction between Lil3 and CHS as shown here. This interaction may provide the structural basis to stabilize GGR against degradation in Arabidopsis thaliana.  (Figs 1 and 3) and F1, F2, and F3 bind Chl PY (Fig 5, grey background). The F2 band accumulates psb29 and D2 indicating that Lil3 participates in the delivery of Chl PY (left arrow). POR, GGR, and Lil3 are detected in all fluorescent complexes F1, F2, and F3 by mass spectrometry or gel-blot analysis (dotted boxes). The presence of CHS, and GGR in fluorescent bands is indicated by accumulation of Chl GG and Chl PY . Lil3 is shown to directly interact with CHS, and POR in a split ubiquitin based Y2H screen (Fig 4). No direct interaction between GGR and Lil3 or POR and CHS was identified.

Interaction of Lil3 with CHS
Characterization of a complex containing CHS and the high-light-inducible Lil homolog HliD in Synechocystis PCC6803 led to the conclusion that HliD is a non-essential component for the function of CHS [16]. Since deletion of HliD increased the Chlide content 6-fold in the deletion strain, the authors concluded that HliD could be responsible for directing the assembly of proteins like Ycf39 into a complex with CHS [16]. As shown here, Lil3 assembles with POR and CHS in barley etioplasts. Phototransformation of etioplasts at low temperature results in accumulation of Chlide and Chl GG in the Lil3 containing F3 band (Fig 5). In the presence of GGPP in organello, or illumination of plants in vivo, the F3 band is shifted into band F1 and F2 in which GGR and Chl PY accumulate (Figs 1, 3 and 5). This indicates that a complex composed of POR, Lil3, and CHS precedes an assembly of increasing amounts of GGR (Figs 5 and 6).
In the split ubiquitin screen, Lil3 shows direct interaction with CHS and with POR. No direct interaction was found between Lil3 and GGR and between POR and CHS. Further, CHS and also POR were found to directly interact with GGR (Figs 4 and 6). These data indicate that Lil3 and GGR co-assemble in etioplasts with POR and CHS in a substrate dependent manner without interacting directly (Figs 5 and 6). The transfer of Chlide from POR to CHS could require transient binding of Chlide to Lil3 for optimized substrate interaction with CHS and fast reloading of POR. High concentrations of POR accumulate at the membrane periphery in higher plant etioplasts [26]. In order to reload POR, Chlide removal and translocation from POR to the membrane integral CHS is required for rapid accumulation of Chl PY [27][28][29].
Interestingly, the rate of Chl accumulation is drastically increased in illuminated etioplasts supplemented with GGPP relative to illuminated plants [30]. Our finding that the availability of GGPP both limits Chl PY accumulation and assembly of Lil3-bands F1 and F2 from the Chlide specific Lil3-F3 band, therefore emphasizes the core position of the membrane integral Lil3 to functionally link the three enzymes. Lil3 directly interacts with CHS and POR and may indirectly interact with GGR via CHS and POR (Fig 6). A lack of direct interaction between Lil3 and GGR has recently also been discussed in rice mutants deficient in GGR [31]. Therefore GGR could interact with Lil3 via CHS. The LHC motif in Lil3 could provide the basis for assembly of Lil3 with CHS. The LHC motif binds Chl in the light-harvesting complex (LHC) [32]. The LHC motif in Lil3 could also function for binding Chlide and Chl. We show direct interaction of CHS with both Lil3 and GGR (Fig 4). Chlorine binding to Lil3 could therefore enable the assembly of a ternary Lil3-Chlide/Chl-CHS complex that anchors GGR to the thylakoid membrane for efficient esterification of Chlide with PYPP (Fig 6).

Structural changes induced by prenylation of Chlide
A comparative MS analysis of the GGPP relative to the Chlide induced changes, revealed a 6and 3-fold normalized change for GGR in native gel bands F1 and F2, respectively. Furthermore, PSII components PsbD and Psb29 increased 9-fold in Lil3-F2 relative to Lil3-F3 (Fig 3). PsbD is one of two reaction center proteins of PSII and present in the first assembly intermediate during reaction center assembly [33][34][35]. The psb29 protein was conserved throughout evolution of oxygenic photosynthetic organisms. A plant psb29 mutant showed slow growth, variegated leaves, and light sensitive PSII activity associated with an uncoupling of the proximal antennae indicating a function of psb29 in the biogenesis of PSII [36]. Our finding indicates that accumulation of Chl PY in the F2 band reflects an assembly of Lil3 with single proteins or with a complex composed of at least psbD, and psb29 (Fig 3).

Mass spectrometry analysis
For identification of peptides from LN-PAGE bands A-H, digests were analyzed in an Orbitrap Velos mass analyzer, at a resolution of 30000. Data dependent analysis (Top 20) was employed and ions with charge states of 2+ and above were selected for fragmentation. Data was processed using Protein Discoverer (version 1.2.ThermoFisher) and the Mascot search algorithm (Matrix Science, London UK) and were searched against the TAIR10 and Uniprot Wheat database on Mar. 14, and Jan.13, respectively, using a fixed modification of carbamidomethyl (C) and a variable modification of oxidation (M).

Peptide data processing
Mascot search results were exported in CSV format and the number of peptide-spectrum matches for each identified protein was determined per LN-PAGE band and variable peptide count numbers per protein compiled in a matrix. Total count number was determined per peptide. All values were standardized and clustered row by row, and column by column using complete linkage in Hierarchical Cluster Explorer 3.5 (HCE3.5) (http://www.cs.umd.edu/hcil/hce/) [20,21]. A Euclidean distance matrix was calculated for all proteins identified by a minimum of 10 counts. Clustering results were analyzed in the form of a heat map and two dendograms, showing clustering of proteins (rows), and bands (columns). Standardized and clustered datasets were analyzed using OriginLab 9.0 (OriginLab Corp., Northampton, USA) and data normalized (0/1) per row. For annotation of protein assembly, and annotation of untagged proteins, dendrograms were cut at an euclidean distance of 0.6. Normalized peptide differences (Chlide induced minus dark state versus Chl induced minus dark state) and ratios (Chlide induced/Dark state versus GGPP induced/Light state) of corresponding bands were determined.
Supporting Information S1 File. Verification of Lil3 antibody specificity. (DOCX) S1 Fig. Specificity of an antibody directed against a Lil3 peptide. Etioplasts were isolated from etiolated barley seedlings illuminated for 10 s with white light. Solubilized membrane extracts corresponding to 1x10 8 plastids were used. For immunoprecipitation, an antibody directed against the Lil3 peptide QSTWQDDSTSGPKK (C, bold) was employed. Protein precipitates were concentrated by centrifugation and denatured proteins separated by SDS-PAGE (A and B). Proteins were detected by scanning for Cy2 labeling (A) and for secondary antibody specific luminescence in gel blots (B). Solubilized membrane extracts were incubated with Lil3 antibody, Sepharose and detergent (lane 2). In the controls, immunoprecipitation was conducted in assays containing Sepharose, Lil3 antibody, and detergent, but no solubilized membrane (lane 1), Sepharose, solubilized membrane extracts corresponding to 1x10 8