Mutation of the Light-Induced Yellow Leaf 1 Gene, Which Encodes a Geranylgeranyl Reductase, Affects Chlorophyll Biosynthesis and Light Sensitivity in Rice

Chlorophylls (Chls) are crucial for capturing light energy for photosynthesis. Although several genes responsible for Chl biosynthesis were characterized in rice (Oryza sativa), the genetic properties of the hydrogenating enzyme involved in the final step of Chl synthesis remain unknown. In this study, we characterized a rice light-induced yellow leaf 1-1 (lyl1-1) mutant that is hypersensitive to high-light and defective in the Chl synthesis. Light-shading experiment suggested that the yellowing of lyl1-1 is light-induced. Map-based cloning of LYL1 revealed that it encodes a geranylgeranyl reductase. The mutation of LYL1 led to the majority of Chl molecules are conjugated with an unsaturated geranylgeraniol side chain. LYL1 is the firstly defined gene involved in the reduction step from Chl-geranylgeranylated (ChlGG) and geranylgeranyl pyrophosphate (GGPP) to Chl-phytol (ChlPhy) and phytyl pyrophosphate (PPP) in rice. LYL1 can be induced by light and suppressed by darkness which is consistent with its potential biological functions. Additionally, the lyl1-1 mutant suffered from severe photooxidative damage and displayed a drastic reduction in the levels of α-tocopherol and photosynthetic proteins. We concluded that LYL1 also plays an important role in response to high-light in rice.


Introduction
Chlorophyll (Chls) molecules, which universally exist in photosynthetic organisms, play a central role in photosynthesis by harvesting light energy and converting it to chemical energy [1]. The Chl biosynthetic pathway was initially studied in Chl mutants of Chlorella [2]. Subsequently, Chls metabolism has been extensively analyzed in various organisms using biochemical and genetic approaches [3,4,5,6]. Because the early enzymatic steps of Chl biosynthesis, from glutamyl tRNA to protoporphyrin IX, are shared with the heme biosynthetic pathway, many essential data regarding the identity of the associated enzymes were obtained from studies of non-photosynthetic organisms such as Escherichia coli [7]. The later steps of Chl biosynthesis are shared with the bacteriochlorophyll biosynthetic pathway [8,9]. Directed mutational analysis using the photosynthetic bacterium Rhodobacter capsulatus had enabled the identification of genes involved in bacteriochlorophyll biosynthesis [5], and the homologous genes had been isolated from oxygenic plants [10]. To date, 27 genes encoding 15 enzymes in the chlorophyll biosynthetic pathway, from glutamyl-tRNA to Chl a and Chl b, have been identified in Arabidopsis, which represents angiosperm plants [11].
Chl consists of two moieties, Chlorophyllide (Chlide) and phytol, which are formed from the precursor molecules 5aminolevulinate and isopentenyl diphosphate, respectively, in two different pathways, i.e., the tetrapyrrole and isoprenoid biosynthetic pathways. Both pathways provide the substrates, Chlide and geranylgeranyl pyrophosphate (GGPP), necessary for the final steps of Chl biosynthesis. The last step of Chl synthesis, after conversion of protochlorophyllide to Chlide, has been studied intensively since geranylgeraniol was first identified as the esterifying alcohol of protochlorophyll a in pumpkin seeds [12]. Soll et al. suggested that there are two pathways for Chl biosynthesis [13]. In one pathway, GGPP synthesized in the chloroplast stroma is esterified to Chlide by Chl synthase in the thylakoid membranes, and the product Chl-geranylgeranylated (Chl GG ) is reduced stepwisely via Chl-dihydrogeranylgeraniol (Chl DHGG ) and Chl-tetrahydrogeranylgeraniol (Chl THGG ) to Chlphytol (Chl Phy ) [13,14,15]. Recombinant Chl synthase, encoded by the G4 gene of Arabidopsis [16] and overexpressed in Escherichia coli, also esterifies Chlide preferentially with GGPP to form Chl GG [17]. The bchP gene product of Rhodobacter sphaeroides is required for the three steps of the isoprenoid moiety of bacteriochlorophyll necessary for the reduction of Chl GG to Chl Phy [6,18,19]. In the other pathway, GGPP is reduced in the envelope membranes to phytyl pyrophosphate (PPP), which is then transferred to the thylakoid membranes, where Chl synthase directly generates Chl Phy [18,19]. Chl synthase derived from the ChlG gene of Synechocystis and bacteriochlorophyll synthase encoded by the Rhodobacter bchG gene give preference to PPP relative to GGPP [20].
The three-step hydrogenation of GGPP into PPP and Chl GG into Chl Phy is catalyzed by NADPH-dependent geranylgeranyl reductase [5,13,17]. Reduced activity of geranylgeranyl reductase leads to the loss of Chl Phy and the accumulation of Chl GG , Chl DHGG and Chl THGG . Geranylgeranyl reductase overexpressed in Escherichia coli catalyzes the stepwise hydrogenation of Chl GG to Chl Phy . Several genes encoding geranylgeranyl reductase were characterized in prokaryotes [5,21,22,23,24] and higher plants such as Arabidopsis [17], tobacco [15], peach [25] and olive [26].
In this study, we characterized a rice mutant lyl1-1 (light-induced yellow 1eaf 1-1) from japonica c.v. Zhonghua 11 (ZH11), displaying a dynamic yellow-green leaf phenotype, reduced level of Chl, arrested development of chloroplasts and hypersensitive to light. Map-based cloning of LYL1 revealed that this gene encodes a geranylgeranyl reductase. A single nucleotide C-to-T substitution in the coding region resulting in an amino acid change from an alanine residue to valine was found in the lyl1-1 mutant. We provided evidence that LYL1 simultaneously participates in the synthesis of Chl Phy and a-tocopherol in rice.

Characterization of a chlorophyll-deficient rice mutant
To investigate the molecular nature of rice chlorophyll metabolism, a light-induced yellow leaf mutant, designated as lyl1-1, was isolated from the progeny of a japonica rice ZH11 treated with 60 Co. Phenotypic observation showed that the lyl1-1 mutant grew slowly and produced premature yellowing leaves under natural conditions. The young leaves from leaf sheaths stayed green without any visible chlorosis ( Figure 1 A-C). However, the leaves rapidly turned yellow in several days. To characterize the yellow leaf phenotype of lyl1-1, we measured the Chl content. The contents of Chl a, Chl b and total Chl in the lyl1-1 mutant were 25.8% to 40.6%, 33.0% to 41.0%, and 30.8% to 40.4% of these in ZH11 plants, respectively, in different growth stages (Table 1). These results indicated that the yellow leaves of the lyl1-1 mutant resulted from reduced Chl levels.
We further investigated the ultrastructure of chloroplasts using transmission electron microscopy. In ZH11 plants, the chloroplasts displayed well-developed membrane systems composed of grana connected by stroma lamellae (Figure 1D, F). Grana stacks in the lyl1-1 mutant, however, appeared less dense and lacked grana membranes compared to those in ZH11. The thylakoid membrane systems of chloroplasts were disturbed in the lyl1-1 mutant, and the membrane spacing was not as clear as that in ZH11 chloroplasts ( Figure 1E, G). Therefore, the development of chloroplast thylakoid was suppressed in the lyl1-1 mutant.
We performed gel blot analysis to examine the abundance of LHC proteins (Light-harvesting chlorophyll-binding proteins) ( Figure 2). All LHCI proteins examined were found to be poorly accumulated in the lyl1-1 mutant. Lhca 3 was almost undetectable. Two major trimeric LHCII proteins, Lhcb1 and Lhcb2, and one monomeric LHCII protein, Lhcb4, were also inhibited in the lyl1-1 mutant. However, the accumulation of Lhcb5 was not affected.

The yellowing of lyl1-1 mutant was caused by high-light stress
To reveal whether the green-yellow transformation of lyl1-1 leaves depend or independent on environmental factors, we tested the response of lyl1-1 to different light and temperature treatments. Plants were first grown under low-light conditions (100 mmol photon m 22 s 21 ) and subsequently transferred to highlight conditions (400 mmol photon m 22 s 21 ) at 27uC. As shown in Figure 3, under low-light conditions, the lyl1-1 mutant displayed a phenotype similar to that of ZH11. The content of total Chl in lyl1-1 was slightly lower than that in ZH11 ( Figure 3A). After the transition to high-light conditions, the total Chl content in lyl1-1 rapidly decreased from 2.93 mg/g to 0.16 mg/g, whereas the Chl content in ZH11 increased from 3.18 mg/g to 3.62 mg/g ( Figure 3A, C). In addition, the changes in Chl a level occurred at a similar rate to that of Chl b in ZH11 and lyl1-1 ( Figure 3B). Experiments in which high-light was replaced by various temperatures (data not shown) indicated that temperature was not responsible for the observed yellowing. Taken together, the yellowing and light hypersensitivity of the lyl1-1 mutant may be caused by high-light stress.
To confirm this conclusion, we carried out a shading experiment. Plants were initially grown under low-light conditions. A black integument covered the center part of leaves to block out light before the plants were transferred to high-light conditions. After 6 days of exposure, the part of lyl1-1 leaf shaded by the integument remained green, but the remaining leaf turned yellow. No significant difference was observed in ZH11 leaf exposed to same conditions ( Figure 3D). This observation directly suggested that the yellowing of lyl1-1 is light-induced and the mutation of LYL1 gene enhances the photosensitivity of rice leaves.

Map-based cloning of LYL1
For genetic analysis of the lyl1-1 mutant, an F 2 population was constructed from the cross between lyl1-1 and 9311. All of the F 1 plants displayed a normal green leaf phenotype. The normal leaf and yellow leaf plants of the F 2 population showed a segregation ratio of 3:1 (X 2 = 0.66,X 2 0.05,1 = 3.84), which suggests that the yellow leaf phenotype in the lyl1-1 mutant is controlled by a single recessive nuclear gene.
The LYL1 gene was initially mapped between the markers W243 and W226 on the long arm of chromosome 2 ( Figure 4A), using the F 2 population derived from lyl1-1 and 9311. A comparison of chromosomal locations and leaf phenotypes indicated that LYL1 is a novel gene and different from previously identified genes related to leaf color alteration. For fine mapping of LYL1, more than 6000 F 2 individuals were developed and new InDel markers between W243 and W226 were designed according to sequence differences between indica and japonica rice (Table S2). Five markers exhibiting polymorphisms between the lyl1-1 mutant and 9311 were used to screen recombinants. Using 1203 recessive plants, the LYL1 gene was subsequently limited to a 33-kb region between the markers W246 and W232 on a single BAC clone, OJ1118_G04 ( Figure 4B). Within this DNA segment, six open reading frames (ORFs) have been predicted according to the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/ cgi-bin/gbrowse/rice/). All genes within this region were amplified and sequenced. A single nucleotide C-to-T substitution at position 182 in the coding region was found in the first exon of LOC_Os02g51080 in lyl1-1. This substitution results in a change from an alanine residue to valine ( Figure 4C). No other DNA sequence change was detected in other candidates. To examine whether the C-to-T mutation was present as a natural variant in other cultivars, we performed CAPS analysis of 22 typical indica and japonica rice cultivars, as the SNP in lyl1-1 removed the original SacII site. And all 22 cultivars exhibited the original restriction fragment ( Figure 4D). Thus, LOC_Os02g51080 is a good candidate gene for LYL1.
To confirm that the SNP mutation in LOC_Os02g51080 is responsible for lyl1-1, we utilized an RNA interference (RNAi) approach to knockdown this gene. Eleven transgenic plants expressing an inverted repeat of LOC_Os02g51080 were gener-ated in Nipponbare (Nipp). Among these, nine plants displayed the Chl deficient phenotype ( Figure 4E-H). In addition, a TOS17 retrotransposon insertion mutant, lyl1-2, was identified (line number, NE1041; Figure. 4C, I). The Tos17 insertion located in the exon 3 of LOC_Os02g51080 and no transcript of LOC_Os02g51080 can be detected in the lyl1-2 mutant ( Figure 4I). As expected, the lyl1-2 mutant exhibited yellow leaves with significantly reduced Chl level ( Figure 4J, K). Therefore, the   yellow leaves phenotype of lyl1 mutant was indeed caused by an SNP mutation in LOC_Os02g51080.

LYL1 encodes a geranylgeranyl reductase with FAD binding domain
Sequence comparison between genomic DNA and cDNA revealed that the LYL1 gene comprises three exons and two introns and encodes a 463-amino acid protein with a molecular mass of approximately 50 kDa. The C-to-T substitution resulted in a change from an alanine residue to valine in the encoded protein in lyl1-1. A protein BLAST search showed that LYL1 encodes a geranylgeranyl reductase with an FAD binding domain. One homolog having 57% sequence identity with LYL1, named as LIL2 (LOC_Os01g16020), was found in rice genome.
To illustrate the domain structure of LYL1 protein, we searched the Pfam database and found that it only contained a pyridine nucleotide-disulphide oxidoreductase domain (CL0063), a signature of the FAD super family. Blast searches also revealed that genes encoding FAD binding proteins exist widely in green plants, unicellular green algae, mosses, lycophytes and angiosperms. Although a whole genome sequence has not yet been found in gymnosperms, several ESTs from Picea sitchensis, P. glauca, Pinus taeda and P. contorta showed high similarity with LYL1 gene. To explore the phylogenetic relationship of these genes, we characterized homologs from the species representing the main lineages of green plants, including the green algae Chlamydomonas reinhardtii and Volvox carteri, the moss Physcomitrella patens, the lycophyte Selaginella moellendorffii and five monocot and five dicot angiosperms (Table S1). The amino acid length of these selected plant proteins ranged from 442 to 524. The FAD binding domain genes of green plants were placed into two groups with 100% bootstrap values in both the ML-and NJ-generated phylogenetic trees ( Figure 5). On the phylogenetic tree, all monocot species that were tested contain genes in both groups, while some dicot species seems to have lost genes. For example, Arabidopsis appears to have lost the gene from subgroup 2, while Medicago lost the group 1 gene. Furthermore, all  the genes from non-seed plants (including green algae, moss and lycophyte) were assigned to group 1, illustrating that subgroup 2 was formed independently in seed plants.
We searched the nr and EST information in NCBI databases, as well as available eukaryotic genome databases, and found that homologs of green plant geranylgeranyl reductase genes exist in bacteria and algae, including red algae, brown algae, diatoms and others. We selected representative homologs from each taxono-mical group of cellular organisms to build a large phylogenetic tree ( Figure S1, S2). The existence of homologous genes in multiple plants, bacteria and algae suggests that there is a widely conserved mechanism for the transformation from Chl GG to Chl phy and GGPP to PPP across divergent species. Expression patterns of LYL1 Quantitative real-time PCR analysis showed that LYL1 is constitutively expressed in organs such as root, stem, leaf, knot and panicle ( Figure 6A). However, the expression in leaf was relatively high, while and the expression in root was almost non-existent or at a very low level, indicating that the LYL1 gene has a specific expression pattern.
We examined the effects of light and dark growing conditions on the expression of LYL1, and found an obvious time-course change in expression levels was observed when plants were grown under light or dark conditions. The LYL1 transcript level was relatively low during the dark period, but rose rapidly after 6 h under low-light condition (400 mmol photon m 22 s 21 ) and achieved a 9-fold increase during 9 h. LYL1 expression decreased upon exposure to darkness and was restored within 6 h ( Figure 6C). To test whether LYL1 is differentially expressed under low-light conditions, we further analyzed the transcript accumulation in leaves exposed to a decreasing light intensity to 100 mmol photon m 22 s 21 . This exposure can also induce the expression level and resulted in a 6-fold increase of LYL1 transcripts at the maximal level. These results clearly demonstrated that LYL1 is an expressed light-responsive gene, which is consistent with its potential biological functions. Actually, most of rice genes involved in the chlorophyll biosynthesis pathway are light-responsive [27,28].

Mutation of LYL1 leads to an accumulation of Chl intermediates and a deficiency of a-tocopherol
Geranylgeranyl reductase is an NADPH-dependent enzyme and responsible for the reduction of GGPP to PPP and the reduction of Chl GG to Chl phy . To identify the function of LYL1 in rice, we analyzed the pigment composition of lyl1-1 using HPLC (Figure 7). In the lyl1-1 plants, Chl a and Chl b species are conjugated with incompletely reduced side chains, including Chl GG , Chl DHGG and Chl THGG , in addition to normal phytylated Chl a (Chl a phy ) and Chl b (Chl b phy ) ( Figure 7A, C, E). A similar change in Chl species was also observed in the lyl1-2 mutant. However, the lyl1-2 plants showed preferential accumulation of Chl GG , and the amounts of Chl DHGG and Chl THGG species were barely detectable ( Figure 7B, D, F). Taken together, these results suggested that the LYL1 gene participates in the last step of the production of Chl phy molecules in green rice seedlings, and reduced activity of geranylgeranyl reductase leads to the accumulation of Chl intermediates.
PPP is also an obligatory precursor for tocopherol synthesis and is directed into the tocopherol-synthesizing pathway through condensation with homogentisate derived from the shikimate pathway. To further explore the consequences of a mutation in LYL1 protein, we examined the tocopherol content in the lyl1 mutants ( Figure 8). HPLC analysis showed that the a-tocopherol levels in lyl1-1 and lyl1-2 were decreased to 12.52% and 8.81% of those in wild type, respectively. This indicated that the reduced activity of geranylgeranyl reductase results in a deficiency of atocopherol levels. The LYL1 gene is also required for the biosynthesis of a-tocopherol in rice.

LYL1 protects against lipid peroxidation and ROS
High-light stress excessively excites lipid peroxidation and generates reactive oxygen species (ROS), which in turn can attack various cellular components. Because the lyl1 mutants displayed an increased sensitivity to high-light stress, we detected the malondialdehyde (MDA) content, an indicator of lipid peroxidation. Under low-light (100 mmol photon m 22 s 21 ) conditions, the content of MDA in the lyl1-1 and ZH11 plants was similar ( Figure 9A). Under high-light (400 mmol photon m 22 s 21 ) conditions, the MDA content in the lyl1-1 leaves was approximate 2.3-fold of that in ZH11 ( Figure 9A). We also compared the levels of ROS, including hydrogen peroxide (H 2 O 2 ) and hydroxyl radicals (OH?), between ZH11 and lyl1-1. As shown in Figure 9, the ROS contents of ZH11 and lyl1-1 mutant under low-light conditions were indistinguishable. However, ROS levels in the lyl1-1 mutant were about 2.9-fold and 3.1-fold of those in ZH11 at high-light exposure ( Figure 9B, C). These results suggested that the lyl1-1 mutant senses a higher level of photodamage than wild type and LYL1 protects plant against lipid peroxidation and ROS.

Discussion
Chls are essential for photosynthesis. They are responsible for harvesting and transferring solar energy in antenna systems, and for charge separation and electron transport in reaction centers [7]. Chl metabolism is a highly coordinated process that is executed via a series of cooperative reactions catalyzed by numerous enzymes [11]. Analysis of the complete genome of Arabidopsis showed that it has 15 enzymes encoded by 27 genes for the biosynthesis of Chl from glutamyl-tRNA to Chl b [29]. However, only seven genes encoding five enzymes involved in Chl biosynthesis have been isolated in rice. Jung   Although two rice mutants, M249 and M134, that accumulate Chl intermediates with incompletely reduced alcohol side chains were previously characterized [36], the genetic properties of hydrogenating enzyme involved in the final step of Chl biosynthesis in rice are still unknown. In this study, we isolated a mutant, lyl1-1, from japonica rice c.v. ZH11 treated with 60 Co. This mutant exhibited dynamic yellow leaves, reduced levels of Chl, arrested development of chloroplasts and a retarded growth rate. Map-based cloning of LYL1 gene revealed that it encodes a geranylgeranyl reductase.
In the lyl1-1 mutant, a C-to-T substitution resulted in the change from an alanine residue to valine in the geranylgeranyl reductase. HPLC analysis indicated that the mutant accumulates Chl with incompletely reduced side chains. Under prolonged illumination, the lyl1-1 plants accumulated intermediates with a similar distribution of side chains in both the Chl a and Chl b groups in the final step of greening (Figure 7). This suggested that the reduction of side chains occurs in a stepwise manner during the conversion of Chl GG to Chl phy via Chl DHGG and Chl THGG in the synthesis of not only Chl a but also Chl b. Our results firstly proposed the pathway for the reduction of Chl GG to Chl phy by a hydrogenating enzyme in rice. We also noticed that the lyl1-1 mutant accumulated all six intermediates and the lyl1-2 mutant showed preferential accumulation of Chl GG . One explanation is that different mutations of LYL1 affect the preference of hydrogenation of the side chain during complete biosynthesis of Chl phy molecules in green seedlings.
Rice plants require high light to optimize photosynthesis during growth. However, the lyl1 mutant showed hypersensitivity to highlight stress (Figure 3). Under prolonged illumination, the total Chl content in lyl1-1 rapidly decreased, whereas the Chl content in ZH11 plants increased. The hypersensitivity of lyl1-2 was much more severe than that of lyl1-1. The lyl1-2 mutant grew very slowly and died after it was transferred to natural sunlight (data not shown). Direct evidence was further provided by a light-shading experiment ( Figure 3). Taken together, our data clearly confirm that the mutation of LYL1 leads to hypersensitivity to high-light, and the yellowing of leaves in the mutant is caused by light illumination. The increased sensitivity to high-light results from the lack of geranylgeranyl reductase activity is in agreement with previous studies in tobacco [37] and Synechocystis sp. PCC 6803 [38].
GGPP is esterified with Chlide to form Chl GG , which is subsequently stepwisely reduced to Chl phy . Alternatively, GGPP can first be reduced to PPP by geranylgeranyl reductase before it is conjugated with Chlide [18,19,25]. PPP forms the hydrophobic carbohydrate side chains of a-tocopherol molecules. The atocopherol levels of lyl1-1 and lyl1-2 were decreased to 12.52% and 8.81% of those of wild type, respectively (Figure 8). This indicated that LYL1 gene also contributes to the biosynthesis of atocopherol in rice. In addition, the content of a-tocopherol was found to coincide with the total amount of Chl. According to existing literatures and our data presented in this study, we proposed a model that LYL1 simultaneously participates in the synthesis of Chl and a-tocopherol in rice ( Figure 10).
Tocopherol is synthesized by all plants and the common functions are its ability to reduce ROS levels in photosynthetic membranes and to limit the extent of lipid peroxidation by reducing lipid peroxyl radicals to their corresponding hydroperoxides [39,40]. Tocopherol has long been speculated to have an essential function in protecting photosynthetic organisms against photooxidative stress [41,42,43]. However, this long-held assumption in photoprotection was broken by the results from the tocopherol-deficient mutants in Arabidopsis and cyanobacteria [44,45]. The photoautotrophic growth and photoinhibition of the tocopherol-deficient mutants in Arabidopsis (vte mutants) and Synechocystis sp. strain PCC 6803 (slr1736 and slr1737 mutants) was indistinguishable from that of the wild type under high-light stress. Subsequently, a mutant of the cyanobacterium Synechocystis sp. PCC 6803 lacking geranylgeranyl reductase, DchlP, was compared to strains with specific deficiency in tocopherol to assess the role of Chl a phytylatation [46]. The tocopherol-less Dhpt strain growed indistinguishably from the wild type under standard light photoautotrophic conditions, and exhibited only a slightly enhanced rate of photosystem I degradation under strong irradiation. Together with previous data, the results demonstrated that, in the DchlP mutant, accumulation of Chl a GG instead of deficiency of tocopherol leads to the instability of photosystem. In this study, rice LYL1 gene mutation reduced the a-tocopherol levels and resulted in yellow plants that had destructive chloroplast membrane system and increased photoinhibition and lipid peroxidation during high-light stress. It is difficult to attribute the observed phenotypes in these experiments directly to reduced a-tocopherol level: mutation of LYL1 also affects Chl level and causes the accumulation of geranylgeranylated Chl derivatives. Characterization of the mutant with specific deficiency in atocopherol will reveal the photoprotective mechanism of LYL1 in rice.
Recently, it was reported that a mutation in LIL3, one type of LHC-like protein, resulted in the accumulation of Chl molecules conjugated with incompletely reduced side chains and a reduction of a-tocopherol levels in Arabidopsis [47]. LIL3 interacts with and stabilizes geranylgeranyl reductase to complete a-tocopherol and Chl biosynthesis. BLAST analysis indicates that there is only one homolog of LIL3 (LOC_Os02g03330) in the rice genome. This gene, OsLIL3, encodes a 250-amino acid protein that has 66% and 58% homology with LIL3:1 and LIL3:2, respectively. We attempted to exam the interaction between the OsLIL3 and LYL1 proteins using the yeast two-hybrid system. However, no significant protein-protein interaction was detected (data not shown). There are several possible explanations for the result. First, the OsLIL3 and LYL1 proteins may not work as a complex in rice. Second, OsLIL3 may function with LYL1 via other proteins. Third, the interaction between OsLIL3 and LYL1 may be too weak to be detected. The molecular nature of the interaction between rice geranylgeranyl reductase and LHC-like proteins should be further studied.
In this study, we characterized a rice light-induced yellow leaf mutant that is hypersensitive to high-light and defective in the Chl phy synthesis. Map-based cloning revealed that LYL1 encodes a geranylgeranyl reductase. Our data suggest that the LYL1 gene functions simultaneously in Chl and a-tocopherol synthesis in rice. Our results highlight the critical functions of LYL1 in the response to light stress and the protection of photooxidative damage in rice.

Genetic analysis and map-based cloning of LYL1
For genetic analysis, an F 2 population derived from a cross between lyl1-1 and 9311, an indica variety, was grown in paddy fields under natural conditions (high light) and the leaf phenotypes could be clearly identified. This segregating population was also used for locating and fine mapping the LYL1 locus. Recessive individuals in the F 2 segregating population were used to screen recombinants. To fine map LYL1, InDel markers were developed based on sequence differences between indica variety 9311 and japonica variety Nipponbare, according to data published in NCBI (http://www.ncbi.nlm.nih.gov). The new polymorphic InDel markers were used to narrow down the region containing LYL1 (Table S2). Candidate genes were amplified and sequenced using gene-specific primers. A single nucleotide substitution of the putative lyl1-1 allele in the mutant was detected with a Cleaved Amplified Polymorphic Sequences (CAPS) marker using the primer pair 59-GAGGAGAAGCCACAGAAACG-39 and 59-TCTTGGTGAGGCAGTAGTAATAAA-39, followed by digestion with SacII. The sequence of LYL1 and lyl1-1 has been deposited into the NCBI/GenBank with an accession number KF305678 and KF305679.
For RNAi analysis, a DNA fragment of LOC_Os02g51080 was amplified by PCR using the primer pair 59-AAAG-GATCCCCGCTGTGCATGGTGTC-39 and 59-AAAACTAG-TATGTCGGGCTTGTGGGT-39. This fragment was cloned into the pMD18-T vector (TaKaRa) and sequentially cloned into the BamHI/SpeI and BglII/XbaI sites of the p1022 vector. Then, the stem-loop fragment was cloned into the p1301UbiNOS vector [48]. The resulting RNAi construct was transformed into A. tumefaciens and used for further transformation.

RNA extraction and quantitative real-time PCR
Total RNA was extracted from various tissues of ZH11 and lyl1-1 plants using Trizol reagent (Invitrogen) and treated with DNase I (TaKaRa) following the manufacturer's protocol. Approximately 1 mg of total RNA from each sample was used for first-strand cDNA synthesis. For quantitative real-time RT-PCR, first strand cDNAs were used as templates in reactions using SYBR Green PCR Master Mix (Takara) according to the manufacturer's instructions. OsActin gene was amplified as a control. Amplification of target genes was carried out using an ABI 7500 Real-time System. PCR was performed with the following primer sets: LYL1, 59-GCGGATGGTGGAGGAGA- 39 and 59-TGCCGATGGTGTTGACG-39; OsActin: 59-GATGACCCA-GATCATGTTTG-39 and 59-GGGCGATGTAGGAAAGC-39.

Transmission electron microscopy
The leaf samples of ZH11 and lyl1-1 plants were harvested from 1-month-old plants grown under natural conditions (high light). Leaf sections were fixed in 2% glutaraldehyde and further fixed in 1% OsO 4 . Tissues were stained with uranyl acetate, dehydrated in ethanol and embedded in Spurr's medium prior to thin sectioning. Samples were stained again and examined with a HITACHI H-600 transmission electron microscope.

Analysis of pigments and a-tocopherol
Chls were extracted from 0.2 g fresh leaves of ZH11 and the mutant with 95% ethanol, and Chl contents were determined with a spectrophotometer according to the previous method [34]. For analysis of Chl intermediates, approximately 5 mg of tissue from the leaves of light-grown seedlings or greening coleoptiles of darkgrown seedlings were weighed and processed in a chilled homogenizer with aqueous acetone, as previously described [36,47]. The homogenates were clarified by centrifugation. The extracts (10,20 ml) were injected onto an ODS-C18 reverse-phase column (Agilent, 4.66250 mm length, 2.5 mm) and eluted at 40uC with 100% methanol at a flow rate of 1.5 ml/min. The fluorescence intensity at 650 nm of chlorophyll in the eluate excited at 440 nm was monitored with a fluorescence detector. The amount of Chl a and b intermediates was calculated according to a previous method [36]. Three independent biological repeats were performed.
Tocopherol was extracted using previous methods [49,50]; approximately 0.3 grams of rice leaves was saponified under nitrogen in a screw-capped tube with 2 ml of potassium hydroxide (600 g/l), 10 ml of ethanol, 2 ml of sodium chloride (10 g/l) and 5 ml of ethanolic pyrogallol (60 g/l) added as an antioxidant. Tocopherol was determined using an Agilent 1200 HPLC. Resolution of vitamin E species was achieved using an Agilent Eclipse XDB-C18 column (4.66150 mm length, 5 mm) and a solvent system consisting of methanol: water (95:5, v/v) with a flow rate of 1.5 ml/min. Sample components were detected and quantified by fluorescence with excitation at 292 nm and emission at 330 nm. A sample volume of 10 ml was injected for chromatographic analysis. Three independent biological repeats were performed.

Western-blot analysis
The thylakoid membrane proteins of leaves were extracted according to a previous method [51]. And the protein content was determined by spectrophotometer using bovine serum albumin (BSA) standard as a reference. About 20 mg proteins were mixed with 5X loading buffer (250 mM Tris-HCl, pH 6.8, 50% glycerol, 10% SDS, 5% 2-mercaptoetnanol, 0.5% bromophenol blue). This mixture was boiled for 5 minutes and loaded onto a 12% SDS-PAGE gel. The proteins were separated and transferred onto a Nitrocellulose Transfer Membrane (Whatman) by electrophoretic cell (Bio-rad) with transfer buffer (25 mM Tris-base, 192 mM glycine, 20% methanol, pH 8.3). The membranes were blocked with Tris-HCl buffer containing 0.15 M NaCl and 5% non-fat dry milk, and were probed with primary antibodies (at a dilution of 1:200 in PBS) specific for LHC I subunits (Lhca1, Lhca2, Lhca3, and Lhca4) and LHC II subunits (Lhcb1, Lhcb2, Lhcb4, and Lhcb5), which were purchased from Agrisera. Then membranes were incubated with the alkaline phosphatase-conjugated secondary antibody (Sigma) (1:2000). The final substrates (NBT and BCIP) were added for color development [52,53].

ROS scavenging and lipid peroxidation determination
The level of lipid peroxidation was estimated in term of MDA content determined by thiobarbituric acid (TBA) reaction. About 200 mg tissue was homogenized with 5 ml 0.25% TBA. The homogenate was boiled for 40 min at 95uC and centrifuged at 13,000 g for 10 min. The absorbance of the supernatant was recorded at 532 nm and corrected by substracting absorbance at 600 nm. For the estimation of H 2 O 2 , about 200 mg tissue was homogenized in 0.9% physiological saline in a chilled pestle and mortar. The homogenate was centrifuged at 10,000 g for 10 min at 4uC and the supernatant was used for the detection of H 2 O 2 . The HO? level was determined spectrophotometrically based on the increase in absorbance of H 2 O 2 at 550 nm. In this assay, one unit of HO? is defined as the capability increasing the accumulation of 1 mmol of H 2 O 2 per milliliter. The HO? level was expressed as unit/mg protein. Protein content was determined according to the method of Bradford using bovine serum albumin as standard [54].
All the experiments were carried out with the kits (Nanjing Jiancheng Bioengineering Institute) according to the manufacturer's instructions. Three independent biological repeats were performed.

Phylogenetic trees analysis
To identify green plant genes encoding geranylgeranyl reductase, BLASTP searches were performed in the Phytozome database with the amino acid sequence of the rice gene LYL1 used as a query. If a protein sequence satisfied E#10 210 , it was selected as a candidate protein.
To identify the geranylgeranyl reductase in eukaryotic genome, BLAST searches against the non-redundant (nr) protein sequence database, NCBI EST database and available eukaryotic genome databases were performed using plant geranylgeranyl reductase sequences as queries. Protein sequences were sampled for further combined phylogenetic analysis from representative groups within each domain of life (bacteria, archaea and eukaryotes) based on BLASTP results against the nr database.
All of the selected representative protein sequences were aligned using Clustal X [55]. The gaps and ambiguously aligned sites were removed manually. Phylogenetic analysis was performed with a maximum likelihood approach using PhyML version 3.0 [56] and a Neighbor-joining method using MEGA [57]. A total of 100 nonparametric bootstrap samplings were performed to estimate the support level for each internal branch. Phylogenetic trees were visualized using the Explorer program of MEGA. Figure S1 Phylogenetic analysis of the LYL1 homologs. The numbers above the branches show bootstrap values for maximum likelihood and distance analysis, respectively. Asterisks indicate values lower than 50%. (TIF) Figure S2 Amino acid sequence alignment of LIL1 and other geranylgeranyl reductase proteins. Residues conserved across three or more sequences are shaded black, and similar residues conserved across three or more sequences are shaded gray. Numbers correspond to amino acid positions. (TIF)