Isolation and characterization of the 637ys mutant
In a previous study, we isolated the yellow-green leaf mutant 502ys from japonica cultivar Nipponbare (NP), which accumulated the Chls with unsaturated side chains, and was resulted from a point mutation causing an amino acid substitution G206S in OsCHLP (LOC_Os02g51080) gene [15]. Here, we obtained a new mutant 637ys from japonica cultivar ZH11 via EMS mutagenesis, which accumulated majority of Chls with unsaturated side chains as well as a small amount of Chlphy (about 10% of Chlphy in the wild type) (Additional file 1: Figure S1). The 637ys mutant displayed a yellow-green leaf phenotype through the whole growth period and grew at a very slow rate. The young leaves from leaf sheaths stayed green in 637ys, but rapidly turned yellow in several days (Fig. 1). Despite 19 days delay to heading compared to wild type ZH11, 637ys showed dramatic declines in major agronomic traits. For instance, while the wild type plants had an average on 7.2 of productive panicles, 637ys plants had only one to at most three, decreasing by 84.7%. The other agronomic traits, plant height, panicle length, No. of spikelets per panicle, seed setting rate, and 1000-grain weight declined by 38.6%, 28.2%, 80.7%, 42.7%, and 33.8% correspondingly (Table 1).
Table 1 Comparison of major agronomic traits between the 637ys mutant and its wild-type ZH11
Traits
|
ZH11 (WT)
|
637ys
|
Compared with WT
|
Days to heading (d)
|
75.0±0.8
|
94.0±2.2
|
+25.3%*
|
Plant height (cm)
|
108.2±1.6
|
66.4±3.5
|
-38.6%*
|
No. of productive panicles per plant
|
7.2±0.5
|
1.1±0.1
|
-84.7%*
|
Length of main panicle (cm)
|
23.8±0.8
|
17.1±1.0
|
-28.2%*
|
No. of spikelets per panicle
|
228.9±6.1
|
44.2±4.1
|
-80.7%*
|
Seed setting rate (%)
|
93.0±1.2
|
50.3±2.8
|
-42.7%*
|
1000-grain weight (g)
|
26.9±0.4
|
17.8±1.0
|
-33.8%*
|
*Significantly different at P = 0.05.
To quantify the yellow-green leaf phenotype of the 637ys mutant, we measured the photosynthetic pigments in the 637ys and ZH11 plants at both seedling and heading stages. The contents of total Chl, Chl a, Chl b, and Caro in 637ys significantly reduced by 50.6% to 58.2%, 47.4% to 57.3%, 63.4% to 61.4%, and 24.0% to 53.4% respectively, compared to those in wild type (Fig. 2). These results suggested that the mutant phenotype resulted from reduced level of photosynthetic pigments.
To investigate if the reduced contents of photosynthetic pigments affect the development of chloroplasts in 637ys, we observed the ultrastructure of chloroplasts under transmission electron microscopy. A number of grana stacks consisting of well-developed grana lamellae connected by stroma lamellae were present in wild type chloroplasts (Fig. 3a, b). However, the chloroplasts were swollen in 637ys. Even if some grana stacks existed, the grana lamellae were less densely spaced than those in wild type and changed into disarray arrangement. Furthermore, stroma density decreased and osmiophilic globules occurred in the stroma in 637ys (Fig. 3c, d). These results revealed that the development of chloroplast was suppressed in the 637ys mutant.
Sensitivity of 637ys mutant to temperature and light intensity
To explore whether the phenotype of 637ys was associated with temperature, we treated the mutant and its wild type in the growth chamber using two different temperature conditions (constant 23 °C and 30 °C). As a result, the 637ys mutant exhibited similar leaf-color phenotype under different temperature conditions (Additional file 2: Figure S2 a1, a2). Its Chl contents significantly reduced, compared to wild type, but there was no obvious difference between low temperature and high temperature, which was similar to those in its wild type (Fig. 4; Additional file 3: Table S1; Additional file 4: Table S2). These data suggested that the phenotype of 637ys was independent upon temperature.
All mutants accumulating the Chls with unsaturated side chains displayed sensitivity to light intensity [14, 30]. Correspondingly, we also investigated the phenotype of 637ys mutant under low light (80 μmol m-2 s-1) and high light (300 μmol m-2 s-1). The mutant displayed yellow-green leaf phenotype under high light condition (Additional file 2: Figure S2 a1-b2). Meanwhile, its Chl contents significantly declined, compared to that under low light condition, while the Chl contents in wild type remained relatively stable (Fig. 4; Additional file 5: Table S3). These data suggested that the phenotype of 637ys depended on light intensity.
Analysis of vitamin E in leaves and grains
Tocopherols and tocotrienols constitute vitamin E. The phytyl-PP forms the side chains of both Chlphy and tocopherols, and the GGPP forms the side chains of ChlGG and tocotrienols [4, 5]. Because of the accumulation of Chls with unsaturated side chains in 637ys mutant, to investigate whether the composition of vitamin E was affected, we analyzed the tocopherol and tocotrienol compositions in leaves and grains in 637ys and its wild type ZH11 by HPLC. In leaves, HPLC profiles of vitamin E showed that α-tocopherol was abundant in wild type, while the elution peak of α-tocopherol in 637ys was much lower than that in wild type and significantly decreased by 89.2% (peak 1 in Fig. 5b, c; Fig. 5f). At the same time, a small amount of γ-tocopherol was detected in the wild type leaves (peak 2 in Fig. 5b), but not in the 637ys mutant. It is noteworthy that a minor peak (peak 7 in Fig. 5c), whose retention time was 0.5 min fewer than the peak of γ-tocopherol in wild type, was detected in the 637ys mutant. We speculated that the minor peak in the 637ys was likely to be an isomer of γ-tocopherol [4]. In addition, tocotrienols in leaves were almost undetectable either in wild type or 637ys mutant (Fig. 5b, c). In grains, HPLC analysis showed that the 637ys mutant had few of α-tocopherol or γ-tocopherol declining by 90.9% and 89.7% respectively (peaks 1 and 2 in Fig. 5d, e; Fig. 5f), but considerable level of tocotrienols comparable to ZH11 (peaks 4, 5 and 6 in Fig. 5d, e; Fig. 5f; Additional file 6: Figure S3) [4, 14]. These results indicated that tocopherols were deficient, but the accumulation of tocotrienols was not affected in 637ys.
Map-based cloning of the 637ys mutant gene
We crossed 637ys with 502ys (chlp) mutant, and the resulting F1 plants all displayed normal green phenotype, which indicated that 637ys and 502ys mutant genes are not allelic. In order to perform genetic analysis for 637ys mutant, 637ys was backcrossed with its wild type ZH11 and crossed with normal green indica cultivar G46B. As a result, all F1 plants exhibited normal green phenotype. F2 populations from the two crosses showed a segregation ratio of 3:1 (χ2<χ20.05 =3.84, P>0.05), suggesting that the yellow-green leaf phenotype of 637ys was controlled by a single recessive gene.
Next, the F2 mapping population from the cross between 637ys and G46B was constructed. Preliminary linkage analysis suggested that the 637ys locus was linked with the SSR marker RM6641 on the short arm of Chromosome 2, and then we used 2 SSR markers and 3 InDel markers (Additional file 7: Table S4) to locate 637ys in a 334-kb region between SSR markers RM110 and RM7033 with 0.2 and 3.7 cM respectively (Fig 6a, b). Within this region, we have further developed a total of 8 InDel and SSR markers which however showed no polymorphism between 637ys and G46B.
Although there are 62 putative genes within the 334-kb region according to the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/annotation_pseudo_current.shtml), the genetic distance between the 637ys locus and RM110 is much shorter than that between 637ys and RM7033, which suggested that the 637ys locus should be close to RM110 (Fig. 6b, c). Therefore we analyzed the genes in this region starting from RM110 based on the annotations and the TargetP and ChloroP (http://www.cbs.dtu.dk/services/TargetP/; http://www.cbs.dtu.dk/services/ChloroP/) [31, 32]. Of particular interest, we found a gene, LOC_Os02g03330, which encodes a light-harvesting like protein with 57% and 58% identities to LIL3:1 and LIL3:2 in Arabidopsis respectively (Fig. 6d). The lil3:1 lil3:2 double mutant of Arabidopsis accumulated ChlGG, ChlDHGG, and ChlTHGG, showing a similar phenotype to the 637ys mutant [18]. Then we sequenced the DNA extracted from 637ys mutant and its wild type, and the results revealed that a G-to-A substitution occurred in position 368 of the gene (Fig. 6d). Furthermore, we sequenced cDNA of LOC_Os02g03330 in 637ys mutant and ZH11. Sequence alignment showed that the substitution occurred at the first nucleotide of the first intron in the DNA sequence of LOC_Os02g03330, and consequently caused 10 bp-intron sequence insertion in cDNA sequence of this gene in 637ys mutant (Fig. 6e). Thereby the reading frame shift in LOC_Os02g03330 resulted in premature translation of its encoded protein in 637ys (Additional file 8: Figure S4a). Therefore, LOC_Os02g03330 was considered as the candidate gene of 637ys mutant, and designated as OsLIL3.
Searching in the rice genome database revealed that OsLIL3 is a single copy gene. Alignment of sequenced DNA and cDNA showed OsLIL3 consists of three exons and two introns and its full length of genomic sequence and cDNA are 2384 bp and 753 bp, respectively. OsLIL3 encodes a 250-amino acid protein with a molecular weight of 27.6 kDa. The OsLIL3 contains a predicted chloroplast transit peptide of 44 amino acids at N-terminus (Additional file 8: Figure S4) [31, 32]. The structure-prediction programs TMHMM and HMMTOP (http://www.cbs.dtu.dk/services/TMHMM/; http://www.enzim.hu/hmmtop/html/submit.html) [33, 34] indicate two transmembrane helices, among which the first helix of OsLIL3 proteins includes the well-conserved LHC motif (Additional file 9: Figure S5). According to multiple alignment of OsLIL3 and its homologues in different species, OsLIL3 has a high similarity to its homologues in monocotyledonous plants, barley (Hordeum vulgare), and maize (Zea mays) and dicotyledonous plants, cucumber (Cucumis sativus) and tobacco (Nicotiana tabacum), with 72%, 72%, 70% and 63% respectively. Phylogenetic analysis showed that OsLIL3 is more closely related to the LIL3 proteins of barley and maize than those of other species (Fig. 7).
Complementation of the 637ys mutant
To confirm that the mutation of OsLIL3 caused the yellow-green leaf phenotype in 637ys mutant, we performed a complementation assay. The construct pC2300-OsLIL3 carrying OsLIL3 driven by rice Actin1 promoter was generated by inserting full length cDNA of OsLIL3 into pCAMBIA2300 vector. Then we introduced the final construct into 637ys mutant by Agrobacterium-mediated transformation, and obtained 11 transgenic lines. These transgenic plants recovered to normal green (Fig. 8a–c). In addition, we determined the Chl compositions in these positive lines. As shown in Fig. 8d–f, positive transgenic lines only accumulated Chlphy instead of Chl conjugated with unsaturated geranylgeraniol side chains. Meanwhile, their levels of tocopherols in leaves reached to those of wild type (Fig. 8g–i). These data suggested that the OsLIL3 gene rescued the deficiency of Chlphy and tocopherols in 637ys, from which we conclude that the mutant phenotype of 637ys was due to the single base pair mutation in the OsLIL3 gene.
Subcellular localization of OsLIL3 protein
OsLIL3 was predicted to contain a chloroplast transit peptide with 44 amino acid residues at its N-terminus by using TargetP and ChloroP (Additional file 8: Figure S4) [31, 32]. In order to prove this prediction, we generated constructs expressing OsLIL3-green fluorescent protein (GFP) fusion protein, pCAMBIA2300-35s-OsLIL3-GFP, transformed rice protoplasts with the final construct and pCAMBIA2300-35s-GFP (as control) respectively, and observed transformed protoplasts under a laser-scanning confocal microscopy. In accordance with what was predicted by TargetP and ChloroP, the green fluorescence of OsLIL3-GFP fusion protein overlapped with the red autofluorescence of Chl in the chloroplasts, while GFP itself was expressed all over the whole cell (Fig. 9). These data provide strong evidence that OsLIL3 is chloroplast targeted.
Expression analysis of OsLIL3 gene
To investigate the expression pattern of the OsLIL3 gene, we examined its transcripts in different tissues of the wild type at both seedling stage and booting stage by qRT-PCR. The results demonstrated that the OsLIL3 was differentially expressed in all tissues, including roots, stems, leaf blades, leaf sheaths, and young panicles. Particularly, leaf blades had the highest levels of transcripts, followed by leaf sheaths and young panicles, while stems and roots had relatively low levels of transcripts (Fig. 10). The results indicated that OsLIL3 was mainly expressed in green tissues.
Expression analysis of genes at seedling stage for photosynthesis and Chl synthesis
Since the Chl compositions changed in the 637ys mutant, to investigate whether expressions of the genes associated with photosynthesis and Chl synthesis were affected, we examined transcript levels of 19 related genes and OsLIL3 at seedling stage in 637ys. Among these genes, six genes are related to photosynthesis [35], including rbcL and rbcS (Rubisco large and small subunits, respectively), CAB1R and CAB2R (Chl a/b-binding proteins of PS II), psaA and psbA (two reaction center polypeptides). Two genes are related to the heme branch, including FC1 and FC2 (ferrochelatase1 and 2) [36]. 11 genes encode enzymes involved in Chl biosynthesis, including HEMA1 (glutamyl-tRNA reductase), CHLD, H and I (D, H and I subunits of Mg chelatase), CHLM (Mg-protoporphyrin IX methyltransferase), CHL27 (Mg-protoporphyrin IX monomethylester cyclase), DVR (divinyl reductase), PORA (protochlorophyllide oxidoreductase), OsCHLP, YGL (CHLG, Chl synthase), and CAO1 (chlorophyllide a oxygenase) [2, 14, 15, 35-39]. However, except that OsLIL3 was significantly down-regulated in the 637ys mutant, no significant change in transcription levels of the other genes detected was found. Interestingly, the expression of OsCHLP was not significantly affected in the 637ys (Fig. 11). The results suggested that the functional defect in OsLIL3 did not affect the transcriptional levels of the aforementioned genes for photosynthesis and Chl and heme synthesis, which were consistent with those from lil3:1 and lil3:2 single mutants and the lil3:1 lil3:2 double mutant [20].
Phenotype and Chl composition of the 502ys 637ys double mutant
Both 637ys and 502ys accumulated Chls with unsaturated side chains and defected in tocopherols [14, 15]. To further explore the association between OsCHLP and OsLIL3 in Chl phytyl biosynthesis in rice, we generated the homozygous 637ys 502ys double mutant by crossing 502ys mutant with 637ys mutant. Under natural sunlight condition, all F1 plants displayed a normal green phenotype. F2 population plants segregated into four phenotypic classes at the early seedling stage: normal green plants, yellow-green plants like 502ys, yellow plants like 637ys, and more severely yellow and smaller plants which were confirmed as 637ys 502ys double mutants by sequencing the mutation sites of OsCHLP in 502ys and OsLIL3 in 637ys (Fig. 12). To avoid competition for light with other plants stronger than double mutants in the F2 population, only 637ys 502ys double mutant plants were retained in the soil. Unlike 502ys or 637ys (Fig. 13b, c), only ChlGG a and ChlGG b were exclusively accumulated, and none of ChlDHGG, ChlTHGG, or Chlphy was detectable in the double mutants by HPLC (Fig. 13d). To investigate the chloroplast development in the double mutants, we also observed the ultrastructure of chloroplasts under transmission electron microscopy. Compared to 637ys and 502ys (Fig. 3) [15], few of well-developed grana stacks existed in the double mutant (Additional file 10: Figure S6). Unfortunately, the double mutants all died at the three-leaf stage (Fig. 12). In addition, we also investigated the 637ys 502ys double mutant grown in a growth chamber under low light at constant 23 °C, and obtained similar results of exclusive accumulation of ChlGG and lethal phenotype at the three-leaf stage (Additional file 11: Figure S7). These results suggested that the complete absence of Chlphy or only the presence of ChlGG in the double mutant could be fatal to rice seedling.