Cloning And Functional Analysis Of A Major Qtl, GS5for Grain Szie/Weight And Chalk5for Chalkiness Rate In Rice

Increasing grain yield and improving grain quality of food crops are two of the most important goals of plant basic and applied science research. In rice (Oryza sativa L.) and many other cereals, grain size, determined by grain length, width, thickness and grain filling, is a major and very stable determinant of1000-grain weight, one of three main components of grain yield. In addition to being a yield trait, rice grain size is also a highly important quality trait, determining grain appearance quality and affecting grain milling, cooking and eating qualities of grain. Grain size is a target trait of both domestication and artificial breeding and can also provide a model system for studying the evolutionary processes underlying domestication of crop plants. Grain chalkiness greatly affects grain appearance, milling, cooking and eating and nutritional qualities of grain, so it is the most important trait of grain qualities. Recently, although no QTL gene (QTG) for grain chalkiness has been cloned, a number of QTGs for grain size and weight in rice have been isolated using a map-based cloning approach, but they are all negative regulators of grain size. Here we have successfully cloned and functionally analyzed a minor QTG, GS5, on the short arm of rice chromosome5positively controlling grain size. And we have also successfully isolated a major QTG, Chalk5, positively controlling grain chalkiness rate in the same interval with GS5. The main results of our study are as fellows:1. We isolated two near isogenic lines (NILs) from BC3F2by fixing qSW5/GW5for the small-grain allele. Compared with NIL(H94), the grains of NIL(ZS97) were8.7%wider and7.0%heavier, leading to a7.4%increase in grain yield per plant. The GS5locus is underlining a dominant allele from Zhenshan97. The grain filling rate of NIL(ZS97) became significantly higher than that of NIL(H94) during grain filling. Thus the increase in grain weight and hence yield per plant resulted from increases in both grain width and grain filling rate. Thus, we used grain width as the target trait for further analysis of the minor QTL, GS5.2. To fine map the GS5locus with minor effects, enough recombinants were identified using two BC3F2segregating populations consisting of9638individuals in two years, and their grain-width phenotypes were determined by progeny testing of the recombinant plants (BC3F3). By combining the analysis of grain-width phenotypes and marker genotypes for each recombinants in the locus, we narrowed GS5locus to an11.6-kb region between markers RM574and S2, with three recombinants between RM574and GS5and four recombinants between S2and GS5, cosegregating with the InDel marker C62. There is only one predicted ORF in this11.6-kb region, which was regarded as the sole candidate for GS5. We identified a full length cDNA corresponding to the ORF. The ORF consisted of ten exons, encoding a putative serine carboxypeptidase belonging to the peptidase S10family, with a PF00450consensus domain.3. For verification of the candidate ORF for GS5, we over-expressed the cDNA of GS5from H94(narrow grain) driven by the35S promoter, and found that there was a remarkable overall increase in grain size (width)(+8.2%) and1000-grain weight (+15.6%) in the transgene-positive plants compared with the negative plants. Cosegregation tests in the T1progeny showed that the expression level of GS5positively correlated to grain width. Wider grains need higher levels of GS5expression. The fact that overexpressing the allele from the narrow-grain parent produced much wider grains indicated that mutations in the coding region of GS5were unlikely the cause for the observed grain-width variation, and thus favored the hypothesis that the grain size differences were regulated by variations in the regulatory sequences of the GS5gene. To substantiate this possibility, we generated an additional transformation construct:the native promoter of the wide-grain GS5allele from Zhenshan97was fused with the cDNA from H94. A remarkable increase in grain width (+6.2%) and1000-grain weight (+12.8%) was observed in the transgene-positive plants compared with the negative plants. The detection of the expression of GS5in the To plants and T1progenies suggested that the promoter from Zhenshan97could produce higher levels of GS5expression and result in wider grains, and that the expression level of GS5positively correlated to grain width. The two transgenes did not affect plant morphology, plastochron or flowering time and the chalkiness phenotype. Moreover, the gs5mutant produced smaller grains than wild type, which cosegregated with the T-DNA insertion and the expression level. Taken together, the results unequivocally confirmed that natural variations in the GS5promoter region were the cause of the GS5effect on grain width.4. GS5comparative sequencing found that two varieties (H94and Minghui63) with narrow grains had identical sequences, which had26variations compared with the sequence of the wide grain variety Zhenshan97. Comparison of the promoter sequences revealed18polymorphisms, including substitutions, deletions and insertions, in one group relative to the other in the2-kb region upstream of the translation start site. Eight variations occurred in the coding region. Six bases were inserted at10-15bp downstream of the translation start site in Zhenshan97relative to that of the H94and Minghui63, resulting in an in-frame increase of two amino acids in the predicted signal peptide. There were also three SNPs in the downstream sequence between the two varietal groups resulting in substitutions of three amino acids, and four SNPs in the introns.5. We next sequenced the GS5promoter regions (2kb, including the six-base deletion in the first exon) and measured grain width for a total of51rice accessions, including35cultivated varieties mostly from China and16accessions of wild rice Oryza rufipogon from a wide geographic range of Asia. The sequences of cultivated varieties could be divided into three haplotypes:H94type (narrow grain), Zhenshan97type (medium grain), and Zhonghua11type (wide grain). Accessions within each group had exactly the same sequences. There were18polymorphic sites between H94and Zhenshan97; Zhonghua11differed from H94and Zhenshan97by22and26sites, respectively. Six wild accessions of very different origins had exactly the same sequence as H94, and also had narrow grain. The remaining9accessions had little similarity with any of the three types. These results suggested that the H94allele of GS5is likely the wild type, and the other two alleles for wider grains are gain-of-function mutant types arising during domestication, functioning to up-regulate GS5expression to increase grain size and accelerate grain filling.6. Genetic transformation and GUS activity assay of the eight GS5promoter-deletion fragments amplified from Zhenshan97and H94indicated that one or more of the four SNPs in the233-bp region from-1241to-1009upstream of the translation start site are very likely responsible for the different expression levels of the two alleles and are functional variations of the GS5QTL.7. The temporal and spatial expression pattern of GS5, in NIL(ZS97) and NIL(H94) using Real-time PCR (qRT-PCR) with total RNA from14tissues, showed that the levels of GS5transcript varied drastically among the tissues. In particular, the transcript was much more abundant in NIL(ZS97) than in NIL(H94) in the8-cm young panicles, in the paella/lemma at2,4and5days before heading, and in the endosperm at10days after fertilization. Such expression differences corresponded well with the critical stages for grain width and grain filling. RNA in situ hybridization in various tissues indicated that GS5has a strong expression in the shoot apical meristem, young spikelet and little floret at various developing stages and the endosperm, consistent with the observed effect on seed size. Comparative subcellular-localization of the GS5proteins in Arabidopsis protoplasts showed that both of the proteins were co-localized with an endoplasmic reticulum (ER)-specific marker Bip protein, indicating the ER localization of the GS5 proteins. This result also implied that the five amino acid changes of the two predicted proteins were not involved in the ER localization signal.8. We examined cross-sections of the central parts of the paella/lemma of the spikelet between NIL(ZS97) and NIL(H94). The inner parenchyma cell layer of NIL(ZS97) contained a substantially greater number of cells than that of NIL(H94)(18.1%more in paella,21.2%more in lemma and20.1%more in total). In addition, the cell size of paella in NIL(ZS97) was21.3%larger than that of NIL(H94), whereas no significant difference was detected in lemma cell size. Thus, GS5positively regulates grain size mainly by increasing the cell number, and to some extent also cell size, leading to enhanced latitudinal growth of grain size.9. We analyzed the expression of25genes related to cell cycle and12genes related to BR in the gs5mutant and GS5over-expressor, relative to the wild-type. The transcript levels of the five putative G1/S phase genes, CDKA1, CAK1, CAK1A, CYCT1and H1, were greatly elevated in the GS5over-expressor compared with the negative segregants. In contrast, their expression was significantly reduced in gs5relative to the wild-type. Thus, GS5functions putatively as a positive modulator upstream of the cell cycle genes by regulating the transcription of the putative Gl/S genes.10. The Chalk5locus is underlining a dominant allele from Zhenshan97(higher white belly rate) by a genetic analysis of a subpopulation from the major-QTL NILs. To fine map the Chalk5locus, using two BC3F2segregating populations consisting of9638individuals, we narrowed Chalk5locus to a16.8-kb region between markers C181and C35, both with three recombinants confirming the mapping results. There are only two predicted ORFs in this16.8-kb region, in which an ORF with an endosperm-specific expression pattern was considered as a reliable candidate gene. We identified a full length cDNA corresponding to the ORF. The ORF consisted of four exons, encoding a vacuolar H+-pyrophosphatase (V-PPase).11. For verification of the candidate ORF for Chalk5, we conducted a complementary transformation experiment, with the Chalk5allele from Zhenshan97driven by its native promoter. A remarkable increase (+26.6%) in grain chalkiness rate (white belly rate) was observed in the transgene-positive plants compared with the negative plants. Cosegregation tests of the T1and T2progenies confirmed the transgenic effects. We also over-expressed the Chalk5allele from Zhenshan97driven by the35S promoter, and found that there was a remarkable overall increase (+18%) in grain chalkiness rate in the transgene-positive plants compared with the negative plants. Cosegregation tests in the T1and T2progenies also confirmed the effects. Meanwhile, the RNA level of the two transgenes from ZpZc and OX is significantly (P<0.05) correlated with the grain chalkiness rate or white belly rate. Taken together, the results confirmed that the ORF was the gene of the Chalk5QTL, and suggested that Chalk5has a positive effect on grain chalkiness, and polymorphisms in the Chalk5promoter region may be the cause of the Chalk5effect.12. The two transgenes and NIL(ZS97) greatly decreased head rice rate, total protein content,1000-grain weight and unit weight, and increased the amylose content (AC) and gel consistency (GC), which showed that Chalk5not only greatly affects grain appearance quality, but also influences grain milling quality, grain cooking and eating quality, nutritional quality as well as grain weight.13. Comparative sequencing of Chalk5found that the two varieties H94and Minghui63, both with low grain chalkiness rate, had the identical sequence of Chalk5gene, which had39variations compared with the sequence of the variety Zhenshan97with high grain chalkiness rate. Comparison of the promoter sequences revealed10polymorphisms, including substitutions, deletions and insertions, in one group relative to the other in the1.8-kb region upstream of the translation start site. Specifically, there is a12-bp deletion of the Zhenshan97allele in the1.0-kb region upstream of the translation start site. Five variations occurred in the exons and resulted in substitutions of two amino acids. There were also twenty-four SNPs in the inrons.14. The temporal and spatial expression patterns of Chalk5, in NIL(ZS97) and NIL(H94) using Real-time RT-PCR with total RNA from11tissues, showed that Chalk5transcript level has a tissue-specific pattern. In particular, the transcript was much more abundant in NIL(ZS97) than in NIL(H94) in the endosperm at5days after fertilization, whereas the opposite was true in the endosperms at10days after fertilization. But almost no transcript was detected in the stem,4-cm and8-cm young panicles and in the paella/lemma before heading in both NILs. Such an expression pattern corresponded well with the critical stages for grain chalkiness formation in the process of grain filling, implying a putative association between promoter sequence types and expression levels.15. Two common nucleotide variations in two potential cis-elements of Chalk5promoter that may create diversity in Chalk5mRNA levels might be a major cause for grain white belly diversity among indica rice. Chalk5controlled white belly rate by affecting the shape and spatial arrangement of endosperm storage substances by SEM and differential expression analysis.In brief, GS5functions as a positive regulator of grain size with minor effects by increasing cell number. Three haplotypes were identified corresponding to different grain width in cultivated rice. Haplotypes for large grain arose as gain-of-function mutations in the promoter region during the course of domestication and breeding. Natural variations in the promoter region of GS5are the cause of the QTL and have contributed greatly to grain size diversity in rice. Chalk5is a major QTL gene positively controlling grain white belly rate for chalkiness among indica rice. Chalk5has large and general effects on many grain quality traits. Natural variation in the promoter region of Chalk5might be a major cause for white belly diversity in indica rice. The results provide molecular evidences for the inconsistency and consistency between rice grain yield and quality caused by the three tightly linked QTGs, GS5, qSW5and Chalk5.