Substitution segment identification and grain size analysis of Z668
Based on the previous development of Z668, the substitution segments and purity of the genetic backgrounds were detected with 10 individuals of Z668 using 11 markers on the substitution segment and 36 markers outside the substitution segment. All substitution segments of the 10 plants of Z668 were identical and no other residual segments derived from R225 were detected. Z668 contained 8 substitution segments from R225, which were distributed on chromosomes 1, 3, 6, 7, 10, 11 and 12, respectively. The total length of substitution segments was 29.08 Mb. The average length was 3.63 Mb. Among these substitution segments, the longest substitution length was 15.7 Mb and the shortest one was 0.625 Mb (Fig. 1).
Compared with Nipponbare, Z668 displayed dwarf (Fig. 2A) and short panicle length (Fig. 2B) and long-narrow grain (Fig. 2C, D). The grain length (10.19 mm), ratio of length to width (3.16) and 1000-grain weight (27.62 g) of Z668 were increased significantly by 39.9%, 31% and 25.9% than those of Nipponbare (Fig. 2E, G, H). While grain width (3.23 mm) of Z668 was decreased by 3.30% than that of Nipponbare (Fig. 2F).
Cytological analysis of glumes in Z668 and Nipponbare
Due to the significant long and narrow grain of Z668 compared with Nipponbare (Fig. 3A and Fig. 2C-F), the cell morphology of Nipponbare and Z668 glumes were analyzed by scanning electron microscopy (SEM) at the heading stage (Fig. 3B-I). The average glume cell length of inner epidermis in Z668 was 51.4 µm longer significantly than that (118.3 µm) of Nipponbare, whereas the cell width was narrower significantly by 3.3 µm than that (48.9 µm) of Nipponbare on average (Fig. 3B, C, F, G). No significant difference in total cell number in the outer epidermis of the glume along the longitudinal axis was observed between Nipponbare and Z668 (Fig. 3D, E, H). These results suggest that the increase in grain length and decrease in grain width of Z668 might have resulted predominantly from cell expansion, and not from changes in cell number.
QTL identification of grain size based on the secondary F2 population of Nipponbare /Z668
A total of 7 QTLs for grain size were identified by the secondary F2 population of Nipponbare /Z668, including 3 QTLs for grain length, 3 for length-to-width ratio and 1 for 1000-grain weight. They were distributed on the corresponding substitution segments of chromosomes 3, 7, and 12 (Fig. 1 and Table 1).
The additive effects of qGL3.4, qGL12-1 and qGL12-2 from R225 increased the grain length of Z668 by 0.60 mm, 0.33 mm and 0.23 mm, respectively, explaining the phenotypic variation by 48%, 33% and 12%, respectively. The additive effects of qRLW3.4, qRLW12-1 and qRLW12-2 from R225 increased the ratio of length to width by 0.20, 0.07 and 0.08, respectively, and explained 58%, 16% and 18% of phenotypic variation, respectively. The additive effect of 1000-grain weight of qGWT7 from R225 increased 1000-grain weight of 0.87g and explained 6.82% of the phenotypic variation.
Development of secondary single segment substitution lines harboring target QTL
Based on the QTL identification, 5 single-segment substitution lines (SSSL, S1-S5) harbored some of these QTLs were further developed in F3 by MAS method. S1 contained the substitution segment RM1254–RM3740–RM5336 of chromosome 1. S2 harbored substitution segment RM400–RM7412 - RM439–RM412 of chromo-
some 6. S3 carried the substitution segment RM8008–RM6574-RM5672–RM8006 of chromosome 7. S4 contained substitution segment RM7564–RM1132-RM234 -RM8261-RM2715–RM2789 of chromosome 7. S5 carried substitution segment RM6288–RM7119–RM1246 of chromosome 12. (Fig. 4A).
Verification and Analysis of Additive Effects of QTLs for Grain Size by the SSSLs (S1–S5)
A total of 13 QTLs for grain size were detected by the five SSSLs, of which 3 QTLs (qGWT7, qGL12-2 and qRLW12-2) were verified, indicating their stable inheritance. 4 QTLs (qGL3.4, qRLW3.4, qGL12-1, qRLW12-1) could not be validated because of no corresponding SSSLs still developed. Thus, 10 novel QTLs (qGL1, qGL6, qGL7, qGW7-1, qGW7-2, qRLW1, qRLW6, qRLW7-1, qRLW7-2 and qGWT6) were only identified by five SSSLs. The results suggested that SSSLs had a higher efficiency of QTL detection than the F2 population of Nipponbare/Z668 because it purified the genetic background in each plant and eliminated the masking of large-effect QTL to small-effect QTL.
The grain lengths (7.64, 7.89, 8.02 and 7.68 mm) of S1 with qGL1 (a = 0.09 mm), S2 carrying qGL6 (a = 0.21 mm), S3 harboring qGL7 (a = 0.28 mm) and S5 containing qGL12-2 (a = 0.11 mm) were significantly longer than that (7.45 mm) of Nipponbare and that (7.56 mm) of S4 without QTL for grain length (Fig. 4B). The grain widths (3.34 and 3.44 mm) of S3 with qGW7-1 (a = − 0.09 mm) and S4 containing qGW7-2 (a = − 0.04 mm) were significantly narrower than that (3.52 mm) of Nipponbare. However, S1, S2 and S5 without QTL for the trait displayed no significant difference in grain width with Nipponbare (Fig. 4C). The ratios of grain length to width (2.20, 2.27, 2.40, 2.20 and 2.20) of S1 with qRLW1 (a = 0.04), S2 carrying qRLW6 (a = 0.08), S3 containing qRLW7-1 (a = 0.14), S4 carrying qRLW7-2 (a = 0.04) and S5 with qRLW12 (a = 0.04) were significantly greater than that (2.12) of Nipponbare (Fig. 4D). The 1000-grain weight (26.0 g) of S2 with qGWT6 (a = 1.57 g) was significantly heavier than that (22.86 g) of Nipponbare, while 1000-grain weight (20.08 g) of S3 carrying qGWT7 (a = − 1.38 g) was significantly lighter than that (22.86 g) of Nipponbare. The 1000-grain weight (22.82, 24.24 and 23.7 g) of S1, S4 and S5 without QTL for the trait exhibited no significant difference with that (22.86 g) of Nipponbare (Fig. 4E).
Genetic Analysis of qGL3.4 for Grain Length
On the basis of QTL mapping, the F3 population of qGL3.4 consisting of 499 individuals were developed by a recombinant plant carrying only qGL3.4 (other grain length loci consistent with Nipponbare). The frequency of grain length in the F3 population displayed basically bimodal distribution, including short grain peak (between 7.01 and 8.05 mm) and long grain peak (between 8.06 and 10.63 mm) (Fig. 5). The chi-square test showed that the plant number (142) with short grains and long grain individuals (357) corresponded to the separation ratio of 1:3 (χ2 = 0.079 < χ2 (0.05, 1) = 3.84). The results suggested that qGL3.4 controlling long grain showed a dominant gene action.
Fine mapping and candidate gene analysis of qGL3.4
Since qGL3.4 as a major QTL explaining 48% of phenotypic variation and a single dominant gene inheritance in secondary single segment substitution line, then 142 recessive individuals with short grains in the F3 population were used for fine mapping of qGL3.4 by 5 new polymorphic SSR markers from 27 newly-designed SSR markers (Fig. 6A). The grain length was 7.68 mm in the 142 recessive plants, which showed no significant difference from that (7.45 mm) in Nipponbare (Fig. 6B). Finally, qGL3.4 was fine-mapped between SSR3 and SSR4 on chromosome 3, with a physical distance of 300 Kb. In the interval, seven possible genes were predicted. Then, by sequencing of the full length DNA of Z668 and Nipponbare, there were 5 genes (candidate genes 1–5) existing substantial sequence differences, and the other two no actual differences (one displaying no any differences and the other existing G to A changes in the middle of intron between Nipponbare and Z668).
Thus, the 5 genes were listed as the candidate genes 1 to 5 of qGL3.4. For candidate gene 1 (a WD-40 repeat family protein), 1 SNP difference was found from A of Nipponbare to C of Z668 at the 120th base of in the coding sequence (CDS), resulting in amino acid from serine of Nipponbare to arginine of Z668; At the same time, 3 bases were missing at the 139th base of the CDS sequence. For candidate gene 2 (a putative zinc finger protein-related), there were 2 bases missing at the 297th base in the 3 'UTR region and insertion of a 3-base fragment (CGC) at the 46th base, a 5-base fragment (GCCGC) at the 47th base, and a single base (C) at the 48th base in the CDS region, respectively. For candidate gene 3 (a zinc finger protein-related), there was a SNP from C to A at base 1311th of the CDS sequence, resulting in amino acids from histidine of Nipponbare to aspartic acid of Z668, and a single base C missing at base 1347th ; Although there was another SNP from T of Nipponbare to C of Z668 at the 1564th base, however, it belonged to a nonsense mutation. For candidate gene 4 (a endonuclease/exonuclease phosphatase family domain containing protein), there were four SNPs in 3 'UTR region, and CA in Nipponbare changed to TG at 120th -121st bases in Nipponbare, which caused changing from histidine of Nipponbare to cysteine of Z668. Although there was a G to A changing at the 200th base, it belonged to a nonsense mutation. At the 225th and 260th bases, it was changed from G and C to T and C, resulting in changing from alanine and isoleucine of Nipponbare to serine and threonine of Z668, respectively. For candidate gene 4, there was a SNP from C to T at the 93th base of the CDS sequence, which did not cause amino acid change. For candidate gene 5 (a putative calcineurin B protein), there was a 3-base fragment deletion at the 299th base of the 3’UTR, and a single base A deletion at the 516th base of CDS, and a change from G of Nipponbare to A of Z668 at the 284th base of the CDS region, resulting in changing from arginine to glutamate,.
Further real-time quantitative fluorescent RT-PCR was performed on candidate genes 1–5 which caused changes in protein-coding region. The results showed that the expression levels of candidate gene 1 in the root, stem, leaf, sheath and panicle of Z668 were significantly lower than that of Nipponbare. There was no significant difference in expression level of candidate gene 2 in panicle of Nipponbare and Z668, however, the expression levels in other tissues of Z668 were significantly higher than those in Nipponbare. The expression level of candidate gene 3 in the leaves of Z668 was significantly higher than that of Nipponbare, while ones in the stem and panicle of Z668 were significantly lower than those of Nipponbare, and no significant difference in the root and sheath between them. The expression levels of candidate gene 4 in the stem and sheath of Z668 were significantly higher than those of Nipponbare, whereas one in the root of Z668 was significantly lower than that of Nipponbare, and no differences in leaves and panicle between them. The expression level of candidate gene 5 in the root and sheath of Z668 was significantly higher than that of Nipponbare, while one in the leaves of Z668 was significantly lower than that of Nipponbare, and no differences in the stem and panicle between them. Taken together, the results suggested that Candidate genes 1 and 3 will be the preferred candidate genes of qGL3.4.