Dissection and validation of minor quantitative trait loci (QTLs) conferring grain size and weight in rice

Grain size and weight contribute greatly to the grain yield of rice. In order to identify minor QTLs conferring grain size and weight, an F2 population derived from a cross between two indica rice lines showing small difference on grain size, Guangzhan 63-4S (GZ63-4S) and Dodda, and its derived F2:3 population were developed and used for QTL analysis. Totally, 36 QTLs for grain size and weight were detected, and 7 were repeatedly detected, of which the number of beneficial alleles was contributed roughly equally by the two parents. In order to further validate effects of QTLs detected, a BC1F2 population derived from a backcross of a mixture of F2 lines with GZ63-4S was developed and subjected to QTL selection. Heterozygous regions of 3 QTLs, qGS3, qTGW6.2 and qGT7 were identified, and corresponding near-isogenic lines (NILs) of each QTL were constructed with three rounds of self-crosses. In the background of NILs, qGS3 was responsible for GL, LWR, GT and TGW, qTGW6.2 was for GL and TGW, and qGT7 was for GT and TGW. These results have laid the foundation of further fine mapping and cloning of underlying genes, and could be of great use in breeding and improvement of rice lines with desirable size and yield.


46
Rice is one of the staple crops worldwide, and feeds more than half of the world's 47 population. In the face of continuously increasing population and reduced arable land, 48 how to further improve the grain yield of rice is a major concern of scientists and 49 breeders. Grain size, characterized by four factors viz., grain length (GL), grain width 50 (GW), length-to-width ratio (LWR) and grain thickness (GT), contributes greatly to 51 grain weight, which is a key determinant of grain yield [1]. Therefore, dissection of 52 the genetic basis that underlies grain size and weight would be of great use in 53 developing rice lines with high grain yield. 54 Considerable efforts have been made to investigate the genetic basis of grain size and 55 weight in the past two decades, and results showed that the four factors of grain size, 56 GL, GW, LWR and GT, and thousand-grain weight (TGW) are quantitative traits, and 57 subjected to control of many genes [2,3]. Up to now, large numbers of quantitative 58 trait loci (QTLs) have been identified, however, only a small proportion of QTLs 59 displaying large effect have been cloned, such as GS3 [4,5], OsMADS1 [6, 7], 60 GL3.3/TGW3 [8][9][10], GW5/GSE5 [11,12], GS5 [13], GW8 [14], GS2/GL2 [15][16][17], 61 GL7/GW7 [18,19], etc. Although the knowledge of molecular regulation of grain size 62 and weight has greatly increased, the mining and cloning of more QTLs, especially 63 minor QTLs, is still of great importance to have a better understanding of underlying 64 mechanisms and provide breeding programs with valuable gene resources. 65 Rice lines displaying large difference on grain size and weight were always selected 66 to develop segregating populations for QTL analysis, which resulted in the repeated 67 detection of several major QTLs/genes. For example, two major genes for grain size, 68 GW2 and GL3.1 were identified and cloned from genetic populations derived from 69 FAZ1 and WY3, of which the TGW values differ by 23.12g [20,21]. The two genes 70 above, together with another two major genes, GS3 and GW5/GSE5, contributed to 71 the huge variation of grain size and weight between N411 and N643, of which the 72 TGW values differ by 54.33g [22]. The existence of major genes is likely to interfere 73 the mapping and validation of minor QTLs, exemplified by the fine mapping of GS5 74 [13]. Therefore, in order to identify minor QTLs for grain size and weight, rice lines 75 displaying small difference should be preferred. 76 Quantitative traits are easily affected by environment, which leads to the instability of 77 QTL detection. Therefore, genetic validation of QTLs is of great necessity in further 78 breeding utilization or cloning. The most frequently used method is evaluation the 79 effect of a QTL using near-isogenic lines (NILs), which are lines that carry 80 segregating regions at target QTL but homozygous regions in the rest of genome [23]. 81 NILs for a QTL are always developed by backcrossing lines carrying the QTL region 82 from donor to the receipt several times until the non-target QTL regions were 83 completely from the receipt, which could achieve the simultaneous improvement of 84 target traits of recipient [24,25]. Another simple method is to select lines carrying 85 segregating target QTL regions from inbred populations that have undertaken several 86 rounds of self-crosses, also known as residual heterozygous lines (RHLs) [26,27].

87
This method is sometimes utilized for absence of laborious hybridization work. The

88
NIL of Ghd8, a major QTL with pleiotropic effects on grain yield, heading date and 89 plant height, was constructed by screening lines carrying segregation target regions 90 from a RIL population of the F 7 generation [28,29].

91
In this study, in order to identify minor QTLs for grain size, two indica rice lines 92 displaying small difference, Guangzhan 63-4S (GZ63-4S) and Dodda were selected to 93 develop the F 2 and derived F 2:3 populations, and QTL analysis of grain size and 94 weight were performed. In order to validate QTL detected, lines carrying 95 heterozygous QTL regions were screened from a BC 1 F 2 population derived from a 96 backcross of a mixture of F 2 lines with GZ63-4S. NILs of three QTLs were developed 97 by a series of self-crosses of screened BC 1 F 2 lines, and further used for evaluation 98 their genetic effect on grain size and TGW. Company, and has been mated with many restorer lines to produce promising hybrid 105 combinations in recent years [30]. Dodda is an indica cultivar with unknown origin, 106 belonging to the core germplasm collections of our lab. The TGW values of GZ63-4S 107 and Dodda differ by less than 10 g (data not shown).

108
As displayed in Fig.1, 1000 F 2 lines were produced from a cross between GZ63-4S 109 and Dodda, and were subjected to selection of the TMS5 locus conditioning 110 thermo-sensitive genic male sterility with a closely linked marker [31]. 214 lines 111 carrying homozygous TMS5 regions were selected to make up the F 2 mapping 112 population, which was further self-crossed to produce the F 2:3 population. Both the F 2 113 and F 2:3 population was exploited to map QTLs for grain size and TGW. In addition, 114 1200 BC 1 F 2 lines were produced by backcrossing a mixture of F 2 lines to 115 followed by a self-cross. These lines were subjected to TMS5 selection, and 250 lines 116 carrying homozygous TMS5 regions were selected to perform heterozygous QTL 117 regions screening with flanking makers in the mapping process (Table 2). BC 1 F 2 lines 118 carrying heterozygous QTL regions were further self-crossed three times to produce 119 the BC 1 F 5 populations, which were utilized to validate the effect of QTLs.

148
Phenotypic variation and correlation of the F 2 and F 2:3 populations 149 GZ63-4S is a typical photoperiod-and thermo-sensitive genic male sterile line, and 150 shows male sterility in the normal growing seasons in Wuhan. Therefore, the seeds 151 could not be harvested, which abolished comparison of grain size and TGW between 152 the two parents. All the five traits of the F 2 and F 2:3 populations showed continuous  All the four grain size factors were significantly positively correlated with TWG in 156 both years, except for LWR (Table 1)   The region flanked by marker LRJ99 and RM232 on chromosome 3 and consisting of 200 four QTLs, was responsible for GL, LWR, GT and TGW in both the F 2 and F 2:3 201 population, and was term qGS3, hereafter.  Table 1). Lines carrying heterozygous regions of three 208 QTLs, qGS3, qTGW6.2 and qGT7, were identified respectively, and were self-crossed 209 three times to produce NIL populations for each QTL.

219
For qGT7, significant differences were observed in the average values of GT among 220 the three genotypes, qGT7 Dodda , qGT7 H and qGT7  , and in that of TGW between 221 qGT7 Dodda and qGT7  in the NIL population (Table 5) In this study, a total of 37 QTLs were identified for GL, GW, LWR, GT and TGW in 245 the F 2 and F 2:3 populations, and 7 QTL regions were repeatedly detected, of which the 246 additive effects were far less than that of cloned major genes for grain size, such as 247 GS3, GL3.1/qGL3, GW5/GSE5 and GW2 [4,5,11,12,[20][21][22], demonstrating minor 248 QTLs for grain size and weight. Moreover, the number of beneficial alleles was  (Table 2, Table 3), suggesting that qGS3 is 281 a stable and pleiotropic QTL for GL, GT and TGW. qTGW6.2 was initially detected 282 as a QTL for TGW, but was validated to have effect on both GL and TGW in the NIL 283 population and act in a dominant manner (Table 4). The failure in detection of 284 qTGW6.2 on GL in F 2 and F 2:3 population may be attributed to the complexity of 285 genome background and the low variation explained, which further supported the 286 necessity of validation of QTLs using NILs. qGT7 was repeatedly confirmed as a 287 QTL for GT, and had no effect on GL and GW in the F 2 , F 2:3 , and NIL populations 288 (Table2 , Table 5). Being one of the four factors of grain size, GT has received less 289 attention, and several cloned genes conditioning GT are responsible for GL and/or 290 GW at the same time, such as GS2, GW8 [17,44]. Therefore, qGT7 is a good 291 candidate for further research of the molecular mechanism underlying GT.