With the structural adjustment and upgrading of the rice industry, the market demand for superior quality geng/japonica rice is increasing every year (Chen et al., 2018). The grain shape of traditional geng/japonica rice in China is mostly short and round, but recently long grain shape has become preferred by the market because of its excellent appearance quality (Xu and Chen., 2016; Huang and Qian., 2017). Currently, the dep1 gene is widely used in the breeding of geng/japonica rice in China due to high-yielding population characteristics, however, the gene also brings in characteristics of short grains and low thousand grain weight (Liu et al., 2018). Therefore, breeding of high-yielding long-grain geng/japonica rice cultivars by incorporating dep1 with major-effect grain shape genes is of high priority in rice industry, as the theoretical basis and required germplasm reports are limited (Chen et al., 2012; Xu and Chen., 2016).
Multiple genes for grain shape have been cloned, some of which have strong effects and wide applications, including the major-effect genes controlling grain length, GS3, qGL3 and GL7/GW7, the thousand grain weight gene TGW6, plant shape gene DEP1 (Fan et al., 2006; Zhang et al., 2012; Ishimaru et al., 2013; Wang et al., 2015 a; Wang et al., 2015 b) and major-effect genes that control grain width, GW2, GW5, GS5 and GW8 (Song et al., 2007; Wan et al., 2008; Li et al., 2011; Wang et al., 2012). Multiple studies have shown that the grain shape genes directly determine the shape of rice grains by regulating the development of glume cells (Fan and Li., 2019), and also indirectly affect grain chalk by determining endosperm development after fertilization (Wang et al., 2008; Li et al., 2014). In addition, a series of endogenous hormones and starch synthesis-related enzymes are also involved in the dynamic development of the glumes and endosperm. For example, brassinosteroids (BR) and indole acetic acid (IAA) are reportedly related to the development of glume cells (Li et al., 2018), whereas the enzyme activities of adenosine diphosphate glucose (ADPG), granule-bound starch synthase (GBSS), soluble starch synthase (SSS) and starch branching enzyme (SBE) on endosperm development is significant (Dong et al., 2008).
In the present study, utilizing CRISPR/Cas9 technology, we created a series of NILs (YF47dep1-gw8, YF47dep1-gs3, YF47dep1-gl7, YF47dep1-qgl3 and YF47dep1-tgw6) in YF47dep1 background. Further, the comparison of grain appearance and yield were contracted, and the related impact factors were also deeply clarified. The objectives of this study included: i) Clarifying the effects of GW8, GS3, GL7, qGL3, and TGW6 genes (partial major-effect grain shape genes) on grain appearance and yield of erect panicle rice, ii) creating a series of long-grain erect-panicle geng/japonica rice gremplasm, and evaluating its breeding application value in aspect of appearance quality and yield, and iii) providing the theoretical basis for breeding long-grain shape geng/japonica rice.
We analyzed 96 geng/japonica rice varieties’ genotype distribution at the GW8, GS3, GL7, qGL3 and TGW6 loci, which were widely used in xian/indica rice breeding programs. As shown in Supplemental Fig. 1, very few long grain type alleles have been used in geng/japonica rice breeding in northern China. Therefore, by utilizing CRISPR/Cas9 technology, we created a series of NILs as our research materials, which carried the mutant alleles at the above loci under the background of geng/japonica rice cultivar, YF47dep1, having an erect panicle architecture. The NILs comprised of YF47dep1-gw8, YF47dep1-gs3, YF47dep1-gl7, YF47dep1-qgl3 and YF47dep1-tgw6, which present single base insertion in target site and led to truncated proteins of various sizes compared with YF47dep1 (wild type, WT) (Fig. 1a). Besides the number of tillers, the plant architecture of NILs had insignificant differences compared with YF47dep1 (Fig. 1b).
The grain lengths and widths were determined at maturity stage (Fig. 2a-d), the grain length of YF47dep1-gw8 was 5.29 mm and around 5.1 mm for the other NILs, which showed significant increase compared to that of YF47dep1(4.9 ± 0.08 mm). The grain width was significantly lower in YF47dep1-gw8 (2.39 ± 0.09 mm) and significantly higher in YF47dep1-gs3 (2.90 ± 0.06 mm), compared to YF47dep1 (2.80 ± 0.09 mm). No significant differences in width were observed in the other lines. The average length to width ratio of YF47dep1 was 1.74 (Fig. 2e), but significantly higher in YF47dep1-gw8 (2.22 ± 0.11 mm), YF47dep1-gl7 (1.81 ± 0.11 mm), YF47dep1-qgl3 (1.84 ± 0.04 mm), and YF47dep1-tgw6 (1.82 ± 0.05 mm). The length to width ratio of the YF47dep1-gs3 grain was statistically the same as that of YF47dep1 due to the increase in its width. Scanning electron microscope analysis showed that the mean length of the glume cells in all of the NILs was significantly greater than that of YF47dep1, with those of YF47dep1-gw8 being the longest (Fig. 2f-i). The cell width in YF47dep1-gw8 was significantly reduced but substantially increased in YF47dep1-gs3, and no significant difference was observed in the other NILs. The cell number in YF47dep1-gw8 and YF47dep1-gl7 was significantly increased, whereas it decreased dramatically in YF47dep1-gs3 and YF47dep1-tgw6.
It has previously been reported that increased levels of IAA and BR might increase fruit size (Li et al., 2018). Therefore, we analyzed the levels of the endogenous hormones IAA and BR during the most vigorous period of glume development (Fig. 2j-k). Of all the NILs, no significant change was seen in YF47dep1-gl7 and YF47dep1-qgl3. However, the levels of IAA and BR were substantially reduced and elevated in YF47dep1-gw8 and YF47dep1-tgw6, respectively. The IAA level was significantly elevated in YF47dep1-gs3, whereas its level of BR remained statistically the same although slightly elevated. Along with the thousand grain weight (TGW) performance in Table 1, we speculated that IAA and BR might promote the TGW by regulating glume development. In erect-panicle varieties, short and round grains with low TGW are prone to form because of the up-regulated expression of dep1 (Sun et al., 2018). In our study, apart from YF47dep1-gw8, the level of dep1 expression was down-regulated in all of the transgenic plants (Fig. 2l), which may serve as a contributing factor to their elongated grain lengths. On the contrary, the level of dep1 was increased in YF47dep1-gw8, which may be related to its significantly decreased TGW (Table 1).
Table 1
The comparison of yield traits in WT and the NILs
Varieties | NEP | NFGP | TGW (g) | Yield (kg∙ha− 1) |
2019 | 2020 | 2019 | 2020 | 2019 | 2020 | 2019 | 2020 |
YF47dep1(WT) | 18 ± 1.62 | 18 ± 1.15 | 132 ± 3.92 | 131 ± 2.12 | 25.9 ± 0.51 | 25.8 ± 0.65 | 10132.84 ± 99.25 | 9860.48 ± 97.87 |
YF47dep1-gw8 | 19 ± 1.58 | 18 ± 0.87 | 128 ± 4.15** | 125 ± 3.33** | 23.2 ± 0.48** | 23.1 ± 0.38** | 9418.59 ± 112.14** | 9168.47 ± 112.14** |
YF47dep1-gs3 | 18 ± 1.34 | 18 ± 1.36 | 125 ± 2.98** | 123 ± 3.15** | 27.5 ± 0.39** | 27.6 ± 0.29** | 10777.61 ± 86.51** | 10494.13 ± 115.21** |
YF47dep1-gl7 | 19 ± 1.87 | 17 ± 0.97 | 127 ± 3.14** | 129 ± 2.97* | 24.4 ± 0.42** | 24.3 ± 0.42** | 9610.35 ± 79.24** | 9287.97 ± 96.21** |
YF47dep1-qgl3 | 16 ± 0.98** | 15 ± 1.15** | 131 ± 3.61 | 127 ± 3.45* | 23.9 ± 0.48** | 23.9 ± 0.61** | 9085.29 ± 96.51** | 8992.38 ± 97.89** |
YF47dep1-tgw6 | 18 ± 1.12 | 19 ± 1.71 | 128 ± 3.28* | 120 ± 4.12** | 26.7 ± 0.54** | 26.6 ± 0.42** | 10774.82 ± 115.41** | 10402.42 ± 93.21** |
NEP, number of effective panicles; NFGP, number of filled grains per panicle; TGW, thousand grain weight. |
The grain chalky characteristics of the NILs were investigated after the grains were processed into polished rice (Fig. 3a-b). Chalkiness degree in all the NILs was significantly lower than that of YF47dep1, which demonstrated an average of 7.31%. The lowest level of chalkiness degree was observed in YF47dep1-gw8 with a value of 1.51 ± 0.41%, followed by YF47dep1-qgl3 (2.28 ± 0.57%) and YF47dep1-gl7 (3.29 ± 0.51%). The chalkiness degree of YF47dep1-gs3 and YF47dep1-tgw6 was relatively high, measured as 5.37 ± 1.19% and 4.08 ± 0.40%, respectively. Also, the length to width ratio and chalkiness degree showed a significant negative correlation (Fig. 3c). This indicated that increasing the length to width ratio of grains and reducing chalkiness degree were harmonized in the erect-panicle background.
To further investigate the effect of endosperm development on grain chalkiness variations, we analyzed the grain filling rate by calculating the changes in endosperm weight during 0–35 days after flowering (Fig. 3d-e). The maximum and average grain filling rates for all the NILs were significantly reduced compared to that of YF47dep1. Of these, YF47dep1-gw8 showed the lowest maximum and average grain filling rates, whereas the measurements for YF47dep1-gs3 and YF47dep1-tgw6 were relatively high. Wang et al., (2012) demonstrated that the activities of ADPG, GBSS, SSS and SBE are closely related to the filling rate of endosperm. Furthermore, we analyzed the activity changes in the starch biosynthesis-related enzymes in the NILs. Compared to YF47dep1, the activities of these enzymes were significantly reduced in all of the NILs (Fig. 3f-i), with the exception that the activity of GBSS in YF47dep1-gs3 and YF47dep1-tgw6 was substantially elevated. Of all the NILs, the activity levels of the four enzymes in YF47dep1-gw8 were the lowest. These results indicated that the up-regulation of these four enzymes was helpful in increasing the filling rate, and the relatively high activity of GBSS in YF47dep1-gs3 and YF47dep1-tgw6 might be due to the high endosperm dry weight and high TGW.
The variation in yield traits of the NILs was compared in two consecutive growing seasons (Table 1). Generally, the yields of YF47dep1-gs3 and YF47dep1-tgw6 improved significantly compared to the control YF47dep1 due to the substantially increased TGW, whereas YF47dep1-qgl3 exhibited the lowest yield due to the significantly reduced effective panicle number and TGW. However, the yields of YF47dep1-gw8 and YF47dep1-gl7 were relatively low owing to the dramatically decreased TGW and grain number per panicle.
While ensuring the high yield of erect-panicle varieties, the primary breeding goal is to develop long grain geng/japonica rice with the desired appearance qualities via the genetic improvement of grain shape (Xu and Chen., 2016; Huang and Qian., 2017). As pioneer research establishments, the China National Rice Research Institute(CNRRI) and Jiaxing Academy of Agricultural Sciences have developed Jiahe 218 and Jiahe 212 with a grain length to width ratio of 3.0 through the aggregation of dep1 and gs3 (Huang and Qian., 2017), which provides a reference for the genetic improvement of grain shape in erect-panicle varieties. However, the Jiahe series rice varieties contain a large amount of tropical japonica background (Huang and Qian., 2017), which is different from our research basing on the temperate japonica background. In the present study, YF47dep1-gw8 grain exhibited the greatest length to width ratio of 2.2, comparing the best-known long grain high quality japonica rice varieties (the length to width ratio is generally more than 2.5) (Huang and Qian., 2017), the grain shape of NILs in this article is still not ideal. We speculated that the simultaneous pyramiding of GS3/TGW6 and GW8 long type allele should be an effective way of developing erect-panicle geng/japonica rice with excellent appearance quality and high yield.
In summary, our results demonstrated that editing of grain shape genes was an effective approach to creating long-grain erect-panicle geng/japonica rice germplasm. Owning to both of the appearance quality and yield improvement, GS3 and TGW6 alleles can be applied directly for breeding long-grain shape geng/japonica rice, and editing GW8 resulted in excellent appearance quality but low yield. Therefore, this gene would be difficult to use directly but can be considered as the core germplasm resource. All this work could provide the required germplasm and theoretical basis for breeding of high-yielding long-grain geng/japonica rice cultivars.