9,311, a BPH-susceptible elite restorer line for HL-CMS and two-line hybrids, was used as the recurrent parent. L1880, a chromosome segment substitution line (CSSL) carrying Gn8.1 gene in 9,311 genetic background, was used as the donor parent for large panicle. Shuhui527, an elite restorer line for WA-CMS and two-line hybrids, was used as the donor parent for Wild-Abortive type restorer gene. Luoyang-6 and Luoyang-9, CSSLs carrying Bph6 and Bph9 in 9,311 genetic background, respectively, were used as the donor parents for BPH-resistance. In addition, seven different types of CMS or GMS sterile lines were used as test parents in this study (Table S1). The brown planthopper populations used for infestation were collected from rice experimental field in Wuhan University.

A seedling bulk test was conducted to evaluate the brown planthopper resistance of rice populations following previously described methods (Huang et al., 2001). The seeds of each rice were pre-germinated to ensure that all seedlings were grown consistently. Detailedly, about 20 seeds were sown in a 9 cm-diameter plastic cup with three replications. At the third-leaf stage, the seedlings were thinned to 10 plants per cup and infested with second to third-instar nymphs of brown planthopper with eight insects per seedling. When all of the seedlings of susceptible control died, the plants of other rice were examined and each seedling was given a score of 1 to 9 according to the method of Huang et al. (2001).

Rice genomic DNA for PCR analysis was extracted from young leaves using the CTAB method (Murray and Thompson, 1980). Seven gene-specific PCR markers, Indel33, RM16994, RM28438, RM10318, RM6100, RM25661, and Indel200-1, tightly linked to the target genes Gn8.1, Bph6, Bph9, Rf3, Rf4, Rf5, and Rf6, respectively, were used in foreground selection of the target genes in each generation of intercross and backcross (Table S2). PCR was performed in a total volume of 10 μl containing 50 ng of DNA template, 1 μl 10 × PCR buffer, 0.25 μM of each primer, 75 μM of each dNTP, and 0.25 units of Taq DNA polymerase (Fermentas). The PCR amplification condition was as follows: one cycle at 94°C for 5 min, followed by 36 cycles of 94°C for 30 s, 50~60°C (dependent on the primers) for 30 s, 72°C for 45 s, and a final incubation at 72°C for 5 min. All the PCR products were analyzed by electrophoresis on agarose gel.

A total of 672 SSR markers distributed evenly on the 12 chromosomes with an average marker interval of 553.4 kb were used for rice genetic polymorphism survey, and 225 SSR markers were found to be polymorphic between the recurrent and donor parents (Table S3). Then, the 225 markers were used to identify the background of the pyramided lines. The physical map was drawn using MapMaker Version 3.0 (Lander et al., 1987). The percentages of chromosome segments from recurrent parent in pyramided lines were reckoned following the graphical genotyping procedure (Xi et al., 2006; Suh et al., 2013).

All of the rice lines were planted in the experimental field of Ezhou in summer, and Hainan in winter during 2011–2016. Among them, the hybrid combinations and their parents used for evaluating the restoring ability and production potential of the new restorer line were planted in the Wuhan University Ezhou Experimental Base in the summer of 2016. Seedlings were transplanted in a five-row plot with 12 plants per row by 20 × 26 cm spacing under a randomized complete block design with three replications. Five representative plants in the middle of each repeat were used to measure growth duration (GD), plant height (PH), panicle number (PN), number of grain (NG), number of grain per panicle (NGP), spikelet fertility (SF), 1,000-grain weight (GW), and yield per plant (YP). The statistical analyses were performed with SPSS Statistics 20 (IBM, USA).

As shown in the schematic flow (Figure 1) of the development of the BPH-resistant wide-spectrum restorer lines, five parents including L1880, Luoyang-6, Luoyang-9, and typical restorer line Shuhui527 and 9,311 were used in this breeding program. Of which, L1880 derived from crosses of 9,311 × O. longistaminata carries big panicle gene Gn8.1, Luoyang-6 and Luoyang-9 carry Bph6 and Bph9, respectively, Shuhui527 carries restoring gene Rf3 and Rf4, and 9,311 carries restoring gene Rf5 and Rf6. During the breeding process, foreground selection was performed from generation F₁ to BC₃F₃ to screen the plants having the desirable alleles of the seven target genes using molecular markers (Table S2). Only progenies carrying all of target alleles were selected for backcross or selfing (Figures S1–S3). Meanwhile, phenotype identification was also conducted so as to eliminate plants with linkage drag traits (Figures S4, S5; Tables S4, S5). In detail, 20 positive F₁ plants with seven target genes were identified and genotyped using the target gene markers from a ~1,000 F₁ population derived from intercross of L1880/Luoyang-9//Shuhui527/Luoyang-6. From which, 16 plants were identified having all of the seven target genes (Gn8.1/Bph6/9/Rf3~6), and further backcrossed as maternal parents with 9,311. In the BC₁ generation, 18 out of 890 plants were identified as our expected candidates after careful genotyping and phenotyping. Similarly, the selected BC₁, BC₂ plants were then backcrossed with 9,311 so as to eliminate the unexpected background. Then, 29 BC₃F₁ plants were selected for selfing until genetically stable in BC₃F₃. Finally, 22 stable BC₃F₃ lines with target genes were developed from 13,580 plants. In consideration of the integrative agronomic traits and the presence of homozygous marker alleles for the seven target genes, 16 lines were selected as potential candidates from the 22 BC₃F₃ progenies (Figure S3; Table S5).

To test whether the selected pyramiding lines can improve the BPH-resistance, we evaluated their resistance to brown planthopper at seedling stage under greenhouse conditions, using Luoyang-6 and Luoyang-9 (donor parents) as resistant control, and 9,311 (recurrent parent) as a susceptible control. After inoculated for 1 week, when all the 9,311 seedlings were dead, the pyramided lines were still green (Figure 2A). Resistance scoring showed that BPH-resistance indexes of the part of pyramided lines ranged from 1.0 to 1.8, significantly less than that of the Luoyang-6 (3.2) and Luoyang-9 (3.3) (Figure 2B). This result implied that these pyramided lines' BPH-resistance were significantly stronger than that of the single-gene donor parents Luoyang-6 and Luoyang-9, although both of them showed high resistant level.

Importantly, comparing with the recurrent parent 9,311, these pyramided lines, especially line 05, 06, and 08 all showed excellent performance in the field test. The traits such as growth period, plant height, panicle number and 1,000-grain weight of the three pyramided lines were almost the same as that of 9,311. However, the panicle size and grain number of 05, 06, and 08 were significantly larger than that of 9,311. This is a desirable agronomic trait for increasing yield. As a result, even if the spikelet fertility of 05, 06, and 08 were slightly declined, their yield per plant were still increased by 77.3, 66.1, and 85.1%, respectively than that of 9,311 (Table 1). Thus, the pyramided lines 05, 06, and 08 were finally chosen as the potential elite BPH-resistant restorer lines based on their excellent performance, and renamed as R345, R346, and R348, respectively (Figure 3).

In order to identify the genetic background of the three pyramided lines, a total of 672 evenly distributed SSR markers were used for PCR identification, and physical maps covering the whole 12 rice chromosomes were constructed (Figure 4). The results showed that each pyramided line carried part of the donors' fragments at the seven target gene loci, even after three rounds of backcrossing with the recurrent parent 9,311. Averagely, the 9,311 genomic fragments took over 91.4, 87.5, and 88.3% in pyramided lines R345, R346, and R348, respectively (Table S3). These results demonstrated that the seven target genes Gn8.1, Bph6, Bph9, Rf3, Rf4, Rf5, and Rf6 was successfully pyramided together through molecular marker selection, and R345 had the cleanest genetic background.

In order to evaluate the restoring ability of the three pyramided lines, we examined the pollen and spikelet fertility of the testcrosses derived from the pyramided restorer lines. All of the six HL-type hybrid crosses (HL-CMS lines Luohong-4A and Jiahong-2A crossed with R345, R346, and R348, respectively) showed the same pollen fertility phenotype as the typical HL-type hybrid combinations Luohong-4A/9311 and Jiahong-2A/9311. Similarly, all of the WA-type and two-line hybrid combinations showed the same pollen fertility as their respective controls, and reached to almost 100% (Figure 5A; Figure S6). Moreover, we observed that the spikelet fertility showed a similar trend as that of the pollen fertility (Figure 5B; Figure S7). These results showed that the pyramided lines R345, R346, and R348 could restore all types of sterile lines, implying these elite BPH-resistant lines had wide-spectrum restoring ability.

To determine if the newly developed BPH-resistant wide-spectrum restorer lines can be well used in hybrid rice production, 30 hybrid rice crosses were constructed by using R345, R346, and R348, and parent restorers 9,311 (for HL-CMS and two-line) and SH527 (for WA-CMS and two-line) to testcross HL-CMS lines Luohong-4A and Jiahong-2A, WA-CMS lines Yuexiang-A and Luofei-A, and two-line sterile lines BPH68S and Guangzhan4S, respectively. Rice yields and agronomic traits including plant height, panicle number, number of grains per panicle, spikelet fertility, and 1,000-grain weight were examined when matured (Table 2; Table S6). Relative to the control of HL-type hybrid rice Luohong-4A/9311, number of grains, number of grains per panicle, and yield per plant of new HL-type combinations Luohong-4A/R345, Luohong-4A/R346, and Luohong-4A/R348 significantly increased. Similarly, for the WA-CMS, hybrid crosses derived from Yuexiang-A and restorer lines R345, R346, and R348 performed better than that of the testcrosses derived from restorer line SH527, showing significant improvements in yield per plant, although the growth duration of the combinations from Yuexiang-A crossed with the three new restorer lines were prolonged. For the two-line hybrid rice, the crosses derived from the three new restorer lines also showed significantly yield advantage over the controls. Then, we tested the resistance of the hybrid rice derived from R345, R346, and R348 and HL-CMS line Yuetai-A (BPH-susceptible), respectively. When the Yuetai-A seedlings almost completely died, Yuetai-A/R345, Yuetai-A/R346, and Yuetai-A/R348 showed normal or only slightly endangered symptoms (Figure 2C), and even a higher resistant level than the resistant controls (Figure 2D). These results indicated that the hybrid rice derived from the new restorer lines not only had strong resistance to the brown planthopper, but also high yield potential.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.