Findings

Hybrid rice has an average yield advantage of 15% to 20% over inbred cultivars. Most commercial hybrid rice has been developed based on a three-line system, namely A (CMS line), B (maintainer line), and R (restorer line). B lines must lack any Rf genes and the seeds of CMS lines are multiplicated by crossing A x B. F1 hybrid seeds are produced by crossing A x R. The resulting F1 hybrid plants are fertile, because an Rf gene is provided by the R line. The most predominantly utilized CMS system is known to employ wild abortive-type CMS (WA-CMS), which has accounted for about 90% of the three-line hybrids produced in China and 100% of the hybrids developed outside of China (Sattari et al. 2007; Huang et al. 2014). A major technical handicap in the development of hybrid rice using WA-CMS is a limited source of maintainer lines, as many Indica Group elite cultivars are known to carry restorer genes for WA-CMS lines; thus, they cannot be used as maintainer lines (Virmani 1994). For example, IR 24 and IR 64 are restorer lines for WA-CMS lines, and they are used as male parents for hybrid seed production (Jing et al. 2001; Cai et al. 2013).

We obtained CMS lines of IR 24 and IR 64 when we employed the Chinese wild rice (CW)-type CMS/Restorer of fertility 17 (Rf17) system (Toriyama and Kazama 2016). CW-CMS is derived from Oryza rufipogon Griff. strain W1 (Katsuo and Mizushima 1958). A CMS-associated mitochondrial gene is orf307 (Fujii et al. 2010). Pollen grains of the CW-CMS lines IR 24, IR 64, and a Japonica Group cultivar, Taichung 65, accumulate starch and look morphologically normal but lack germination ability (Fujii and Toriyama 2005; Toriyama and Kazama 2016). Fertility of the CW-CMS lines is gametophytically restored by a single nuclear gene, Rf17, which is identified to be a reduced expression allele of RETROGRADE-REGULATED MALE STERILITY (Fujii and Toriyama 2009; DDBJ accession number AB481199). To broaden the combination of male and female parents for hybrid rice production, seven elite Indica Group rice cultivars were tested for their acquisition of CMS by using the CW-CMS/Rf17 system.

The IR 64 nuclear background restorer line CWR-IR 64, which carried CW-cytoplasm and Rf17Rf17 (Toriyama and Kazama 2016), were successively backcrossed with the elite Indica Group cultivars of NSIC Rc 160 (a high-quality eating cultivar), NSIC Rc 240 (a high-yielding cultivar) in the Philippines, Ciherang (a high-yielding cultivar) in Indonesia, BRRI dhan 29 (a high-yielding cultivar in Bangladesh), Pusa Basmati (an aromatic cultivar with short culm) in India, Samba Mahsuri (a high-yielding fine-grain rice cultivar) in India, and NERICA-L-19 (a high-yielding cultivar with the IR 64 genetic background) in Africa. A local Basmati line was also used as a recurrent parent, which was provided by the National Institute of Agrobiological Sciences Genebank (Tsukuba, Japan) as WRC42 (Kojima et al. 2005). The presence or absence of the Rf17 allele was detected by a SNP in the promoter region 2286 bp upstream of the initiation codon; Rf17 carried T, while rf17 carried A (Fujii and Toriyama, 2009; Toriyama and Kazama 2016). Plants with rf17rf17 were selected as CMS candidates, while plants with Rf17Rf17 were selected as restorer candidates after the self-pollination of plants with Rf17rf17 (Fig. 1).

Fig. 1
figure 1

Flowchart of developing new CMS lines and restorer lines. [ms-CW] indicates the cytoplasm

Full size image

The plants were grown in a biotron as previously described (Ohnishi et al. 2013). The filled and unfilled grains of bagged panicles were counted to calculate the seed setting rate. As shown in Table 1, the backcrossed lines with rf17rf17 at generations BC3F1 and BC4F1 were completely sterile for NSIC Rc 160, NSIC Rc 240, Ciherang, BRRI dhan 29, and NERICA-L-19. Those at generations BC2F1 and BC3F1 were also completely sterile for Pusa Basmati. In contrast, the BC3F1 plants of Samba Mahsuri with rf17rf17 set seeds segregating a plant with a higher seed setting rate of 79.6%. A progeny of this plant (BC3F2) also showed a higher seed setting rate of 77.8%.

Table 1 Seed setting rates (%) of backcrossed lines and pollen parents in 2017 and 2018
Full size table

The backcrossed lines of NSIC Rc 160, NSIC Rc 240, Ciherang, BRRI dhan 29, and NERICA-L-19 with Rf17Rf17 at generations BC2F2 and BC2F3 showed higher seed setting rates comparable to those of each pollen parent, indicating that the fertility was recovered by the presence of Rf17 (Table 1). The plants that were backcrossed with local Basmati three times (BC3F3) followed by the backcrossing with Pusa Basmati (BC1F2) also showed higher seed setting rates of approximately 80% in the presence of Rf17 (Table 1).

Plants were also cultivated in an isolated glasshouse and a paddy field at Tropical Agriculture Research Front, Japan International Research Center for Agricultural Sciences located subtropical Ishigaki island from March to July, 2019, together with their pollen parents. Seed setting rates were evaluating using open pollinated panicles without bagging. For CMS lines, BC4F1 plants with rf17rf17 of NSIC Rc 160, NSIC Rc 240, BRRI dhan 29, and NERICA-L-19 and Pusa Basmati are completely male sterile except for one panicle each of NSIC Rc160 and NERICA-L-19, which set 1 and 2 seeds, respectively, when grown in an isolated glass house (Additional file 1: Table S1). For restorer lines, BC2F3 plants of NSIC Rc 160, NSIC Rc 240, Ciherang, BRRI dhan 29, and NERICA-L-19, and plants backcrossed with local Basmati three times followed by backcrossing once with Pusa Basmati showed higher seed setting rates, which values were not significantly different from those of their pollen parents, except for NERICA-L-19 showing variation of lower seed setting rates depending on individual BC2F3 plants (Additional file 1: Table S1). A seed setting rate of lines backcrossed with Samba Mahsuri was 52.6% for BC3F2 plants with rf17rf17, while 74.0% for BC2F3 plants with Rf17Rf17 (Additional file 1: Table S1), suggesting Samba Mahsuri might have a Rf gene with weak function of fertility restoration.

To know whether Samba Mahsuri carried another allele of the Rf17 gene, a 5-kb genomic region including the 3.9-kb promoter and 0.5-kb coding sequences of the RF17 gene of Samba Mahsuri and IR 64 were amplified by PCR using the following primers: 5′-AAGAGATGACGGTGCAGTTC-3′, 5′-TCGTTCACCACGGTAGATAGACTCAT-3′, 5′-CCCACATCTTCTCCTTGCATAATCC-3′, and 5′-GGGGCTCCCTAGGTGGCTAA − 3′. The nucleotide sequence of Samba Mahsuri was completely identical to that of IR 64. Because IR 64 does not have fertility restoration abilities, the rf17 allele of IR 64 and Samba Mahsuri was non-functional for fertility restoration. This result indicated that Samba Mahsuri had a new fertility restorer gene different from the Rf17 gene.

In conclusion, we produced CMS lines of NSIC Rc 160, NSIC Rc 240, Ciherang, BRRI dhan 29, NERICA-L-19, and Pusa Basmati, which did not set any seeds. The fertility was fully recovered by the presence of Rf17. The CW-CMS/Rf17 system will be useful for the production of CMS lines of various Indica Group rice cultivars and for hybrid rice breeding.