Introduction

Rice (Oryza sativa) is the most widely consumed staple food, feeding more than half of the global population. Compared with inbred varieties, hybrid rice lines produce 10–20% higher yields, and hybrid rice is cultivated on 57% of the rice planting areas in China (Yuan 2014). Hybrid rice lines include three- and two-line hybrids, which are developed by applying cytoplasmic male sterility (CMS) and environmentally sensitive genic male sterility, respectively (Cheng et al. 2007; Huang et al. 2014). A CMS line, a maintainer line, and a restorer line, which carries a fertility restorer gene (Rf), are required to develop three-line hybrids. Three representative CMS types, including wild abortive (WA), Honglian (HL), and Chinsurah Boro II (BT), are commonly used to generate three-line hybrid rice. The WA-type CMS belongs to the sporophytic CMS/Rf system, and the HL- and BT-type CMS belong to the gametophytic CMS/Rf system, and these three CMS types are considered to have different maintainer–restorer relationships (Chen and Liu 2014; Huang et al. 2014; Li et al. 2007). However, there are potential problems regarding the relationships among the three CMS/Rf systems. For example, WA-type restorers can restore the fertility of BT- and HL-type CMS lines, but not vice versa (Zhu et al. 2010). Additionally, BT- and HL-type CMS lines exhibit similar maintainer–restorer relationships (Tan et al. 2008; Zhang et al. 2016). Therefore, additional studies should be conducted to determine the roles of mapped and cloned Rf genes during the fertility restoration of different types of CMS lines, which may help to determine the relationships among the three CMS/Rf systems.

To date, Rf3 and Rf4 for WA-CMS lines, Rf5 and Rf6 for HL-CMS lines, and Rf1a/Rf1b for BT-CMS lines have been mapped, and except for Rf3, all of them have been cloned (Hu et al. 2012; Huang et al. 2012, 2015; Komori et al. 2004; Tang et al. 2014; Wang et al. 2006; Zhang et al. 1997). Cloning of these genes revealed that Rf5 is the same as Rf1a, and Rf6 can restore the fertility of BT-CMS lines (Hu et al. 2012; Huang et al. 2015; Komori et al. 2004; Wang et al. 2006; Zhang et al. 2016). All the cloned Rf genes encode pentatricopeptide repeat (PPR) proteins, and the amino acid sequence encoded by Rf4 is highly similar to that of RF1A (Hu et al. 2012; Huang et al. 2015; Komori et al. 2004; Tang et al. 2014; Wang et al. 2006). Previously, we determined that Rf1a, a BT-type fertility restorer gene, can partially restore the fertility of HL-type japonica CMS lines, and that Rf5 and Rf6 can restore the fertility of BT-type CMS lines (Zhang et al. 2016, 2017). Above all, Rf5 (Rf1a) and Rf6 are involved in the fertility restoration of BT-type and HL-type CMS lines, which are useful to characterize the similar maintainer–restorer relationships among the BT- and HL-type CMS lines. Rf5 and Rf6 were identified from ‘93-11’, an elite restorer line for HL-type and two-line hybrid rice, which exhibits high yield and superior quality. In breeding practice, ‘93-11’ is not a restorer for WA-type CMS lines, but was found to carry some Rf genes with minor effects for the fertility restoration of WA-type CMS (Tang et al. 2014). The improved ‘93-11’ without minor Rf genes for WA\CMS would be termed as a maintainer to develop WA-type hybrids. Therefore, identifying minor Rf genes for WA-type CMS in ‘93-11’ is of great interest. Until now, there have been few studies of minor Rf genes for WA-type CMS. Considering that Rf5 and Rf6 are located in the multigene clusters that encode PPR proteins, these targeted Rf genes for WA-type CMS lines located around the Rf5 and/or Rf6 loci should be studied as a priority.

In the present study, we used a marker-assisted selection (MAS) strategy to develop near-isogeneic lines (NILs) for Rf5 and Rf6, the polygene pyramid lines that combine Rf5 with Rf6, and bred ‘WA-NipponbareA’ (WA-NipA), a WA-type japonica CMS line using ‘Nipponbare’ as the maintainer. Then, we constructed several populations to obtain plants carrying WA-type cytoplasm and the different genotypes at the Rf5 and Rf6 loci, and the morphologies of the pollen grains and anthers of these plants were observed. We provided information regarding whether the Rf genes located around the Rf5 and Rf6 loci are involved in the fertility restoration of WA-type CMS lines, and whether these genes function to reduce the WA352 (the mitochondrial gene conferring CMS-WA) transcript levels. The results will promote the breeding of WA-type CMS lines to develop three-line hybrids.

Materials and methods

Plant materials

‘Nipponbare’ (Nip), a genome sequenced japonica cultivar and is the maintainer for BT-CMS, HL-type CMS, and WA-CMS lines. ‘93-11’, a genome sequenced typical indica cultivar, is the restorer for BT-type CMS and HL-type CMS lines, but is not a restorer for WA-type CMS lines. In our previous study, a set chromosome segment substitution lines (CSSLs) from the cross between ‘93-11’(donor parent) and Nip (recipient parent) was developed, and these lines were genotyped using resequencing (Zhang et al. 2017). BT-NipA, which has the same nuclear background as Nip, but has a BT-type sterile cytoplasm, was used as the female parent to cross with the CSSLs, and two Rf genes for BT-type CMS, Rf5 and Rf6 from ‘93-11’, were identified and finely mapped (Zhang et al. 2017). Four markers closely linked with Rf5 (i.e., STS10-27 and STS10-16) and Rf6 (i.e., STS8-4 and STS8-32) were developed. Using MAS, NILs for Rf5 and Rf6 were developed and designed as NILRf5and NILRf6, respectively. Polygene pyramid lines (PPLs) PPLRf5+Rf6 that combined Rf5 with Rf6 were then developed using MAS. For breeding WA-NipA, Nip was used as the male parent in a cross with WA-LiuqianxinA (a WA-type japonica CMS line), which was followed by six backcrosses with Nip from 2011 to 2015. In the rice-growing season of 2015, WA-NipA plants were crossed with NILRf5, NILRf6, and PPLRf5+Rf6, and the F1 hybrids were sterile. Thus, the corresponding F1 plants were further crossed with NILRf5, NILRf6, and PPLRf5+Rf6, respectively, to generate the WA-NipA/NILRf5//NILRf5, WA-NipA/NILRf6//NILRf6, and WA-NipA/PPLRf5+Rf6//PPLRf5+Rf6 populations to obtain plants harboring different genotypes at the Rf5 and Rf6 loci. In the rice-growing season of 2016, the above three backcross populations were planted, and the target plants were identified using MAS.

Field experiment

All plant materials used in this study were sown in May 20, and the 30-days-old seedlings were transplanted into the field at the experimental farm of Yangzhou University (32°24′′N, 119°26′′E) in Jiangsu province. Each plot consisted of 2–20 rows separated by 25 cm, with each row consisting of 10 plants, separated by 20 cm, and the management of the field experiments was conducted according to the normal procedures for rice.

Fertility scoring

Mature anthers of WA-NipA plants and plants with different genotypes in the WA-NipA/NILRf5//NILRf5, WA-NipA/NILRf6//NILRf6, and WA-NipA/PPLRf5+Rf6//PPLRf5+Rf6 populations were observed under an optical microscope. Pollen grains from mature anthers were stained with a 1% iodine–potassium iodide (I2–KI) solution and observed using an optical microscope. At the flowering stage, two plants for each genotype and two panicles from each plant were bagged. Natural and bagged spikelet fertility levels of one plant were measured as the average seed-setting rates, which were calculated by counting the filled and unfilled grains of two panicles harvested 20 days after flowering from one plant.

DNA extraction and PCR amplification

Genomic DNA was isolated from the fresh leaves using the cetyltrimethylammonium bromide method (Rogers and Bendich 1985). The markers closely linked with Rf5 and Rf6 were developed in our previous study (Zhang et al. 2017), and primers were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). The molecular marker analysis was carried out in a 20-µL reaction containing 2.0 µL of reaction buffer, 0.1 mmol/L of each dNTP, 1.0 U Taq polymerase, 0.2 µmol/L primer, 20 ng of template DNA, and ultra-pure water to a final volume of 20 µL. The amplification reaction consisted of one cycle at 94 °C for 4 min; followed by 30 cycles at 94 °C for 45 s, 55 °C for 45 s, 72 °C for 50 s; with a final extension step at 72 °C for 5 min. The amplification products were separated by electrophoresis through a 3.0% (w/v) agarose gel containing ethidium bromide, and visualized using the GEL DOC 1000 system (Bio-Rad Laboratories, Hercules, CA, USA).

RNA isolation and quantitative real-time PCR

Total RNA was extracted from leaves, culms, and panicles using a Plant RNA Kit (Tiangen, Beijing, China). The Perfect Real Time Prime Script RT reagent (Takara, Dalian, China) was used for first-strand cDNA synthesis in a 20-µL reaction containing 5 µg of total RNA. The quantitative real-time PCR (qPCR) was conducted in triplicate (i.e., with RNA from three independent extractions) using a CFX96 Real-Time PCR system (Bio-Rad Laboratories, Hercules, CA, USA). Data were analyzed according to the 2−∆∆CT method (Livak and Schmittgen 2001). The mitochondrial atp6 gene was used as an internal reference to determine the relative WA352 expression levels. The primers used in this study are listed in Supplementary Table S1.

Results

Construction of NILRf5, NILRf6, and PPLRf5+Rf6

During 2004–2013, a population comprising 127 CSSLs (N1–N127) was developed using crossing and back crossing, assisted by 352 molecular markers, and the genotypes of these CSSLs were identified using a high-throughput resequencing strategy (Zhang et al. 2011, 2017). BT-NipA was crossed with these CSSLs, and two Rf genes, Rf5 and Rf6, were identified and finely mapped (Zhang et al. 2017). Based on resequencing data, N91 (Fig. 1a) harboring an Rf6-containing region between STS 8-4 and STS 8-32 (IRGSP-1.0 position 0.38 Mb and 0.56 Mb) was selected to cross with BT-NipA and N114 (Fig. 1b) harboring an Rf5-containing region between STS 10-27 and STS 10-16 (IRGSP-1.0 position 18.67 Mb and 19.17 Mb), and the testcross F1s from N91 and N114 produced offspring by self-pollination. Using MAS, the plants carrying the genotypes of Rf6Rf6 were selected from the BT-NipA/N91 F2 self-crossed progeny, and were designed as NILRf6. Except for the introgression segment covering the Rf5 locus, N114 carried another introgression segment (IRGSP-1.0 position 2.12 Mb and 10.02 Mb) on Chr.10, and two markers (RM25222 and RM25384) were selected to detect this introgression segment. Using MAS, plants specific for the Rf5-containing region and carrying the genotypes of Rf5Rf5 were selected from the BT-NipA/N114 F2 population, and were designed as NILRf5. In 2014, BT-NipA/N91F1 plants were crossed with BT-NipA/N114 F1 plants, and MAS experiments were conducted to obtain plants harboring the genotype Rf5rf5Rf6rf6 and only the introgression segments covering the Rf5 and Rf6 loci in the BT-NipA/N91//BT-NipA/N114 population. In 2015, an F2 population comprising 200 plants was generated, and plants harboring the genotype Rf5Rf5Rf6Rf6 in the F2 population were obtained and designed as PPLRf5+Rf6. All these materials had a similar genetic background to Nip, and exhibited similar agronomic traits to Nip.

Fig. 1
figure 1

Genome constitution of CSSLs based on resequencing data analyses. a Genome constitution of N91 (carrying Rf6). Each blue line and red line represent single nucleotide polymorphisms of ‘Nipponbare’ and ‘93-11’, respectively. The red box in Chr.8 indicates a substituted fragment derived from 93-11. b Genome constitution of N114 (carrying Rf5). The red boxes in Chr.10 indicate substituted fragments derived from 93-11. (Color figure online)

Full size image

Generation of plants carrying different genotypes at the Rf5 and Rf6 loci

During 2010–2015, WA-NipA was bred using Nip as the maintainer with continuous backcrossing. In 2015, NILRf5, NILRf6, and PPLRf5+Rf6 were used as the male parents to cross with WA-NipA. The resulting F1 plants were backcrossed with NILRf5, NILRf6, and PPLRf5+Rf6 plants, respectively, to produce three BC1F1 populations, including WA-NipA/NILRf5//NILRf5, WA-NipA/NILRf6//NILRf6, and WA-NipA/PPLRf5+Rf6//PPLRf5+Rf6. There were 47, 59, and 58 plants in these three populations, respectively. Plants harboring one of eight genotypes (i.e., Rf5rf5rf6rf6, Rf5Rf5rf6rf6, rf5rf5Rf6rf6, rf5rf5Rf6Rf6, Rf5rf5Rf6rf6, Rf5rf5Rf6Rf6, Rf5Rf5Rf6rf6, and Rf5Rf5Rf6Rf6) were detected from the BC1F1 populations using MAS, and there were more than three plants for each genotype (Table 1). The plants with different genotypes exhibited similar agronomic traits to those of WA-NipA (Fig. 2a).

Table 1 Numbers of plants harboring different genotypes in the BC1F1 populations
Full size table
Fig. 2
figure 2

Plant and anther morphology, and pollen grain stainability, of plants with WA-type cytoplasm and different genotypes at the Rf5 and Rf6 loci. a Morphology of plants with the WA-type cytoplasm and different genotypes at the Rf5 and Rf6 loci. b Anthers of plants with WA-type cytoplasm and different genotypes at the Rf5 and Rf6 loci. c Pollen grains of plants with WA-type cytoplasm and different genotypes at the Rf5 and Rf6 loci. Pollen grains were stained with 1% I2–KI. Scale bars = 10 cm (a), 1 mm (b), and 50 μm (c). Genotypes (left to right): WA-NIPA, WA-NIPA-Rf5rf5rf6rf6, WA-NIPA-Rf5Rf5rf6rf6, WA-NIPA-rf5rf5Rf6rf6, WA-NIPA-rf5rf5Rf6Rf6, WA-NIPA-Rf5rf5Rf6rf6, WA-NIPA-Rf5Rf5Rf6rf6, WA-NIPA-Rf5rf5Rf6Rf6, and WA-NIPA-Rf5Rf5Rf6Rf6

Full size image

Rf genes around the Rf5 and Rf6 loci are involved in fertility restoration of WA-type CMS lines

In 2016, we analyzed the anther and pollen grain morphologies, as well as the seed-setting rates, of all plants in WA-NipA/NILRf5//NILRf5, WA-NipA/NILRf6//NILRf6, and WA-NipA/PPLRf5+Rf6//PPLRf5+Rf6 populations. Plants carrying the genotypes of Rf5rf5rf6rf6, Rf5Rf5rf6rf6, rf5rf5Rf6rf6, rf5rf5Rf6Rf6, and Rf5rf5Rf6rf6 produced milky white, slender, and stunted anthers that contained shrunken pollen grains. In contrast, the anthers of WA-NipA-Rf5Rf5Rf6rf6, WA-NipA-Rf5rf5Rf6Rf6, and WA-NipA-Rf5Rf5Rf6Rf6 plants were yellow and engorged, with pollen grains that could be stained with 1% potassium iodide (Fig. 2b, c). None of these plants were able to set seeds on the bagged panicles, and had extremely low seed-setting rates (≤ 1.1%) on the unbagged panicles. These results suggested that Rf genes around the Rf5 and Rf6 loci are involved in the fertility restoration of WA-type CMS and are influenced by a gene dosage effect during the recovery of normal anther and pollen grain morphologies in the WA-CMS lines. Therefore, we concluded that Rf genes around both the Rf5 and Rf6 loci are responsible for the weak capability of ‘93-11’ to restore the fertility of WA-type CMS lines. Thus, it would be necessary to breed WA-type maintainers with the genetic background of ‘93-11’ by eliminating the Rf genes around the Rf5 and Rf6 loci.

Rf genes function to reduce the WA352 transcript level

To examine the mechanism underlying the fertility restoration of WA-CMS lines by Rf genes around the Rf5 and Rf6 loci, we first used qPCR to examine the expression patterns of WA352, which is a mitochondrial gene responsible for the male sterility of WA-CMS lines (Luo et al. 2013). The qPCR data for the different WA-NipA plant tissues indicated that WA352 was highly expressed in young panicles, but expressed at low levels in the other examined tissues (Fig. 3a), which was the same as that described in a former study (Luo et al. 2013; Tang et al. 2014). We then investigated WA352 expression in the young panicles of plants carrying different genotypes. The WA352 transcript levels were lower in plants carrying the Rf5 and/or Rf6 loci than that in WA-NipA plants. Additionally, many Rf genes were associated with lower relative WA352 expression levels, which was consistent with the morphology of the pollen grains and anthers (Fig. 3b). Therefore, Rf genes around Rf5 and Rf6 loci are involved in decreasing WA352 mRNA levels to suppress WA352-mediated male sterility, which is similar to the role of Rf4 (Tang et al. 2014).

Fig. 3
figure 3

Functional analysis of Rf5 and Rf6 in the fertility restoration of WA-type CMS. a Relative WA352 transcript abundance, as analyzed by qPCR. b Relative WA352 transcript levels in the young panicles of plants with WA-type cytoplasm and different genotypes at the Rf5 and Rf6 loci. The atp6 gene was used as an internal control. The data are shown as the mean ± standard deviation (n = 3)

Full size image

Discussion

Since the 1970s, three-line hybrid rice varieties have been used widely in China. The application of the three-line hybrid breeding system depends on the identification of CMS lines containing recessive Rf alleles and a corresponding Rf allele-containing restorer line. To date, WA-, HL-, and BT-type cytoplasms were the typical sterile cytoplasms used for breeding CMS lines, and five Rf genes including Rf4, Rf5 and Rf6, Rf1 (Rf1a/Rf1b) have been isolated from rice restorer plants (Hu et al. 2012; Huang et al. 2015; Komori et al. 2004; Tang et al. 2014; Wang et al. 2006). In breeding practice, it used to be thought that there is a one-to-one match between each type CMS line and the restorer line, and each Rf functions specifically in a particular type of CMS system, which has a unique phenotype caused by the specific type of mitochondrial DNA and corresponding nuclear gene(s) (Chen and Liu 2014; Tang et al. 2017). In indica, WA-type CMS plants usually exhibit stunted anthers, typical abortive pollen grains, and no seed setting on bagged panicles. ‘93-11’ is not a restorer for WA-type CMS lines, and the allele of Rf4, the major fertility restorer gene for WA-type CMS in ‘93-11’, was identified to be rf4 (Tang et al. 2014). However, ‘93-11’ was identified to carry some minor Rf genes of WA-type CMS, because the F1 plants from the crosses between WA-type CMS lines and ‘93-11’ exhibited normal anthers and stained pollen grains. In a previous study, only Rf5 and Rf6, the fertility restoration genes for HL-type CMS, were isolated from ‘93-11’. It was believed that there might be no genes around Rf5 and Rf6 loci for the fertility restoration of WA-type CMS in ‘93-11’. In the present study, we generated plants carrying the background of WA-NipA and different genotypes at the Rf5 and Rf6 loci. As WA-NipA is a stable WA male sterile line, exhibiting stunted anthers and typical abortive pollen grains, it presumably does not carry Rf genes for WA-type CMS. WA-NipA-Rf5rf5Rf6Rf6, WA-NipA-Rf5rf5Rf6Rf6, and WA-NipA-Rf5Rf5Rf6Rf6 plants exhibited normal anthers and stained pollen grains, and these plants carried only the introgression segments covering the Rf5 and Rf6 loci from ‘93-11’. These results indicated that Rf genes in ‘93-11’ having minor effects on the fertility restoration of WA-type CMS lines are located around the Rf5 and Rf6 loci, which would be very useful to develop pairs of WA male sterile and maintainer lines with the background of ‘93-11’. These findings also provided new insights into the maintainer–restorer relationships among different types of CMS.

It was previously reported that WA352 is the candidate CMS-associated gene in WA-type CMS (Luo et al. 2013). Rf4, the major Rf gene for WA-type CMS, encodes a PPR-containing protein that generally binds to RNA and is involved in RNA metabolism in mitochondria or plastids (Lurin et al. 2004). Rf4 was demonstrated to reduce the level of WA352 transcripts during fertility restoration of WA-type CMS (Tang et al. 2014). In previous studies, a cluster of PPR genes was observed at the Rf5 locus on chromosome 10 and the Rf6 locus on chromosome 8 (Hu et al. 2012; Komori et al. 2004; Tang et al. 2014; Wang et al. 2006). In the present study, qPCR analysis showed that the WA352 transcript level varied between WA-NIPA and other lines carrying the Rf5 and/or Rf6 loci. The WA352 transcript levels were decreased in plants carrying the Rf5 and/or Rf6 loci, indicating that the Rf genes around the Rf5 and Rf6 loci contributed to the decrease in WA352 transcript levels. Accordingly, we presumed that Rf genes located around the Rf5 and Rf6 loci that influenced the fertility restoration of WA-type CMS lines in ‘93-11’ might encode PPR-containing proteins. The cloning of rice fertility restorer genes showed that Rf5 is the same gene as Rf1a. Additionally, Rf5 and Rf6, both of which were identified from HL-type indica restorer, and similar to Rf1a from CMS-BT, could restore the fertility of BT-type CMS lines (Hu et al. 2012; Huang et al. 2012, 2015; Zhang et al. 2017). These results suggested that the functions of Rf5 and Rf6 are not specific to restore the fertility of HL-type CMS lines. Furthermore, the amino acid sequences encoded by Rf4 and Rf5 are highly similar, and Rf6 is only one member of the cluster at Rf6 locus has the function to restore fertility to HL-type CMS line (Huang et al. 2015; Tang et al. 2014). Therefore, we hypothesized that the Rf genes around the Rf5 and Rf6 loci that influenced the fertility restoration of WA-type CMS lines might actually be Rf5 and Rf6. More studies are required to test this hypothesis.

In the CMS/Rf systems, two or more fertility restorer genes are required to produce viable pollen. For example, the fertility recovery of maize CMS-T was influenced by the cumulative action of Rf1 and Rf2a (Wise et al. 1999), and the fertility was partially recovered in the presence of genes Rf8 and Rf* with Rf2a (Dill et al. 1997). In rice, for the full recovery of fertility of WA-CMS lines, Rf3 on chromosome 1 and Rf4 on chromosome 10 are required (Ahmadikhah and Karlov 2006; Jing et al. 2001; Luo et al. 2013; Ngangkham et al. 2010; Yao et al. 1997; Zhang et al. 1997). In the present study, we also determined that the functions of Rf genes around the Rf5 and Rf6 loci are mediated by dosage effects during the fertility restoration of WA-CMS lines. These results may be useful to further characterize the restoration of fertility in WA-type CMS lines, and provide new information for the breeding of maintainers and restorers for WA-type CMS.