Introgression and molecular tagging of Rf ₄, a new male fertility restoration gene from wild sunflower Helianthus maximiliani L.

Abstract

Cytoplasmic male sterility (CMS) and its fertility restoration (Rf) genes are critical tools for hybrid seed production to utilize heterosis. In sunflower, CMS PET1 and the associated Rf gene Rf ₁ is the only source extensively used in commercial hybrid production. The objective of this research was to develop new sources of CMS and fertility restorers to broaden the genetic diversity of hybrid seed production. We identified a new type of CMS, named as CMS GIG2, from an interspecific cross between Helianthus giganteus accession1934 and H. annuus cv. HA 89. Based on reactions to a set of standard Rf testers, CMS GIG2 is different from all previously reported CMS types, including the CMS GIG1 from another H. giganteus accession. We also identified an Rf gene for CMS GIG2 from wild species H. maximiliani accession 1631. The CMS GIG2 and its restoration gene were introduced into HA 89 background through recurrent backcross and single plant selection techniques. Genetic analysis revealed that the CMS GIG2-Rf system is controlled by a completely dominant gene, named as Rf ₄, and the gene additive and dominance effects were estimated as 39.9 and 42.2%, respectively, in the HA 89 background. The gene Rf ₄ was mapped onto linkage group 3 with simple sequence repeat (SSR) markers and RFLP-derived STS-marker, and is about 0.9 cM away from the SSR marker ORS1114 based on a segregation population of 933 individuals. The CMS GIG2-Rf ₄ system tagged by molecular markers provides an alternative genetic source for hybrid breeding in the sunflower crop.

Introduction

Sunflower (Helianthus annuus L.) is one of the most important oil-producing crops worldwide. High seed and oil yields are the main objectives of sunflower breeding programs, and the utilization of heterosis has been one of the most important tools for breeding high yield. Heterosis or hybrid vigor is often obtained through the use of complete sets of stable male-sterile lines (A line), maintainers (B line) and efficient restorer lines (R line) in higher plants. Use of cytoplasmic male sterility (CMS) eliminates the need for hand emasculation and ensures large-scale hybrid seed production. Since the first CMS reported in sunflower in 1969, 72 CMS sources have been identified (Serieys 2005). However, only the first CMS PET1 developed by Leclercq (1969) from H. petiolaris Nutt. has been extensively used to produce almost all commercial sunflower hybrids. Inheritance studies indicated that almost all restoration lines for CMS PET1 in sunflower breeding carry the same restorer gene Rf ₁ (Serieys 1996, 2005). The lack of new CMS and fertility restoration (Rf) genes for use in commercial hybrid production is partly due to the limited number of stable CMS cytoplasms and their corresponding Rf genes, as well as the time-consuming effort of converting lines to CMS and the incorporation of new Rf genes into adapted cultivated lines. In order to prevent cytoplasmic uniformity and reduce genetic vulnerability of sunflower hybrids to everchanging environmental stress and disease, alternative CMS and restorer sources need to be utilized in sunflower breeding programs.

The CMS phenotype can occur spontaneously or from induced mutation, as well as from the progenies of inter- or intraspecific crosses (Kaul 1988). The CMS system is a common phenomenon found in over 150 plant species (Laser and Lersten 1972). While a CMS line can be relatively easily produced, finding restorer genes has been proven much more difficult, especially when using the currently available cultivated sunflower. However, fertility restorer genes have been isolated from the same wild species population which provided the sterile cytoplasm or from other wild species (Leclercq 1969; Vranceanu and Stoenescu 1971, 1978). Kinman (1970) discovered the male fertility restorer gene Rf ₁ in the line T66006-2-1-B, which was selected from a composite cross involving the wild annual sunflower. Since then, many restoration lines, such as all the USDA-ARS RHA lines, RHA 271, 272, 273, 274, 275, 276, 278, 279 and 296 carry the Rf ₁ gene derived from T66006-2-1-B (Korell et al. 1992; Serieys 2005; Jan et al. 2002). An allelic test between T66006-2-1-B and MZ01398, an obsolete local cultivar, led to the discovery of a second major dominant gene Rf ₂ from MZ01398 (Vranceanu and Stoenescu 1971, 1978).

Numerous studies have indicated that sunflower male fertility restoration is generally controlled by a limited number of major restoration genes. In most cases, the Rf is under the control of a single dominant gene (Rf ₁), or two complementary dominant genes (Rf ₁ and Rf ₂) and modifiers (for review, see Serieys 1996, 2005). Since Rf ₂ is present in nearly all the inbred lines, including maintainer lines of CMS, only the Rf ₁ gene was considered to be the Rf gene needed in sunflower hybrid production (Dominguz-Gimeneze and Fick 1975; Leclercq 1984; Horn et al. 2003; Jan and Vick 2007). There has been confusion about the nomenclature of restoration genes in some of the early sunflower work where the “Rf ₂” was named for different restorers in different publications (Fick and Zimmer 1974; Vranceanu and Stoenescu 1978; Leclercq 1984). Recently, a Rf gene Rf ₃, which is different from both Rf ₁ and Rf ₂, was assigned to the confection restorer RHA 280 (Jan and Vick 2007).

The most studied CMS/Rf system in sunflower is the CMS PET1 and the corresponding restoration gene Rf ₁. CMS PET1 is associated with recombination events in the mitochondrial (mt) genome which involved an inversion/insert rearrangement resulting in the creation of a novel mt gene, orfH522 (Horn et al. 1991; Köhler et al. 1991; Laver et al. 1991), which encodes a ~16-kDa polypeptide. The level of ~16-kDa protein is significantly reduced in the presence of the dominant restorer genes (Monéger et al. 1994; Gagliardi and Leaver 1999). Although a number of mt genes associated with CMS have been characterized, the identification of Rf genes has proven elusive. To date, the only Rf genes to be identified and cloned are: maize (Zea mays L.) Rf ₂ for restoration of the cms-T-type male sterility (Cui et al. 1996; Liu et al. 2001); Rf of petunia (Petunia sp.) for the RM cytoplasm (Bentolila et al. 2002); Rfk ₁ (Rfo) of radish (Raphanus sativus L.) for the restoration of Ogura (ogu) CMS in Brassica napus (Brown et al. 2003; Desloire et al. 2003); and rice (Oryza sativa L.) Rf-1 for restoration of CMS-Boro II cytoplasm (Komori et al. 2004; Wang et al. 2006). However, map-based cloning of Rf genes has not been successful in sunflower because of the lack of precise mapping of the gene region.

In sunflower the gene Rf ₁, for restoration of PET1-CMS cytoplasm, was mapped to linkage group 6 of the RFLP map and the consensus map in 130 F₂ individuals from the cross CX × RHA 266 (Gentzbittel et al. 1995, 1999), to linkage group 2 of the RFLP genetic map in 93 F₂ individuals from the cross RHA 271 × HA 234 (Jan et al. 1998), and to linkage group 13 of the simple sequence repeat (SSR) map (Tang et al. 2002) using RAPD, AFLP and SSR markers with a 183-plant F₂ population of RHA 325 × HA 342 (Horn et al. 2003; Kusterer et al. 2005). All these markers linked with Rf ₁ provide an essential basis for genetic analysis of restoration genes and marker-assisted selection in sunflower breeding.

In 1995, an interspecific hybridization program was started in our lab with the objective of incorporating useful traits from wild Helianthus species into cultivated sunflower. A wild species H. giganteus was crossed with HA 89. After backcrossing five cycles with HA 89, a new alloplasmic CMS line with H. giganteus cytoplasm and HA 89 nucleus was developed. Subsequently, the search for its corresponding Rf gene and the incorporation into cultivated sunflower were undertaken. This study reports the discovery of the new CMS and its corresponding restoration gene, and the molecular tagging of the Rf gene on existing molecular maps. This work will ultimately lead to the cloning of the restorer allele, as well as facilitating the understanding of the genetics of the restoration of the male fertility and nuclear–cytoplasm interaction.

Materials and methods

Plant materials and breeding scheme

The perennial wild sunflower H. giganteus accessions 1934 (Ames 1934, PI 503250) and H. maximiliani accession 1631 and the cultivated sunflower variety HA 89 and its nuclear male sterility (NMS) mutant NMS HA 89 were used as parents in this research. These parental materials were obtained from the germplasm collections in our laboratory. Diploid H. giganteus 1934 was crossed with HA 89 (Fig. 1), and 12 F₁ seedlings were recovered through embryo rescue. All F₁ plants had 2n = 34 chromosomes and were treated with colchicine to induce chromosome doubling. At flowering, one of the 12 F₁ plants was identified as male-sterile, while the others were male-fertile, based on the morphology of the anthers and pollen stainability. The male-sterile F₁ plant was backcrossed with HA 89. All ten BC₁F₁ progenies from this cross were male-sterile and had 2n = 51–52 chromosomes, suggesting that chromosome doubling was successful. The BC plants were backcrossed with HA 89 repeatedly to reduce the chromosome number to 2n = 34, the chromosome number of the cultivated sunflower. All plants from the BC₁ to BC₅ generations were male-sterile, indicating that the CMS trait from H. giganteus 1934 was introduced into the HA 89 background. To search for the Rf gene for the CMS from H. giganteus 1934, BC₅F₁ plants were crossed with seven interspecific amphiploids (Amp) including NMS HA 89 × H. maximiliani 1631, H. cusickii × P21, H. atrorubens × HA 89, H. mollis × P21, H. grosseserratus × P21, H. pumilus × P21, and H. angustifolius × P21.

Fig. 1

Breeding scheme for introduction of the male sterile cytoplasm (CMS) and its fertility restoration (Rf) gene from the wild species into the HA 89 background

In order to test the relationship of the CMS from H. giganteus 1934 with the reported types of CMS, selected BC₄F₁ plants were crossed with a set of 19 standard lines (Armavir, HA 89, HA 290, P21, HA 821, Hopi Dye, Issanka, Luch, RCMG1, RCMG2, RCMG3, RHA 266, RHA 274, RHA 280, RHA 294, RHA 801, Seneca, Smena, and VNIIMK) in 2001. An average of 30 progeny plants from each cross were grown in the field in 2003 and were examined for pollen fertility at the flowering stage.

Genetic analysis

To understand the genetic control of the new CMS-Rf system, an initial F₂ segregation population of 113 individuals was generated from one single F₁ plant from the cross H. giganteus/7*HA 89/3/H. giganteus/6*HA 89//H. maximiliani 1631 (Amp) in 2005 (Fig. 1). Some of the male-fertile F₂ plants were selected for progeny testing to corroborate the genetic model developed based on the initial F₂ population. About 30 plants from each F₃ line grown under field conditions were identified for pollen fertility in 2006.

Pollen fertility for individual F₂ or F₃ plants was first identified visually after during flowering. The plants with normal, dehiscent anthers and abundant pollen were scored as male-fertile (F); otherwise, they were identified as male-sterile (S). To confirm the visual pollen fertility, all individuals were also microscopically examined for pollen stainability using Alexander’s method (Alexander 1969). Pollen grains from dehiscent anthers were suspended in a drop of the stain solution for 2–3 h, and normally stained pollen grains in three fields per slide were counted. The well filled fertile pollen grains will be stained red and the non-filled aborted pollen grains will be stained green. Approximately 400–500 pollen grains from each plant were examined to estimate the percentage of fertile pollen. Plants having an average of <5 and >30% pollen stainability were classified as male-sterile and male-fertile, respectively. Most male-sterile individuals were completely sterile with severe atrophy of anthers and non-stainable (aborted) pollens. The Chi-square test was used to determine the segregation pattern of male fertility phenotypes in the F₂ and F₃ populations.

Molecular mapping of the fertility restoration gene

The above F₂ population of 113 individuals was used to map the restoration gene on a low-resolution genetic map with SSR markers. Genomic DNA was extracted from fresh young leaves from individual plants according to the protocol described in Zhang et al. (2006). The bulked segregant analysis strategy (Michelmore et al. 1991) was used to screen for polymorphic SSR markers for the mapping population. Equal amounts of DNA from each of ten male-sterile and male-fertile F₂ individuals were bulked to develop the sterile and fertile pools. In total 200 SSR markers randomly selected from the 17 linkage groups (Tang et al. 2002; Yu et al. 2002, 2003) were screened for polymorphism between male-fertile and male-sterile pools. The polymorphic markers were used to genotype the 113 F₂ plants. In the genomic region containing the restoration gene, 11 single- or low-copy RFLP markers from linkage group 11 (Jan et al. 1998), which cross-referenced linkage group 3 of the SSR genetic map (Yu et al. 2003), were also selected to design sequence tagged sites (STS) primers to screen for more polymorphic markers. The STS primers were designed using PRIMER 3.0 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). These STS markers converted from RFLP sequences, tentatively designed using the “STS” prefix and then the original RFLP marker name. The sequence of the RFLP-based STS primer which produced polymorphic markers was as follows: STS10B1, 5′-AAGATCCCATCAAACCACCA-3′ and 5′-GGTCCACCTTGCGAAGATAA-3′.

PCR amplification was performed according to Tang et al. (2002) with minor modifications. The 15-μl mixture contained 1.5 μl of 10× PCR buffer, 0.6 μl of 25 mM MgCl₂, 1.2 μl of 2.5 μM dNTPs, 0.4 μl of 10 pmol each primer, 2 μl of 10 ng/μl DNA sample, 0.2 μl (5 unit/μl) Taq DNA polymerase (Qiagen, Valencia, CA, USA), and 0.1% Tween-20. An MJ Research single or a Bio-Rad dual 96-well thermal cycler was used along with the following “touchdown” profile: 95°C for 3 min; then 12 cycles of 94°C for 45 s, 64°C for 45 s, −0.5°C/cycle, and 72°C for 1 min; followed by 35 cycles at 94°C for 45 s, 58°C for 45 s, and 72°C for 1 min, with a final extension step at 72°C for 20 min. The amplified product was electrophoresed on a 6.5% denaturing polyacrylamide gel at 60 W for 2.0 h (0.5× TBE), and scanned with a Typhoon 9410 variable mode imager (Molecular Dynamics Inc., CA, USA) after staining with GelRed nucleic acid gel stain (Biotium, Inc., CA, USA).

An F₂-derived F₃ population of 1173 plants was grown in the field in 2007 to estimate the genetic distance of the restoration gene from the cosegregating markers detected with the initial mapping population. The selected F₂ plants were heterozygous for the Rf gene based on the progeny test. All F₃ plants were also visually and microscopically examined for pollen fertility as described above.

A partial linkage map for the restoration gene region was constructed using the software MAPMAKER/EXP version 3.0b (Whitehead Institute, Cambridge, MA, USA) (Lander et al. 1987). A constant LOD score of 3.0 and a maximum recombination frequency of 0.40 were used. The “error detection” command was used to check for mistakes in scoring and data entry, and for double crossover events. The Kosambi function was used to obtain the genetic distances in centiMorgan (cM) (Kosambi 1944). The linkage group and RFLP loci were named according to Tang et al. (2002) and Jan et al. (1998). Linkage group maps were drawn using the MapChart software (Voorips 2002).

The restoration gene, named as Rf ₄, was estimated for gene additive (a) and dominance (d) effects on pollen fertility based on the F₂-derived F₃ population. The estimations were made using the method described by Kearsey and Pooni (1996):

$$ a = 1/2{\text{M}}_{Rf4Rf4} - 1/2{\text{M}}_{rf4rf4} , $$

$$ d = {\text{M}}_{Rf4rf4} - 1/2{\text{M}}_{rf4rf4} - 1/2{\text{M}}_{Rf4Rf4} , $$

where, M _Rf4Rf4, M _Rf4rf4, and M _rf4rf4 are mean percentages of pollen stainability for the genotypes Rf ₄ Rf ₄, Rf ₄ rf ₄, and rf ₄ rf ₄, respectively. Standard errors for the parameters a and d were estimated as:

$$ s_a = 1/2\left( {s_{Rf4Rf4}^2 + s_{rf4rf4}^2 } \right)^{1/2} , $$

$$ s_d = \left( {s_{Rf4rf4}^2 + 1/4s_{Rf4Rf4}^2 + 1/4s_{rf4rf4}^2 } \right)^{1/2} , $$

where, s ² _Rf4Rf4 , s ² _Rf4rf4 , and s ² _rf4rf4 are variances of the means M _Rf4Rf4, M _Rf4rf4, and M _rf4rf4, respectively. Significance of the a and d estimates was determined by Student’s t test.

Results

Isolation of a male sterility cytoplasm and screening the corresponding Rf gene

The initial CMS plant was obtained from the F₁ hybrid from the cross between H. giganteus 1934 and HA 89 to isolate the male sterility cytoplasm (Fig. 1). The F₁ hybrid was backcrossed with the recurrent parent HA 89 for five generations. All the BC progenies remained male-sterile. An alloplasmic CMS line with mostly HA 89 genome was recovered in the BC₅ generation, suggesting that a CMS source derived from H. giganteus 1934 had been isolated and incorporated into the cultivated sunflower HA 89 background.

BC₅F₁ plants were pollinated by a set of 19 genetically diverse cultivated lines to test the Rf pattern for the CMS from H. giganteus 1934. All the progenies of the crosses were male-sterile, suggesting the lack of restorer genes for the CMS H. giganteus 1934 and also indicating that this CMS from H. giganteus 1934 is different from all previously reported CMS types, including CMS GIG1 which can be restored by RHA 280, RHA 294 and RHA 801 (Serieys 1996). Therefore, it was concluded that this CMS derived from H. giganteus 1934 was a new CMS source. According to the FAO codification, the new CMS source was designated as CMS GIG2.

To further search for the Rf gene for the CMS GIG2, BC₅F₁ plants were crossed with seven interspecific Amp (see “Materials and methods”). The results showed that Rf genes were identified in four Amp involving wild species H. maximiliani 1631, H. atrorubens, H. grosseserratus, and H. angustifolius, but the other Amp behaved as a CMS GIG2 maintainer. Surprisingly, a single male-fertile plant with 2n = 34 was obtained after crossing a BC₅F₁ plant with the Amp NMS HA 89 × H. maximiliani 1631 (Fig. 1), while all other nine progenies of the same cross were also male-fertile but had 2n = 51 chromosomes, indicating the presence of Rf genes in this amphiploid. Because of the time-consuming process of continuous backcrossing of the 2n = 51 male-fertile progenies to reduce the 2n chromosome number and to maintain male-fertility, we focused on the 2n = 34 male-fertile plant to further study the inheritance of the restoration genes. As mentioned above, the standard set of 19 cultivated lines carrying different Rf genes for various CMS sources, such as RHA 266 and RHA 274 carrying the Rf ₁ gene, HA 89 and HA 821 carrying the Rf ₂ gene, and RHA 280 carrying the Rf ₃ gene (Jan and Vick 2007) were not able to restore CMS GIG2. This indicated that the Rf gene derived from amphiploid H. maximiliani 1631 is a novel Rf gene and is different from Rf ₁, Rf ₂, and Rf ₃.

Inheritance of the male fertility restoration gene

A single F₁ plant with normal male fertility was grown in a greenhouse to develop a 113-plant F₂ segregation population to understand the inheritance of CMS GIG2 and its Rf gene. All individual plants were classified as fertile or sterile based on the quantity of stainable pollen from dehiscent anthers. In some cases, no pollen was visible to the naked eye, but under the microscope, pollen grains were observed after staining with Alexander’s stain. Therefore, we examined all individuals visually and microscopically. Of the 113 F₂ plants, the segregation of 32 homozygous male-fertile, 49 heterozygous male-fertile and 32 male-sterile individuals fit the expected 1:2:1 ratio (0.3 < P < 0.5) (Table 1). In addition, progeny tests conducted with 48 F₂ male-fertile lines in the field showed that 30 families were heterozygous for male fertility and 18 families were homozygous, suggesting their parental F₂ plants were heterozygous and homozygous, respectively (Table 1). The 30 F₃ families derived from heterozygous male-fertile F₂ plants segregated into 607 male-fertile and 207 male-sterile, fitting an expected 3:1 ratio closely (0.7 < P < 0.8), while 18 F₃ families derived from the homozygous male-fertile F₂ plants produced 492 male-fertile plants and no male-sterile plant, fitting an expected 1:0 ratio. All the results from the F₂ and F_2:3 generations indicated that a single dominant gene, named as Rf ₄, controls the Rf of CMS GIG2.

Table 1 Fitness test for segregation ratios of the pollen sterility/fertility trait controlled by the Rf ₄/rf ₄ locus in the F₂ and advanced F₃ populations (Fig. 1)

Molecular mapping and genetic effects of Rf ₄

Two hundred ORS SSR primer pairs randomly chosen from 17 linkage groups of the SSR genetic map (Tang et al. 2002; Yu et al. 2002, 2003) were used for screening polymorphisms between male-fertile and male-sterile bulks. After PCR amplification, 182 primer pairs produced clear bands, 18 primer pairs produced unclear bands or no band, and only nine gave polymorphic bands between the two bulks. The lower polymorphism obtained between the two bulks compared with other strategies may be due to their similar genetic backgrounds after repeated backcrosses. However, those identified polymorphic markers are expected to be more likely linked to the target gene Rf ₄. Our results through genotyping the F₂ population indicated that ORS1114 on linkage group 3 of the SSR genetic map was linked to the Rf ₄ locus. Subsequently, 15 ORS markers from linkage group 3 and 11 RFLP-derived STS primers from linkage group 11 (Jan et al. 1998) were selected to screen two bulks and F₂ individuals. In total, four DNA markers near the Rf ₄ locus were identified, including two codominant markers (ORS1114 and ORS13) and two dominant markers (ORS822-3 and STS10B1). All the polymorphic markers were mapped to the distal region of linkage group 3 of the SSR genetic map (Tang et al. 2002; Yu et al. 2002, 2003) (Fig. 2). The order of loci in this region is similar to the SSR genetic map. SSR marker ORS13 and ORS1114 cosegregated with the Rf ₄ locus, while STS10B1 and ORS822-3 mapped 3.1 cM and 31.7 cM, respectively, from the Rf ₄ locus on the same side.

Fig. 2

The map position of the male fertility restoration gene Rf ₄. The linkage groups constructed with different mapping populations were aligned based on the shared markers. a Part of linkage group 3 of the RFLP-SSR map (Yu et al. 2003); b partial linkage map for the Rf ₄ region developed based on a population of 113 F₂ plants; c partial linkage map for the Rf ₄ region based on a segregation population of 933 plants; and d part of linkage group 3 of the SSR map (Tang et al. 2003). Genetic distances are shown in centiMorgan (cM)

Since no recombinants between the Rf ₄ locus and ORS1114 marker were detected using 113 F₂ plants, a large population of 1173 F₃ plants derived from heterozygous F₂ plants was genotyped using ORS1114. The pollen stainability of 1173 individuals showed a bimodal distribution with a clear valley from 5 to 30% (Fig. 3). The classification of male-fertile and male-sterile plants followed the standard of <5 and >30% as male-sterile and male-fertile, respectively. The segregation ratio of 883 male-fertile and 290 male-sterile agreed with the expected 3:1 ratio (0.8 < P < 0.9) (Table 1). These results further confirmed a single dominant gene control of the Rf of CMS GIG2. The recombinant frequency between ORS1114 and Rf ₄ was low, with seven recombinants among 933 F₃ individuals identified. As a result, the SSR marker ORS1114 was 0.9 cM away from the Rf ₄ locus (Fig. 2).

Fig. 3

Distribution of pollen stainability for three genotypes of the Rf ₄ fertility restoration locus in the F₂-derived F₃ population. The Rf ₄ locus was represented by its tightly linked marker ORS1114 (Fig. 2d)

Gene additive and dominance effects of the Rf ₄ locus on pollen fertility were about 40 and 42%, respectively, as estimated based on the tightly linked marker ORS1114 in the F₃ population (Table 2). Both the gene additive and dominance effects were similar in magnitude, indicating that Rf ₄ is a completely dominant gene controlling pollen fertility restoration in the HA 89 genetic background.

Table 2 Gene additive (a) and dominance (d) effects of the Rf ₄ locus on pollen fertility estimated based on the tightly linked marker ORS1114 in the F₂-derived F₃ population grown in 2007

Discussion

Enrichment of CMS sources and male fertility restorers

To broaden the genetic base and reduce the vulnerability of the cultivated sunflower to diseases and environmental stress, identification of new CMS sources and the associated restorers has been a continuous objective for commercial sunflower production in the world (Serieys 1996). Even though 72 CMS sources have been reported in the literature, inheritance studies of restoration were performed for only 19 CMS sources (Serieys 2005). Moreover, characterization and molecular mapping of the Rf genes was conducted only for the Rf ₁ gene of CMS PET1, which is extensively used in hybrid sunflower production.

The newly identified CMS GIG2 is clearly different from CMS PET1, and represents a new source of male-sterile cytoplasm. More importantly, a single dominant Rf gene Rf ₄ derived from a diploid perennial wild species H. maximiliani 1631 can efficiently restore the male fertility of CMS GIG2. Field experiments in 2007 showed that the pollen stainability of the enlarged population ranged from 33 to 100%, with 86% of the plants possessing pollen stainability >70%, and nearly half (47%) of the fertile plants possessing >90% pollen stainability (Fig. 3). However, due to environmental factors, some individuals showed poor pollen fertility both in the greenhouse or field.

While CMS GIG2 cytoplasm was crossed with 19 cultivated lines involving different restoration genes, all the progenies were male-sterile, suggesting that the restorer of CMS GIG2 cytoplasm did not exist in the cultivated germplasms evaluated. However, such genes were present in four Amp involving wild species. This is consistent with the conclusion that male fertility restoration of CMS is present at a higher frequency in wild species or their related hybrids than in current cultivars (Vranceanu and Stoenescu 1978). The reasons might be due to alloplasmic CMS lines containing the cytoplasm from wild species and the nucleus from the cultivated species, and this nuclear–cytoplasm conflict can most likely be eliminated by introducing specific nuclear genes from wild species.

Previous studies indicated that the pollen fertility restoration in sunflowers is genetically controlled by a single Rf gene or a series of two independent complementary genes in different crosses (Serieys 1996). However, polygenic restorers, which have two or more Rf genes, should be avoided in breeding programs because of their instability in different environments and their complicated inheritance (Vranceanu and Stoenescu 1978). Thus, a single dominant Rf gene is highly preferred for sunflower breeding. Rf ₄ is a single major gene for restoration of CMS GIG2 and pollen fertility can be recovered to 100%. It can be concluded that CMS GIG2 and the Rf ₄ gene are a new potential system that could be used to substitute for the current PET1 cytoplasm, if necessary, for hybrid sunflower production because of its stable male sterility, efficient male fertility restoration, and a single dominant gene controlling fertility restoration. In addition, closely linked codominant SSR marker ORS1114 can easily distinguish the homozygous and heterozygous alleles, and will facilitate marker-assisted selection for the Rf ₄ gene in breeding programs. Nevertheless, additional studies of the CMS GIG2 cytoplasm for agronomic traits such as seed yield, oil quality, and disease resistance would be necessary to assess its suitability for commercial sunflower hybrids.

In the search for the restoration gene for CMS GIG2, BC₅F₁ plants with 2n = 34 were crossed as female with several Amp (2n = 68). Theoretically, all their progenies should be 2n = 51. However, a single male-fertile plant was obtained after crossing with an amphiploid NMS HA 89 × H. maximiliani 1631, and the chromosome number was 2n = 34. At present, we do not have adequate data to explain how or why this happened. Whether the restoration gene from the amphiploid H. maximiliani was incorporated into the CMS BC₆ generation through a rare recombination event or chromosome translocation or others is unknown. Further cytogenetic research will be needed to understand the mechanism.

Molecular tagging of Rf ₄

The fertility restorer gene Rf ₁ of CMS PET1was mapped to linkage group 13 of the SSR genetic map (Tang et al. 2002) by using SSR marker ORS1030 and other markers (Horn et al. 2003; Kusterer et al. 2005), and to linkage group 2 of the RFLP genetic map (Jan et al. 1998), which is cross-referenced to linkage group 13 of the SSR genetic map (Gedil et al. 2001; Yu et al. 2003). Thus, Rf ₁ was assigned to linkage group 13 of the SSR genetic map. In the present studies, the Rf ₄ locus was mapped to linkage group 3 of the SSR genetic map, which also cross-referenced to linkage group 11 of the RFLP genetic map (Jan et al. 1998). Therefore, the Rf ₄ gene needed for fertility restoration of CMS GIG2 is the first fertility restorer gene to be identified on linkage group 3 of the SSR genetic map. It is obviously different from Rf ₁ of CMS PET1 based on mapping analysis. This further explains why all the restorer lines for the CMS PET1 do not restore the pollen fertility for CMS GIG2.

The isolation of the Rf genes through a map-based cloning strategy has been successfully used in various plant species (Bentolila et al. 2002; Brown et al. 2003; Wang et al. 2006). Fine mapping of the target gene region and constructing a large insert DNA library are two essential steps. BAC/BIBAC libraries with 8.9 × genome coverage have recently been constructed by Feng et al. (2006). A closely linked SSR marker ORS1114 was mapped <1 cM from Rf ₄ with 933 individuals. It can be expected that the use of closely linked marker ORS1114 with backcross breeding will produce near isogenic lines (NILs) of the Rf ₄ gene. Further fine mapping covering the Rf ₄ gene region and screening the BAC/BIBAC libraries will greatly facilitate cloning of the Rf ₄ gene in the near future.