- Split View
-
Views
-
Cite
Cite
Andrea Garavito, Romain Guyot, Jaime Lozano, Frédérick Gavory, Sylvie Samain, Olivier Panaud, Joe Tohme, Alain Ghesquière, Mathias Lorieux, A Genetic Model for the Female Sterility Barrier Between Asian and African Cultivated Rice Species, Genetics, Volume 185, Issue 4, 1 August 2010, Pages 1425–1440, https://doi.org/10.1534/genetics.110.116772
- Share Icon Share
Abstract
S1 is the most important locus acting as a reproductive barrier between Oryza sativa and O. glaberrima. It is a complex locus, with factors that may affect male and female fertility separately. Recently, the component causing the allelic elimination of pollen was fine mapped. However, the position and nature of the component causing female sterility remains unknown. To fine map the factor of the S1 locus affecting female fertility, we developed a mapping approach based on the evaluation of the degree of female transmission ratio distortion (fTRD) of markers. Through implementing this methodology in four O. sativa × O. glaberrima crosses, the female component of the S1 locus was mapped into a 27.8-kb (O. sativa) and 50.3-kb (O. glaberrima) region included within the interval bearing the male component of the locus. Moreover, evidence of additional factors interacting with S1 was also found. In light of the available data, a model where incompatibilities in epistatic interactions between S1 and the additional factors are the cause of the female sterility barrier between O. sativa and O. glaberrima was developed to explain the female sterility and the TRD mediated by S1. According to our model, the recombination ratio and allelic combinations between these factors would determine the final allelic frequencies observed for a given cross.
INTRINSIC postzygotic reproductive barriers are a very common and important phenomenon, driving the establishment and conservation of species (Widmer et al. 2009). They are observed as a reduction in the hybrids' fitness, evidenced by a reduced viability or fertility. So far, a limited number of genes acting as postzygotic barriers have been identified in Drosophila (Ting et al. 1998; Presgraves et al. 2003; Brideau et al. 2006; Masly et al. 2006; Phadnis and Orr 2009; Tang and Presgraves 2009), and in other animal systems (Wittbrodt et al. 1989; Lee et al. 2008; Mihola et al. 2009), but also in Arabidopsis (Josefsson et al. 2006; Bomblies et al. 2007; Bikard et al. 2009) and rice (Chen et al. 2008; Long et al. 2008).
In rice improvement, hybrid inviability and sterility are major obstacles for the common utilization of closely related species in breeding programs, impairing the exploitation of the rich genetic diversity found within the Oryza sativa complex (known as genome group AA), and the beneficial effect of the high level of heterosis observed in the F1 plants. One of the most relevant examples of this strong limitation comes from the African cultivated rice species O. glaberrima Steud. This species represents an interesting source of drought tolerance (Sarla and Swamy 2005), weed competitiveness (Dingkuhn et al. 1998), and nematode and virus resistances (Ndjiondjop et al. 1999; Soriano et al. 1999). Several of these traits have been already mapped (Lorieux et al. 2003; Ndjiondjop et al. 2003); however, their introduction to O. sativa has been hampered by the strong sterility barrier between the two species (Sano et al. 1979). This barrier is the result of several loci that might interact or act separately to render the F1 hybrids pollen sterile and partially female fertile. Among these, the S1 locus has the strongest effect over the fertility of the hybrids (Sano 1990).
Despite the strong hybrid sterility, obtaining fertile plants derived from O. sativa × O. glaberrima crosses has been possible by performing successive backcrosses followed by selfing (Ghesquière et al. 1997). Nevertheless, this fertility recovery is associated with the presence of the homozygote O. glaberrima S1 allele (S1g), and is lost again when recrossing with O. sativa (Heuer and Miezan 2003). Furthermore, a strong transmission ratio distortion (TRD) of markers linked with S1, in favor of the O. glaberrima alleles, results as a consequence of the systematic elimination of the O. sativa alleles from the descendants (Ghesquière et al. 1997; Doi et al. 1998b; Lorieux et al. 2000; Aluko et al. 2004).
As an alternative intended to unlock the full potential of O. glaberrima to rice breeding, the “iBridges” project, supported by the Generation Challenge Program (http://www.generationcp.org) aims to develop a set of O. sativa × O. glaberrima hybrids that would be fertile when crossed with O. sativa. To develop these interspecific bridges, genetic factors affecting their fertility must be identified and characterized so as to better understand the nature of the sterility barrier. Recently, the complex nature of the S1 locus was suggested, and the male component was mapped to an interval equivalent to 45 kb on the genome of O. sativa cv. Nipponbare (Koide et al. 2008c). However, the nature and the genomic positions of the components affecting female fertility are still unknown.
In this article we describe the fine genetic and physical mapping of the female component of the S1 locus, using an approach based on the evaluation of the degree of TRD caused by the elimination of the S1 O. sativa alleles. This methodology enabled us to identify a 27.8-kb interval on the short arm of the O. sativa chromosome 6 as the location of the female component of the S1 locus. Furthermore, the presence of additional factors involved in the sterility barrier mechanisms was predicted, and their localization was deduced. Direct comparison of the S1 interval in O. glaberrima with the O. sativa orthologous sequence revealed the presence of several significant differences and a possible candidate factor. On the basis of our data, a model involving the epistatic interaction of the additional factors with S1 was developed to explain the female sterility and the TRD found in the interspecific hybrids.
MATERIALS AND METHODS
Plant materials:
BC1F1 populations were obtained from crosses between the O. sativa accessions IR64 (O. sativa ssp. indica), Caiapo (CA) (tropical japonica), Curinga (CU) (tropical japonica) and Nipponbare (NIP) (temperate japonica), and the three O. glaberrima accessions, TOG5681, Ac103544 (MG12), and CG14. Backcrosses were made as follows: IR64/TOG5681//IR64 (referred as IR64xTOG5681), CA/MG12//CA (CAxMG12), CU/CG14//CU (CUxGG14), and NIP/MG12//NIP (NIPxMG12). A first set of 125 individuals from the cross IR64xTOG5681 was generated, sowed, and genotyped. Later on, 114 individuals from the cross CAxMG12, 101 from CUxCG14 and 59 from NIPxMG12 were generated, as well as a second set of 334 plants from the IR64xTOG5681 cross. Additional plants issued from crosses NIP/TOG5681//NIP (12) and CA/TOG5681//CA (34) were also genotyped. Seeds were first germinated in vitro in MS medium (Toshio and Folke 1962). Ten-day-old seedlings were transplanted in the greenhouse and afterward brought to the field at the International Center for Tropical Agriculture (CIAT, Cali, Colombia) headquarters.
Molecular marker analysis:
To derive a new O. sativa × O. glaberrima genetic map, 140 markers belonging to version 1 of the Universal Core Genetic Map (UCGM) for rice (Orjuela et al. 2010) were evaluated in the first 125 BC1F1 individuals of the IR64xTOG5681 population. Marker saturation was carried out in the S1 region on chromosome 6 (Sano 1990; Lorieux et al. 2000) (supporting information, Table S1). The rest of the chromosome was also saturated at a lower density. SSR markers were chosen on the basis of the physical location on the O. sativa cv. Nipponbare chromosome 6 (TIGR v. 6 data, (Ouyang et al. 2007). Primers were synthesized according to the SSR information available (International Rice Genome Sequencing Project 2005). Further saturation near S1 was performed using new additional markers (Table S2) found with the MISA script (Thiel et al. 2003), and InDel markers were designed by comparing the Nipponbare sequence with the orthologous CG14 sequence (see below). Primers were designed with primer3 (Untergasser et al. 2007), confirmed to amplify a unique genomic region in Nipponbare using the Primer Blaster tool (Droc et al. 2009), and evaluated in the populations.
PCR reactions were carried out with an annealing temperature and magnesium concentration ranging from 1.5 to 2.5 mm optimized for each marker (Table S1 and Table S2), as previously described (Orjuela et al. 2010). Electrophoreses of PCR products were made in 4% agarose and revealed with ethidium bromide for polymorphisms greater than 12 bp, and in a Li-Cor sequencer (Li-Cor Bioscienes) for smaller polymorphisms, using an M13 tag (IRD700 and IRD800).
Statistical analysis and mapping approach:
Genetic map construction was made using the MapDisto v. 1.7 program (http://mapdisto.free.fr/). A minimal LOD score of 3.0 was retained to identify the linkage groups. The order of the markers was determined using the Order, Ripple, and Bootstrap commands. The Kosambi function (Kosambi 1944) was used to convert the recombination fractions to centimorgans (cM) and vice versa. The deviation from the expected Mendelian ratio (1:1) for each locus was determined by segregation χ2 tests. Genotyping error candidates were detected using the tools provided in the Color Genotypes module, with a threshold of 0.005.
To fine map the female component of the S1 locus, we used the estimation of the female TRD (fTRD) as an indicator of the closeness of a given marker to the factor causing the distortion. Indeed, it is expected that the maximum fTRD occurs at the location of the transmission ratio distortion locus (TRDL), the precision of the mapping depending on the population size, the marker saturation, and the level of TRD. We conducted a simulation study to estimate the precision of this approach (see File S1, Simulation.doc). Simulation and computations were performed using the MapDisto program (http://mapdisto.free.fr/). fTRD was measured using the kf statistic, which represents the proportion of plants inheriting the overrepresented allele after female segregation of the heterozygote. Because the TRD on the S1 locus in our BC1F1 populations depends solely on female gametogenesis in the F1, the kf and the k statistics are equivalent and were thus calculated as the number of heterozygotes over the total number of BC1F1 individuals.
Physical mapping of the S1 locus:
For the O. glaberrima cv. CG14, several BAC clones of the S1 region were chosen on the basis of the available finger print contig (FPC) data from the OMAP project (Kim et al. 2008). Clones showing strong sequence identity in at least one of their BAC end sequences (BES) to the orthologous Nipponbare region were identified by the BLAST algorithm (Altschul et al. 1997). PCR confirmation was made with the same markers used for the genotyping of the S1 region. A set of BAC clones was chosen, and their overlapping was confirmed by PCR amplification with primers derived from their BES.
Sequence analysis, comparison and gene annotation method:
BAC sequencing was performed by the Sanger method. The final BAC sequence was first analyzed using BLAST against different public and local plant nucleotide and protein databases. Coding regions were ab initio predicted using the FGENESH program (Salamov and Solovyev 2000) and then confirmed by comparative analysis with annotated gene models and proteins in O. sativa cv. Nipponbare, downloaded from the TIGR database (Ouyang et al. 2007). Predicted gene structures were manually evaluated by alignment with rice EST and full-length cDNA (FLcDNA) public sequences (Kikuchi et al. 2003). Detailed analysis was performed with the EMBOSS Analysis software (Rice et al. 2000).
Putative transposable elements (TEs) were first identified and annotated by RepeatMasker searches (http://www.repeatmasker.org) against local databases of rice TEs downloaded from the REPBASE (Jurka et al. 2005), and from the TIGR repeat database (Ouyang and Buell 2004). De novo prediction of TEs was performed according to the structure of the different class of TEs. The final annotation of the BAC sequences was performed using the Artemis tool (Rutherford et al. 2000), and comparison with the Nipponbare genome was accomplished using dot-plot alignments with the Dotter software (Sonnhammer and Durbin 1995).
Calculation of nonsynonymous and synonymous nucleotide substitution rates:
Pairwise alignments of orthologous genes between O. glaberrima and O. sativa were conducted using the Needle tool (Rice et al. 2000) to estimate the degree of gene structure conservations. Rates of nonsynonymous and synonymous nucleotide substitutions (Ka/Ks) were calculated using the PAML package (Yang 1997) implemented in the PAL2NAL program (Suyama et al. 2006). Orthologous genes with clear distinct gene structures were removed from the analysis.
RESULTS
Genetic map of the IR64xTOG5681 cross, marker segregation and its association with sterility loci between O. sativa and O. glaberrima:
Using a population of 125 individuals from the IR64xTOG5681 cross, a survey of the entire genome was completed to identify markers with fTRD and to compare their colocalization with previously described sterility loci between the two species. Using markers from the UCGM, an 1879.56-cM map length was obtained (Figure 1). Possible inversions were detected in centromeric regions (chromosomes 1, 5, 7, 8, 9, 11) and in the short arm of chromosome 10 (Figure S1). Several sites showing a significant deviation from the 1:1 expected segregation (P < 0.05) were also found (Figure 1). The O. glaberrima alleles were overrepresented on chromosomes 3, 6, and 11, while the O. sativa alleles were more common than expected on chromosomes 1, 2, and 7.
On the basis of marker information available from Gramene (Liang et al. 2008), markers surrounding the different sterility loci between the two species were localized on the map. All but two sites presenting fTRD (long arms of chromosomes 2 and 6) colocalized with the different loci described so far (Figure 1). The strongest fTRD (P < 10−5) was found at the S1 locus. Nonetheless, several sterility loci did not colocalize with markers showing fTRD. These results corroborate the relation between TRD and sterility loci.
Mapping of the female component of the S1 locus by measuring the TRD:
Recently, Koide and Colleagues (2008c) proved that the TRD of the Wx locus in pollen produced by heterozygote plants for the S1 region was due to the abortion of gametes carrying the highly linked O. sativa S1 allele (denoted in this article as S1s and equivalent to S1a (Sano 1990)). They also inferred that the S1 locus has at least two components, one controlling male TRD (mTRD) and the other one fTRD. Additionally, the male component of the locus was mapped to a 45-kb interval in the chromosome 6 of Nipponbare.
To map the female component of the S1 locus, we applied an approach based on the direct measurement of marker deviation from the expected Mendelian ratio, caused by the elimination of female gametes carrying the S1s allele. Due to meiotic recombination, the degree of TRD of a given marker would be inversely proportional to the genetic distance between it and the distorter factor. By constructing a highly saturated genetic map around S1, and measuring precisely the fTRD observed for each marker, it would be possible to restrict the genomic location of the distorter factor, since the markers closest to S1 would show the highest fTRD.
To examine the fTRD level, a saturated genetic map was obtained by two succeeding marker saturations in the S1 region, the first one with approximately 1 SSR/100 kb, carried within the segment limited by the markers showing the highest fTRD in a previous work (RM190 and RM204, Lorieux et al. 2000), and the second one with 1 SSR/40 kb within the interval bounded by the two markers with the highest TRD (RM19357 and RM19369). Using the 125 IR64xTOG5681 BC1F1 population, an fTRD peak of kf = 0.970 was found, limited by markers RM19357 and RM19367 (Figure 2A). On the basis of these data, the location for the female component of S1 was attributed to an interval equivalent to 160 kb in the Nipponbare genome.
Mapping of the female component of the S1 locus in the O. sativa ssp. japonica background:
As the O. sativa genetic background affects the level of female sterility caused by S1 (Sano 1990; Koide et al. 2008c), it is necessary to analyze crosses involving both indica and japonica to better understand the nature of the sterility barrier. For that purpose, three O. sativa ssp. japonica × O. glaberrima BC1F1 populations were developed, and the chromosome 6 genetic map was constructed and saturated.
A peak in the fTRD of kf = 0.912 (limited by RM19353 and RM19367), kf = 0.911 (RM190–RM5199), and kf = 0.881 (RM19357–RM19367), was found for crosses CAxMG12 (Figure 2B), CUxCG14 (Figure 2C), and NIPxMG12 (Figure 2D), respectively. Also, different levels of significant TRD were found in the centromeric region according to the cross. This marker evaluation shows that the fTRD caused by S1 is located in the same genomic region in different O. glaberrima and O. sativa indica or japonica accessions, as it involves the same molecular markers for all of the crosses. However, the degree of female gamete elimination varies between the different japonica populations and is lower than the one found for the indica cross.
Construction of a physical map around the S1 locus in O. glaberrima:
To compare the region between the two species, a physical map of the S1 locus of O. glaberrima was established. Twenty-six BAC clones from the O. glaberrima cv. CG14 BAC library were chosen, on the basis of data available from the OMAP project (Kim et al. 2008). PCR amplifications with the same markers from the genetic map around S1 confirmed that 14 belonged to the genomic region. A minimum tiling path (MTP) of 8 clones (covering 800 kb in the Nipponbare genome) was established after PCR confirmations of the overlaps between the clones (Figure 3A). On the basis of the amplification profile, clone OG-BBa0049I08 was selected for complete sequencing, as it contains the markers with the highest fTRD. The final 165-kb sequence for the BAC (EMBL accession number FP340543) was obtained with 12.5× coverage and an error rate below 1 base/100 kb.
Fine mapping of the female component of S1:
The factor causing fTRD was mapped in the same genomic region for the two O. sativa backgrounds. To map it more finely, new polymorphic SSR and InDel markers were designed by sequence comparison between the two species. High saturation with new markers (close to one marker per gene) was completed in the existing populations and in 334 new BC1F1 individuals from the IR64xTOG5681 cross.
An fTRD peak limited by markers RM19357 and RM19359 was found for both IR64xTOG5681 (kf = 0.980) and CAxMG12 (kf = 0.921) crosses (Figure 3, B and C). For the CUxCG14 and NIPxMG12 crosses, the values for kf were 0.911 (RM190–RMC6_22639) and 0.881 (RM19357–RM19367), respectively (Figure 3, D and E). Consequently, the segment between markers RM19357 and RM19359 was defined as the location for the factor affecting female fertility of the hybrids.
To reduce the size of the candidate interval, the recombination events were examined around the fTRD peak using additional markers designed for that purpose. Surprisingly, recombination occurred within the same small segment for the three individuals limiting each side of the interval (Figure 3F). This approach allowed us to reduce the S1 female component interval to 27.8 kb in O. sativa and to 50.3 kb in O. glaberrima. This segment is positioned within the interval described to contain the male component of the locus (Koide et al. 2008c). The fine mapping of the female component of S1 indicates that the factor(s) responsible for the female and male semi-sterility of the interspecific hybrids are located in the same interval, in both O. sativa subspecies.
Chromosome 6 genotype of viable S1s female gametes produced by the F1 hybrids:
Overall, only a small number of female S1s gametes produced by the F1 hybrid are able to survive the allelic elimination exerted by S1. To characterize these gametes, the complete markers' genotype of the S1s/S1s plants (each originated from a viable S1s female gamete) was examined in detail all along chromosome 6. The 34 S1s/S1s BC1F1 plants from the four populations under study, as well as two individuals from two additional small populations, were analyzed in detail (Table 1). All of the plants shared an O. sativa chromosomal segment of at least 7.3 cM (r = 0.0725) around S1s, equivalent to 886 kb in the genome of Nipponbare (Table 1). The probability of observing no recombination events between the bounds of this segment in 36 lines is
Populations . | Minimal O. sativa inherited region around S1s . | N . | N (S1s) . | f (S1s) . | σ f (S1s) . | f ± σ f(S1s) . | f ± 2σ f(S1s) . |
---|---|---|---|---|---|---|---|
indica × O. glaberrima | |||||||
IR64 × TOG5681 | RM19350–RM3183 | 459 | 9 | 0.020 | 0.006 | 0.013–0.026 | 0.007–0.033 |
japonica × O. glaberrima | RM190–RM3805 | 274 | 25 | 0.091 | 0.017 | 0.073–0.108 | 0.056–0.126 |
Caiapo × MG12 | 114 | 9 | 0.079 | 0.025 | 0.054–0.104 | 0.028–0.129 | |
Curinga × CG14 | 101 | 9 | 0.089 | 0.028 | 0.061–0.117 | 0.032–0.146 | |
Nipponbare × MG12 | 59 | 7 | 0.119 | 0.042 | 0.077–0.161 | 0.034–0.203 | |
Additional individuals | |||||||
Caiapo × TOG5681 | 34 | 1 | 0.029 | 0.029 | 0.000–0.058 | 0.000–0.087 | |
Nipponbare × TOG5681 | 12 | 1 | 0.083 | 0.080 | 0.004–0.163 | 0.000–0.243 | |
O. sativa × O. glaberrima | RM19350–RM3805 | 779 | 36 | 0.046 | 0.008 | 0.039–0.054 | 0.031–0.061 |
Populations . | Minimal O. sativa inherited region around S1s . | N . | N (S1s) . | f (S1s) . | σ f (S1s) . | f ± σ f(S1s) . | f ± 2σ f(S1s) . |
---|---|---|---|---|---|---|---|
indica × O. glaberrima | |||||||
IR64 × TOG5681 | RM19350–RM3183 | 459 | 9 | 0.020 | 0.006 | 0.013–0.026 | 0.007–0.033 |
japonica × O. glaberrima | RM190–RM3805 | 274 | 25 | 0.091 | 0.017 | 0.073–0.108 | 0.056–0.126 |
Caiapo × MG12 | 114 | 9 | 0.079 | 0.025 | 0.054–0.104 | 0.028–0.129 | |
Curinga × CG14 | 101 | 9 | 0.089 | 0.028 | 0.061–0.117 | 0.032–0.146 | |
Nipponbare × MG12 | 59 | 7 | 0.119 | 0.042 | 0.077–0.161 | 0.034–0.203 | |
Additional individuals | |||||||
Caiapo × TOG5681 | 34 | 1 | 0.029 | 0.029 | 0.000–0.058 | 0.000–0.087 | |
Nipponbare × TOG5681 | 12 | 1 | 0.083 | 0.080 | 0.004–0.163 | 0.000–0.243 | |
O. sativa × O. glaberrima | RM19350–RM3805 | 779 | 36 | 0.046 | 0.008 | 0.039–0.054 | 0.031–0.061 |
f (S1s), observed S1s frequency; σ, standard deviation of f (S1s); confidence intervals of the S1s/S1s BC1F1 plants in indica and japonica crosses.
Populations . | Minimal O. sativa inherited region around S1s . | N . | N (S1s) . | f (S1s) . | σ f (S1s) . | f ± σ f(S1s) . | f ± 2σ f(S1s) . |
---|---|---|---|---|---|---|---|
indica × O. glaberrima | |||||||
IR64 × TOG5681 | RM19350–RM3183 | 459 | 9 | 0.020 | 0.006 | 0.013–0.026 | 0.007–0.033 |
japonica × O. glaberrima | RM190–RM3805 | 274 | 25 | 0.091 | 0.017 | 0.073–0.108 | 0.056–0.126 |
Caiapo × MG12 | 114 | 9 | 0.079 | 0.025 | 0.054–0.104 | 0.028–0.129 | |
Curinga × CG14 | 101 | 9 | 0.089 | 0.028 | 0.061–0.117 | 0.032–0.146 | |
Nipponbare × MG12 | 59 | 7 | 0.119 | 0.042 | 0.077–0.161 | 0.034–0.203 | |
Additional individuals | |||||||
Caiapo × TOG5681 | 34 | 1 | 0.029 | 0.029 | 0.000–0.058 | 0.000–0.087 | |
Nipponbare × TOG5681 | 12 | 1 | 0.083 | 0.080 | 0.004–0.163 | 0.000–0.243 | |
O. sativa × O. glaberrima | RM19350–RM3805 | 779 | 36 | 0.046 | 0.008 | 0.039–0.054 | 0.031–0.061 |
Populations . | Minimal O. sativa inherited region around S1s . | N . | N (S1s) . | f (S1s) . | σ f (S1s) . | f ± σ f(S1s) . | f ± 2σ f(S1s) . |
---|---|---|---|---|---|---|---|
indica × O. glaberrima | |||||||
IR64 × TOG5681 | RM19350–RM3183 | 459 | 9 | 0.020 | 0.006 | 0.013–0.026 | 0.007–0.033 |
japonica × O. glaberrima | RM190–RM3805 | 274 | 25 | 0.091 | 0.017 | 0.073–0.108 | 0.056–0.126 |
Caiapo × MG12 | 114 | 9 | 0.079 | 0.025 | 0.054–0.104 | 0.028–0.129 | |
Curinga × CG14 | 101 | 9 | 0.089 | 0.028 | 0.061–0.117 | 0.032–0.146 | |
Nipponbare × MG12 | 59 | 7 | 0.119 | 0.042 | 0.077–0.161 | 0.034–0.203 | |
Additional individuals | |||||||
Caiapo × TOG5681 | 34 | 1 | 0.029 | 0.029 | 0.000–0.058 | 0.000–0.087 | |
Nipponbare × TOG5681 | 12 | 1 | 0.083 | 0.080 | 0.004–0.163 | 0.000–0.243 | |
O. sativa × O. glaberrima | RM19350–RM3805 | 779 | 36 | 0.046 | 0.008 | 0.039–0.054 | 0.031–0.061 |
f (S1s), observed S1s frequency; σ, standard deviation of f (S1s); confidence intervals of the S1s/S1s BC1F1 plants in indica and japonica crosses.
In the cross with the lowest S1s transmission (IR64xTOG5681), the 9 S1s/S1s plants share a much larger O. sativa chromosomal segment (with the exception of a heterozygote <7 cM singleton in one line, from RM19414 to RM204). This interval, limited by markers RM19350 and RM3183, starts 3.3 cM before and ends 48.2 cM after S1. The probability of observing no recombination events between the bounds of this segment in 9 lines
Sequence annotation and comparison of the S1 locus between O. glaberrima and O. sativa:
The 164,664-bp sequence of the O. glaberrima BAC OG-BBa0049I08 was carefully annotated and analyzed. TEs and predicted genes are shown in Table S3 and Table S4 and in Figure S2, respectively.
To study the sequence divergence of the S1 locus, fine comparative genomic approaches were carried out between CG14 and Nipponbare. Nucleotide sequences surrounding the male component of the locus mapped by Koide et al. (2008c) (between markers E0506 and E1920) were extracted from both genomes and analyzed in detail (Figure 4). Alignments of the 44.8- and 68.8-kb sequences showed an overall conservation of nucleotide sequences, with breaks that expanded the O. glaberrima segment (Figure 4). This 1.5× increase was mainly due to the insertion of 15.3 kb from 19 TEs, suggesting a higher local accumulation in comparison to the whole BAC sequence. Within the O. glaberrima 50.38-kb interval of the female component of the S1 locus, seven genes and three pseudogenes were predicted (Table 2). Among the predicted genes, two are hypothetical proteins (49I08.7 and 49I08.75), three exhibited similarities with “early nodulins,” one exhibited with ribosome biogenesis regulatory proteins (49I08.8), and one exhibited with F-box proteins (49I08.11). The three pseudogenes and the two hypothetical proteins were located between predicted genes 49I08.6 and 49I08.11; three of them (49I08.75, 49I08.9, and 49I08.10) corresponded to nonautonomous chimeric Mutator-like transposable elements (pack-MULEs) carrying fragments of cellular genes (Jiang et al. 2004).
O. glaberrima gene name . | Best BLASTN homology . | Best BLASTN EST homology . | Best BLASTX homology (swiss prot) . | Protein domain . | Putative function . | Putative O. sativa (Nipponbare) orthologous gene . | % of nucleotide identity . | % of protein identity . | % of similarity . | Length (bp) . | Ka (dN) . | Ks (dS) . | Ka/Ks . |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
OG-BBa0049I08.5 | AK122162 O. sativa flcDNA (2e-148) | EL586675 O. sativa (2e-162) | Q02921 Glycine max Early nodulin 93 (8e-20) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04940 | 65 | 54.1 | 59.5 | — | — | — | — |
OG-BBa0049I08.6 | AK121791 O. sativa flcDNA (1e-157) | CI240527 O. sativa (1e-158) | Q02921 G. max Early nodulin 93 (4e-14) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04950 | 89.1 | 76.8 | 78.4 | 376 | — | — | — |
OG-BBa0049I08.7 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.75 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.8 | NM_001063299 O. sativa Os06g0142000 (0.0) | CB622551 O. sativa (0.0) | Q9SH88 A. thaliana Ribosome biogenesis regulatory protein (3e-48) | — | Putative protein | LOC_Os06g04970 | 98.6 | 97.7 | 98.4 | 2073 | 0.0112 | 0.0237 | 0.4728 |
OG-BBa0049I08.85a | AP002838 O. sativa BAC (2e-156) | CT858525 O. sativa (8e-67) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.9a | AP004090 O. sativa BAC (8e-69) | CT844843 O. sativa (3e-68) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.10a | AC121363 O. sativa BAC (3e-128) | CB649669 O. sativa (1e-57) | P47927 A. thaliana Floral homeotic protein APETALA 2 (4e-8) | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.11 | AK073027 O. sativa flcDNA (0.0) | CB658549 O. sativa (0.0) | Q9FJT2 A. thaliana F-box/FBD/LRR-repeat (5e-4) | — | Putative F-box protein | LOC_Os06g04980 | 98.1 | 96.5 | 97.7 | 1542 | 0.0173 | 0.0198 | 0.8715 |
OG-BBa0049I08.12 | AK122162 O. sativa flcDNA (2e-175) | CI252819 O. sativa (2e-175) | Q02921 G. max Early nodulin 93 (1e-17) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04990 | 99.7 | 100 | 100 | 348 | 0.0000 | 0.0142 | 0.001 |
O. glaberrima gene name . | Best BLASTN homology . | Best BLASTN EST homology . | Best BLASTX homology (swiss prot) . | Protein domain . | Putative function . | Putative O. sativa (Nipponbare) orthologous gene . | % of nucleotide identity . | % of protein identity . | % of similarity . | Length (bp) . | Ka (dN) . | Ks (dS) . | Ka/Ks . |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
OG-BBa0049I08.5 | AK122162 O. sativa flcDNA (2e-148) | EL586675 O. sativa (2e-162) | Q02921 Glycine max Early nodulin 93 (8e-20) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04940 | 65 | 54.1 | 59.5 | — | — | — | — |
OG-BBa0049I08.6 | AK121791 O. sativa flcDNA (1e-157) | CI240527 O. sativa (1e-158) | Q02921 G. max Early nodulin 93 (4e-14) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04950 | 89.1 | 76.8 | 78.4 | 376 | — | — | — |
OG-BBa0049I08.7 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.75 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.8 | NM_001063299 O. sativa Os06g0142000 (0.0) | CB622551 O. sativa (0.0) | Q9SH88 A. thaliana Ribosome biogenesis regulatory protein (3e-48) | — | Putative protein | LOC_Os06g04970 | 98.6 | 97.7 | 98.4 | 2073 | 0.0112 | 0.0237 | 0.4728 |
OG-BBa0049I08.85a | AP002838 O. sativa BAC (2e-156) | CT858525 O. sativa (8e-67) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.9a | AP004090 O. sativa BAC (8e-69) | CT844843 O. sativa (3e-68) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.10a | AC121363 O. sativa BAC (3e-128) | CB649669 O. sativa (1e-57) | P47927 A. thaliana Floral homeotic protein APETALA 2 (4e-8) | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.11 | AK073027 O. sativa flcDNA (0.0) | CB658549 O. sativa (0.0) | Q9FJT2 A. thaliana F-box/FBD/LRR-repeat (5e-4) | — | Putative F-box protein | LOC_Os06g04980 | 98.1 | 96.5 | 97.7 | 1542 | 0.0173 | 0.0198 | 0.8715 |
OG-BBa0049I08.12 | AK122162 O. sativa flcDNA (2e-175) | CI252819 O. sativa (2e-175) | Q02921 G. max Early nodulin 93 (1e-17) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04990 | 99.7 | 100 | 100 | 348 | 0.0000 | 0.0142 | 0.001 |
The e-value for the BLAST is presented in brackets. Predicted genes located in the S1 locus, between markers RM19357 and RMC6_22028.
Pseudogene.
O. glaberrima gene name . | Best BLASTN homology . | Best BLASTN EST homology . | Best BLASTX homology (swiss prot) . | Protein domain . | Putative function . | Putative O. sativa (Nipponbare) orthologous gene . | % of nucleotide identity . | % of protein identity . | % of similarity . | Length (bp) . | Ka (dN) . | Ks (dS) . | Ka/Ks . |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
OG-BBa0049I08.5 | AK122162 O. sativa flcDNA (2e-148) | EL586675 O. sativa (2e-162) | Q02921 Glycine max Early nodulin 93 (8e-20) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04940 | 65 | 54.1 | 59.5 | — | — | — | — |
OG-BBa0049I08.6 | AK121791 O. sativa flcDNA (1e-157) | CI240527 O. sativa (1e-158) | Q02921 G. max Early nodulin 93 (4e-14) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04950 | 89.1 | 76.8 | 78.4 | 376 | — | — | — |
OG-BBa0049I08.7 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.75 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.8 | NM_001063299 O. sativa Os06g0142000 (0.0) | CB622551 O. sativa (0.0) | Q9SH88 A. thaliana Ribosome biogenesis regulatory protein (3e-48) | — | Putative protein | LOC_Os06g04970 | 98.6 | 97.7 | 98.4 | 2073 | 0.0112 | 0.0237 | 0.4728 |
OG-BBa0049I08.85a | AP002838 O. sativa BAC (2e-156) | CT858525 O. sativa (8e-67) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.9a | AP004090 O. sativa BAC (8e-69) | CT844843 O. sativa (3e-68) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.10a | AC121363 O. sativa BAC (3e-128) | CB649669 O. sativa (1e-57) | P47927 A. thaliana Floral homeotic protein APETALA 2 (4e-8) | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.11 | AK073027 O. sativa flcDNA (0.0) | CB658549 O. sativa (0.0) | Q9FJT2 A. thaliana F-box/FBD/LRR-repeat (5e-4) | — | Putative F-box protein | LOC_Os06g04980 | 98.1 | 96.5 | 97.7 | 1542 | 0.0173 | 0.0198 | 0.8715 |
OG-BBa0049I08.12 | AK122162 O. sativa flcDNA (2e-175) | CI252819 O. sativa (2e-175) | Q02921 G. max Early nodulin 93 (1e-17) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04990 | 99.7 | 100 | 100 | 348 | 0.0000 | 0.0142 | 0.001 |
O. glaberrima gene name . | Best BLASTN homology . | Best BLASTN EST homology . | Best BLASTX homology (swiss prot) . | Protein domain . | Putative function . | Putative O. sativa (Nipponbare) orthologous gene . | % of nucleotide identity . | % of protein identity . | % of similarity . | Length (bp) . | Ka (dN) . | Ks (dS) . | Ka/Ks . |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
OG-BBa0049I08.5 | AK122162 O. sativa flcDNA (2e-148) | EL586675 O. sativa (2e-162) | Q02921 Glycine max Early nodulin 93 (8e-20) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04940 | 65 | 54.1 | 59.5 | — | — | — | — |
OG-BBa0049I08.6 | AK121791 O. sativa flcDNA (1e-157) | CI240527 O. sativa (1e-158) | Q02921 G. max Early nodulin 93 (4e-14) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04950 | 89.1 | 76.8 | 78.4 | 376 | — | — | — |
OG-BBa0049I08.7 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.75 | — | — | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.8 | NM_001063299 O. sativa Os06g0142000 (0.0) | CB622551 O. sativa (0.0) | Q9SH88 A. thaliana Ribosome biogenesis regulatory protein (3e-48) | — | Putative protein | LOC_Os06g04970 | 98.6 | 97.7 | 98.4 | 2073 | 0.0112 | 0.0237 | 0.4728 |
OG-BBa0049I08.85a | AP002838 O. sativa BAC (2e-156) | CT858525 O. sativa (8e-67) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.9a | AP004090 O. sativa BAC (8e-69) | CT844843 O. sativa (3e-68) | — | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.10a | AC121363 O. sativa BAC (3e-128) | CB649669 O. sativa (1e-57) | P47927 A. thaliana Floral homeotic protein APETALA 2 (4e-8) | — | Hypothetical protein | No orthologous gene | — | — | — | — | — | — | — |
OG-BBa0049I08.11 | AK073027 O. sativa flcDNA (0.0) | CB658549 O. sativa (0.0) | Q9FJT2 A. thaliana F-box/FBD/LRR-repeat (5e-4) | — | Putative F-box protein | LOC_Os06g04980 | 98.1 | 96.5 | 97.7 | 1542 | 0.0173 | 0.0198 | 0.8715 |
OG-BBa0049I08.12 | AK122162 O. sativa flcDNA (2e-175) | CI252819 O. sativa (2e-175) | Q02921 G. max Early nodulin 93 (1e-17) | pfam03386, Early nodulin 93 ENOD93 protein | Putative ENOD93 protein | LOC_Os06g04990 | 99.7 | 100 | 100 | 348 | 0.0000 | 0.0142 | 0.001 |
The e-value for the BLAST is presented in brackets. Predicted genes located in the S1 locus, between markers RM19357 and RMC6_22028.
Pseudogene.
The gene content in the segment was found to be variable between the two species. Neither the pseudogenes nor the hypothetical proteins were found at the collinear positions in O. sativa (Table 2 and Figure 4). In contrast, hypothetical gene Os06g04960 from O. sativa was not found to be collinear with O. glaberrima. Re-annotation of the O. sativa segment showed that this gene overlapped different predicted TEs and that it was split away in O. glaberrima by the insertion of additional TEs and the predicted genes 49I08.7 and 49I08.75. All five collinear protein-coding genes showed a high conservation in gene structure, as introns–exons were strictly conserved. The only exception occurred in the first intron of gene 49I08.11, with the insertion of a 448-bp nonautononous transposon. All of the orthologous genes within the segment were found to be highly conserved at the nucleotide and protein levels, with sequence identities ranging from 96 to 100% and from 82.8 to 100%, respectively.
The Ka/Ks rates were calculated for 17 orthologous predicted genes present in the CG14 BAC and Nipponbare sequences (Table 2 and Table S4); the other genes were not analyzed due to the absence of one of the orthologous genes, due to deep gene structure differences, or because one of the two was annotated as a pseudogene. Most of the analyzed genes appeared to be under purifying selection, with the exception of four genes outside the S1 interval (49I08.1, 49I08.19, 49I08.21, and 49I08.22) whose high Ka/Ks values seem to indicate a strong positive selection or a relaxed selective constraint. Despite an overall purifying evolution of genes in the sequenced BAC, four genes show high nonsynonymous changes. Two of them are located within the S1 interval (49I08.8 and 49I08.11).
DISCUSSION
TRD and its association with interspecific sterility loci:
Different processes such as the nonrandom segregation of chromosomes during meiosis (meiotic drive) (Fishman and Willis 2005), abortion of haploid gametes (as in hybrid sterility) (Orr and Irving 2005), or abortion of zygotes after fertilization (as in hybrid inviability) (Matsubara et al. 2003) could deviate the final allelic frequency in a population from the expected Mendelian ratio. The strong association between markers showing TRD and pollen sterility QTL has been already demonstrated in an interspecific cross of tomato (Moyle and Graham 2006). This association suggests that hybrid sterility may arise from incompatibilities between different factors that play a role in gamete development, resulting in the elimination of specific allelic combinations.
In our genetic map, the interpolated positioning of genes and QTL causing O. sativa × O. glaberrima hybrid sterility evidences their colocalization with markers showing fTRD (Figure 1). Differences in the intensity and the direction of the fTRD were found for chromosomes 1, 2, 3, 4, 7, and 12, in comparison with the genetic map obtained by J. Li et al. (2008), and for chromosomes 4, 5, and 11, in comparison with the genetic map obtained by Doi et al. (1998b). Some of those differences could be explained by a rapid evolution of the sterility barriers, generating different alleles of the incompatibility factors within species (Sweigart et al. 2007; Widmer et al. 2009). Colocalization of fTRD sites with pollen sterility loci could imply that these loci also have a role in female gamete development or in postpollination, since fTRD cannot be caused by loci affecting only male gametogenesis. Occurrence of fTRD in genomic regions where no sterility loci have been described (chromosomes 2 and 6) could be explained either by the existence of a new sterility locus segregating in this cross or by some mechanism other than gamete elimination biasing the segregation. The TRD in the long arm of chromosome 6, which has also been described for another O. sativa × O. glaberrima cross (J. Li et al. 2008), involves different markers in the two O. sativa subspecies (Figure 2), indicating that different factors may cause it. This observation would suggest that the regions near and below the chromosome 6 centromere might contain polymorphic factors within the species, possibly involved in the viability of gametes, zygotes, or endosperm in the O. sativa × O. glaberrima hybrids. Interestingly, the same genomic regions have been reported to bear loci for endosperm abortion, in hybrids between O. rufipogon and O. sativa (Matsubara et al. 2003; Koide et al. 2008b), and of female and male gametes (Koide et al. 2008a). It is plausible that a related mechanism, or different alleles of the same loci, work in hybrids of the two cultivated rice species, as factors involved in reproductive barriers between different species can be closely linked or can be allelic (Hu et al. 2006).
Use of the fTRD as a method to map sterility loci:
Hybrid sterility is a common phenomenon in rice. However, only a few factors responsible for inter- or intraspecific sterility have been fine mapped (W. Li et al. 2006, 2008; D. Li et al. 2007; Zhao et al. 2007; Koide et al. 2008c), and only two have been recently identified (Chen et al. 2008; Long et al. 2008). Indeed, the mapping strategy that has been used requires isolating the genetic region into near isogenic lines and carrying out the production and phenotypic analysis of many recombinant plants.
To reduce the time and effort to map sterility loci, we developed a simpler approach that makes it possible to quickly fine map these genes on the basis of the effect that they have on their own segregation ratio (Phadnis and Orr 2009). The approach is based on the direct measurement of the TRD for each marker evaluated on a sequentially saturated map. Due to the hitchhiking effect with the distorter locus (Chevin and Hospital 2006), the variations on marker TRD are inversely proportional to the genetic distance between them and the factor causing the distortion. This leads to a maximal TRD in the genomic region containing the TRDL, which, in the case of S1, is the factor causing the allelic elimination of gametes.
Different algorithms have been previously developed to predict the genetic positions of TRDL (Lorieux et al. 1995a,b; Vogl and Xu 2000; Zhu et al. 2007) and their utilization allowed to find the approximate location of several TRDL in a whole genome scale, showing a good colocalization of distorted markers with the predicted locations for the TRDL (Bouck et al. 2005; Hall and Willis 2005). However, they sometimes fail to detect the presence of TRDL in highly distorted locations (Fishman et al. 2001). Our method differs from these previous methodologies in that the use of sequential saturations achieving a locally highly saturated map (in our case close to one marker per gene) makes the estimation of the TRDL position unnecessary, since the highest TRD is a direct evidence of the cosegregation with the distorter.
Additionally, this methodology has several advantages in comparison to the traditional mapping strategies of sterility loci, which typically involve large, advanced backcross populations evaluated for pollen and spikelet fertility. First, it is faster, as it is based on the analysis of BC1F1 or F2 populations. Second, it is accurate as the TRD could be caused by the gamete eliminator factor itself (Phadnis and Orr 2009). Third, the approach could be used to fine map not only hybrid sterility loci causing TRD, but also any TRDL located in a genomic region with a good recombination ratio, as long as enough markers can be designed in the species under study. However, its application is limited to cases where highly significant differences in the level of TRD are observed.
The complex nature of the S1 locus:
The sterility barrier due to the S1 locus has been described to involve two separate components influencing male and female gametogenesis and unlinked modifiers that enhance fTRD on O. sativa ssp. indica (Koide et al. 2008c). Using our mapping methodology in four BC1F1 populations, a region of 27.8 kb in O. sativa cv. Nipponbare, and another one of 50.3 kb in O. glaberrima cv. CG14 were identified as the location of the female component of the S1 locus. Surprisingly, in indica and japonica this component maps exactly within the interval where the male component was mapped (Koide et al. 2008c). This colocalization would indicate that a unique or two contiguous factors could be involved in the elimination of male and female S1s gametes in both O. sativa backgrounds. In the case of a single factor affecting both male and female gametogenesis, its expression or function may be different, or it may interact with additional gene products in female gametogenesis, to cause total elimination of S1s from pollen, and partial allelic elimination in megagametophytes.
On the basis of the genotype of the surviving S1s female gametes, the presence of two linked elements affecting their viability (S1A and S1B) was detected, and evidence for an additional factor (S1C) was also found in one cross. Because S1 alone is capable of inducing mTRD but not fTRD (Koide et al. 2008c), it is therefore possible that an epistatic interaction between the conspecific alleles of S1, S1A, and S1B might be the main cause of fTRD and female sterility in O. sativa × O. glaberrima hybrids. In this scenario, the additional S1C factor would have a supplementary deleterious effect over the S1s gametes, strengthening the interspecific sterility barrier in certain crosses by reducing the viability of recombinant gametes (Orr and Turelli 2001). The higher S1s elimination found in crosses involving TOG5681 with indica and japonica accessions (Table 1) may suggest that the elevated fTRD value does not depend exclusively on the indica or japonica background as previously suggested, but also on the presence of the S1C incompatibility found between certain O. glaberrima and O. sativa accessions. Nevertheless, this concept needs confirmation because of the low number of plants issued from these additional crosses.
The hypothesis that a conspecific epistatic interaction between linked factors is necessary and responsible for the hybrids' female sterility and fTRD explains quite well previous observations describing how the loss of heterozygosity by recombination on the upstream, downstream, or both O. glaberrima segments near S1 cause the recovery of female fertility and the elimination of fTRD on S1s/S1g plants (Sano 1990; Koide et al. 2008c). Furthermore, it has already been shown that a similar, complex conspecific epistasis causes sterility and TRD in Drosophila pseudoobscura intersubspecific hybrids, where sterility was observed only in males possessing an entire X-chromosome block derived from one of the parental subspecies (bogotana), and not in the ones carrying independently each one of the involved loci (Orr and Irving 2005).
On the basis of our data, the location for S1A and S1B can be predicted between markers RM19350 and S1 (210 kb segment in Nipponbare) and between S1 and RM3805 (654 kb), respectively. As for S1C, it might be located within the 9.2-Mb segment between RM204 and RM3183. Interestingly, the intersubspecific sterility locus S5 is located within this large region (Chen et al. 2008).
Predicted model for the reproductive sterility barrier between O. sativa and O. glaberrima mediated by S1:
The origin of postzygotic barriers has been explained by the Bateson–Dobzhansky–Muller (BDM) model as the accumulation of incompatibilities in epistatic interactions between genes (Bateson 1909; Dobzhansky 1936; Muller 1942). As previously mentioned, a conspecific epistasis between S1A, S1, and S1B may be the cause of the female hybrid sterility between the two cultivated rice species. It is then plausible that BDM incompatibilities between S1Ag and S1Bg with S1s may play a part in the allelic elimination, as the F1 plants produce viable S1As–S1g or S1g–S1Bs, but no S1Ag–S1s or S1s–S1Bg gametes.
In light of the available data, we considered several possible models where BDM incompatibilities during megasporogenesis and megagametogenesis are the cause of the female sterility barrier between O. sativa and O. glaberrima (see File S2, Genetic models.xlsx). The different models are all based on the following data and information: (1) the localization of S1 found with the fTRD analysis, (2) the systematic transmission of an O. sativa gene block around S1s that suggests the presence of at least one additional locus on each side of S1, (3) the maximal recombination rates within and around the S1A–S1B interval [S1A to S1 (r1), S1 to S1B (r2), and S1B to the centromere (r5)], (4) the loss of fTRD and of female sterility after recombination around S1 (Sano 1990; Koide et al. 2008c), and finally, (5) the cytological observations during female gametogenesis of the hybrids, showing two types of aborted female gametes: the ones characterized by the absence of embryo sacs, probably due to an abortion during or just after meiosis, and the ones where embryo sacs have less than seven cells (Bouharmont et al. 1985; Koide et al. 2008c).
After consideration of the different models, we retained model 1 (see explanation of the different models in File S2, Genetic models.xlsx), as it is the only one that predicts well the three following parameters: (1) The S1s observed frequency [f (S1s)], (2) the total rate of gamete survival (D) observed as the final seed set of the hybrids, and (3) the observed recombination rates between S1A–S1 (r1) and S1–S1B (r2). The proposed model states that in the presence of S1Ag–S1g–S1Bg only sister cells (1n-2c or 1n-1c) that carry at least one copy of S1g would be viable after each meiotic division (Figure 5). When recombination exchanges S1Ag or S1Bg, the epistatic interaction acting against S1s would cease, allowing the development of the corresponding S1s daughter cells. In consequence, the only S1s meiocytes that could survive are the ones whose sister cell bears the S1As–S1g–S1Bg, S1Ag–S1g–S1Bs or S1As–S1g–S1Bs recombinant haplotypes. Nevertheless, at the end of megagametogenesis, only S1s megaspores carrying the S1As–S1s–S1Bs configuration would lead to the formation of a functional embryo sac.
Although the predictions made by Model 1 are the closest to the observed parameters, the predicted values of f (S1s) are slightly lower than the ones observed for the three japonica × glaberrima crosses. The estimated f (S1s) (around 3%) lies at the inferior bound of the observed values. The S1As–S1s–S1Bs gametes seem to show some additional viability, meaning that the effect of the S1 region could show some level of incomplete penetrance. This enhanced viability rate is still difficult to explain considering the available data and could be due to: (1) unknown nonlinked factors (that we were not able to detect), which rescue some of the S1s gametes, and are segregating simultaneously with the three epistatic loci, or (2) intraspecific polymorphism on the epistatic factors, as seen in other postzygotic barriers (Sweigart et al. 2007), likely to cause differences in the levels of the BDM incompatibility. The existence of a sterility locus (S10) observed in crosses between indica and japonica at the same location of S1 (Sano et al. 1994; Zhu et al. 2005) favors the hypothesis of an allelic variation. The possible differences that might exist between the indica and japonica alleles for the S1A, S1 and S1B loci cause a certain level of sterility between the two O. sativa subspecies and may alter the allelic elimination of female gametes produced by the O. sativa × O. glaberrima hybrids.
In summary, the predictions made by our model support the hypothesis that female sterility in the O. sativa × O. glaberrima hybrids is caused by the allelic elimination of gametes, due to a BDM incompatibility in an epistatic interaction between S1 and the additional factors. Differences in the observed fTRD between indica and japonica could be due to polymorphisms within the two subspecies in S1A, S1, S1B, or S1C, or to the existence of other unlinked factors that rescue some of the gametes. Accordingly, the final allelic frequencies for a given cross would be determined by the alleles confronted in a given cross, by the recombination ratio between S1A and S1B, and by the presence of the S1C-mediated incompatibility.
Genomic variations in the S1 orthologous region and candidate genes:
Divergent evolution between closely related species may be the cause of reproductive isolation. In Drosophila, a high level of gene divergence and strong positive selection appeared to be mechanisms influencing the evolution of speciation genes (Ting et al. 1998; Presgraves et al. 2003; Barbash et al. 2004; Brideau et al. 2006). Recent studies conducted with different species of the Oryza genus indicated an extensive conservation of the microcollinearity, with local small rearrangements, or structural variations due to insertions of transposable elements (Ammiraju et al. 2008; Lu et al. 2009). In contrast, the accumulation of structural genomic variations observed within the interval of the S1 locus in O. glaberrima might imply a fast evolution of the segment, despite the recent divergence of the Asian and African rice species (approximately 0.7 MYA; Ma and Bennetzen 2004; Ammiraju et al. 2008).
We used orthologous comparative genomic approaches between O. glaberrima and O. sativa at the S1 interval to explore structural changes and nucleotide divergence of orthologous coding regions as putative candidates for the locus. Ten genes are present in the 50.3-kb S1 region of O. glaberrima and thus constitute the candidates for the S1 locus. From these 10, 2 predicted genes attract our attention on the basis of their putative functions, differential presence on the species, or high sequence divergence. The first one, carried by a pack-MULE and present only in the O. glaberrima S1 region, is gene 49I08.10. This gene shows similarities with an APETALA2 (AP2) transcription factor. Mutations in AP2 and other related proteins in Arabidopsis cause the arrest of mega-sporogenesis after the first meiotic division (Byzova et al. 1999). Pack-MULEs are known to have an important role in rice genome evolution, as they capture and relocate gene fragments in other genomic contexts (Jiang et al. 2004). A significant number of these elements are transcribed, being frequently associated with small RNAs that may modulate both pack-MULEs and paralogous gene expression (Hanada et al. 2009). If the complete paralogous AP2 gene has a function similar to that of the Arabidopsis genes, this pack-MULE could affect its expression on the heterozygotes, in a dose-dependent mechanism similar to hybrid dysgenesis (Michalak 2009), causing the abortion of female gametes.
The other interesting candidate is the one with the highest Ka and Ka/Ks values in the S1 interval (Table 2), gene 49I08.11, belonging to the super-family of F-box proteins. Members of this family constitute protein complexes known as SCF (Skp1–Cullin–F-box) involved in the control of a wide range of processes (Xu et al. 2009). Several F-box genes and their associated proteins have been related with the progression of the cell cycle, especially during sporogenesis and gametogenesis (Wang and Yang 2006; Pesin and Orr-Weaver 2008; Gusti et al. 2009). Intraspecific sequence divergence of members of such protein complexes can cause hybrid sterility, as demonstrated for the complex composed by an F-box and a SUMO E3 ligase-like protein, that controls the intersubspecific hybrid sterility mediated by the Sa locus in O. sativa (Long et al. 2008). Additionally, homologous F-box genes closely related to our putative candidate are involved in the self-incompatibility system of Hordeum bulbosum (Kakeda 2009), proving that such genes can actually act as reproductive barriers in cereals. From the two putative candidate genes described, this F-box protein is the one that fits best with the genetic model of a BDM incompatibility in an epistatic interaction developed in this article to explain the female gamete elimination caused by S1. Taking into account their ability to tightly interact with other proteins, the rising of incompatibilities after divergences in the coding sequence of the interacting proteins seems to be a feasible mechanism for this postzygotic reproductive barrier. Moreover, considering that the strong hybrid sterility between O. sativa and O. glaberrima occurs independently of the direction of the interspecific cross (Sano et al. 1979; Bougerol and Pham 1989; Sano 1990), while hybrid dysgenesis, the possible mechanism explaining the role of gene 49I08.10 in the hybrid dysfunction, is a unidirectional phenomenon, the F-box protein appears as a more consistent candidate.
Perspectives:
In this article, we have described and modeled the complex nature of the allelic female gamete elimination mediated by the S1 locus. The knowledge generated by our results will contribute to a better understanding of plant interspecific sterility barriers and will also open the door to an efficient use of O. glaberrima in rice breeding programs, as is intended with the iBridges project. Hybrid plants inheriting the O. sativa chromosomal segment that contains the S1, S1A, and S1B factors involved in the allelic elimination can be used to develop lines with all the potential genes of interest from O. glaberrima that would produce a more fertile progeny when crossed with O. sativa. More information is needed to characterize all the factors involved, as well as the molecular mechanism behind the allelic gamete elimination in hybrids from Asian and African rice species.
Footnotes
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.110.116772/DC1.
Sequence data from this article have been deposited with the EMBL Library under accession number FP340543.
Footnotes
Communicating editor: M. Kirst
Acknowledgements
Our thanks go to R. Wing (AGI, Arizona) for providing the O. glaberrima BAC clones and J. L. Goicoechea for the in silico confirmation of the MTP; M. Morales for the in vitro germination of seeds and tissue harvest; and M. F. Alvarez and L. Melgarejo for their contribution in the construction of the IR64xTOG5681 genetic map. We also thank two anonymous reviewers for their constructive comments that helped us to improve significantly the manuscript.This research was supported by the Generation Challenge Program (Grant G4007.01). A grant from Département Soutien et Formation (DSF) Institut de Recherche pour le Développement (IRD) and International Center for Tropical Agriculture (CIAT)provided the Ph.D. scholarship for A. Garavito.
References
Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang et al.,
Aluko, G., C. Martinez, J. Tohme, C. Castano, C. Bergman et al.,
Ammiraju, J. S. S., F. Lu, A. Sanyal, Y. Yu, X. Song et al.,
Barbash, D. A., P. Awadalla and A. M. Tarone,
Bateson, W.,
Bikard, D., D. Patel, C. Le Mette, V. Giorgi, C. Camilleri et al.,
Bomblies, K., J. Lempe, P. Epple, N. Warthmann, C. Lanz et al.,
Bouck, A., R. Peeler, M. L. Arnold and S. R. Wessler,
Bougerol, B., and J. L. Pham,
Bouharmont, J., M. Olivier and M. D. Chassart,
Brideau, N. J., H. A. Flores, J. Wang, S. Maheshwari, X. Wang et al.,
Byzova, M. V., J. Franken, M. G. M. Aarts, J. de Almeida-Engler, G. Engler et al.,
Chen, J., J. Ding, Y. Ouyang, H. Du, J. Yang et al.,
Chevin, L. M., and F. Hospital,
Dingkuhn, M., M. P. Jones, D. E. Johnson and A. Sow,
Dobzhansky, T.,
Doi, K., K. Taguchi and A. Yoshimura,
Doi, K., A. Yoshimura and N. Iwata,
Doi, K., K. Taguchi and A. Yoshimura,
Droc, G., C. Perin, S. Fromentin and P. Larmande,
Fishman, L., and J. H. Willis,
Fishman, L., A. J. Kelly, E. Morgan and J. H. Willis,
Ghesquière, A., J. Sequier, G. Second and M. Lorieux,
Gusti, A., N. Baumberger, M. Nowack, S. Pusch, H. Eisler et al.,
Hall, M. C., and J. H. Willis,
Hanada, K., V. Vallejo, K. Nobuta, R. K. Slotkin, D. Lisch et al.,
Heuer, S., and K. M. Miezan,
Hu, F. Y., P. Xu, X. N. Deng, J. W. Zhou, J. Li et al.,
International Rice Genome Sequencing Project,
Jiang, N., Z. Bao, X. Zhang, S. R. Eddy and S. R. Wessler,
Josefsson, C., B. Dilkes and L. Comai,
Jurka, J., V. V. Kapitonov, A. Pavlicek, P. Klonowski, O. Kohany et al.,
Kakeda, K.,
Kikuchi, S., K. Satoh, T. Nagata, N. Kawagashira, K. Doi et al.,
Kim, H., B. Hurwitz, Y. Yu, K. Collura, N. Gill et al.,
Koide, Y., M. Ikenaga, N. Sawamura, D. Nishimoto, K. Matsubara et al.,
Koide, Y., M. Ikenaga, Y. Shinya, K. Matsubara and Y. Sano,
Koide, Y., K. Onishi, D. Nishimoto, A. R. Baruah, A. Kanazawa et al.,
Kosambi, D.,
Lee, H.-Y., J.-Y. Chou, L. Cheong, N.-H. Chang, S.-Y. Yang et al.,
Li, D., L. Chen, L. Jiang, S. Zhu, Z. Zhao et al.,
Li, J., P. Xu, X. Deng, J. Zhou, F. Hu et al.,
Li, W., R. Zeng, Z. Zhang, X. Ding and G. Zhang,
Li, W., R. Zeng, Z. Zhang, X. Ding and G. Zhang,
Liang, C., P. Jaiswal, C. Hebbard, S. Avraham, E. S. Buckler et al.,
Long, Y., L. Zhao, B. Niu, J. Su, H. Wu et al.,
Lorieux, M., B. Goffinet, X. Perrier, D. G. León and C. Lanaud,
Lorieux, M., X. Perrier, B. Goffinet, C. Lanaud and D. G. León,
Lorieux, M., M. N. Ndjiondjop and A. Ghesquiere,
Lorieux, M., G. Reversat, S. X. Garcia Diaz, C. Denance, N. Jouvenet et al.,
Lu, F., J. S. S. Ammiraju, A. Sanyal, S. Zhang, R. Song et al.,
Ma, J., and J. L. Bennetzen,
Masly, J. P., C. D. Jones, M. A. F. Noor, J. Locke and H. A. Orr,
Matsubara, K., T. Khin and Y. Sano,
Michalak, P.,
Mihola, O., Z. Trachtulec, C. Vlcek, J. C. Schimenti and J. Forejt,
Moyle, L. C., and E. B. Graham,
Ndjiondjop, M. N., L. Albar, D. Fargette, C. Fauquet and A. Ghesquiere,
Ndjiondjop, M. N., L. Albar, D. Fargette, C. Brugidou, M. P. Jones et al.,
Orjuela, J., A. Garavito, M. Bouniol, J. Arbelaez, L. Moreno et al.,
Orr, H. A., and S. Irving,
Orr, H. A., and M. Turelli,
Ouyang, S., and C. R. Buell,
Ouyang, S., W. Zhu, J. Hamilton, H. Lin, M. Campbell et al.,
Pesin, J. A., and T. L. Orr-Weaver,
Phadnis, N., and H. A. Orr,
Presgraves, D. C., L. Balagopalan, S. M. Abmayr and H. A. Orr,
Ren, G., P. Xu, X. Deng, J. Zhou, F. Hu et al.,
Rice, P., I. Longden and A. Bleasby,
Rutherford, K., J. Parkhill, J. Crook, T. Horsnell, P. Rice et al.,
Salamov, A. A., and V. V. Solovyev,
Sano, Y.,
Sano, Y.,
Sano, Y., Y.-E. Chu and H.-I. Oka,
Sano, Y., R. Sano, M. Eiguchi and H. Y. Hirano,
Sarla, N., and B. P. M. Swamy,
Sonnhammer, E. L. L., and R. Durbin,
Soriano, I. R., V. Schmit, D. S. Brar, J.-C. Prot and G. Reversat,
Suyama, M., D. Torrents and P. Bork,
Sweigart, A. L., A. R. Mason and J. H. Willis,
Taguchi, K., K. Doi and A. Yoshimura,
Tang, S., and D. C. Presgraves,
Thiel, T., W. Michalek, R. Varshney and A. Graner,
Ting, C.-T., S.-C. Tsaur, M.-L. Wu and C.-I. Wu,
Toshio, M., and S. Folke,
Untergasser, A., H. Nijveen, X. Rao, T. Bisseling, R. Geurts et al.,
Vogl, C., and S. Xu,
Wang, Y., and M. Yang,
Widmer, A., C. Lexer and S. Cozzolino,
Wittbrodt, J., D. Adam, B. Malitschek, W. Maueler, F. Raulf et al.,
Xu, G., H. Ma, M. Nei and H. Kong,
Yang, Z.,
Zhang, Z., P. Xu, F. Hu, J. Zhou, J. Li et al.,
Zhao, Z. G., L. Jiang, W. W. Zhang, C. Y. Yu, S. S. Zhu et al.,
Zhu, C., C. Wang and Y. M. Zhang,
Zhu, S., L. Jiang, C. Wang, H. Zhai, D. Li et al.,