Index INTRODUCTION MATERIALS AND METHODS Plants. Allelic identification. DNA amplification and sequencing. Phylogenetic analysis. Polymorphism near the OsGA20ox2 gene. Hybridization experiments. RESULTS AND DISCUSSION References

INTRODUCTION

The process of artificial selection is crucial in crops for understanding the importance of genetic diversity that is a major source of varietal improvement (Harlan 1992; Evans 1993; Tanksley and McCouch 1997). Beneficial phenotypes are fixed by artificial selection, in which some rare or hidden agronomic traits within a population become advantageous in response to sudden changes in the natural environment or human society (Harlan 1992; Smith 2001). The interplay of shared history and microevolutionary forces is not usually discriminated in crops; however, one of the well known events in agriculture is the “Green Revolution” in wheat and rice (Peng et al. 1999; Hedden 2003). High-yielding varieties (HYVs) were developed in wheat and rice by altering the architecture of plants to produce a short-stature that results in lodging resistance and a high seed production under heavy application of nitrogen fertilizer (Harlan 1992; Khush 1999). Identification of the genes responsible for these traits proved that the genes regulate the gibberellin (GA) biosynthetic pathway (Peng et al. 1999; Hedden 2003).

In rice, a recessive semi-dwarf 1 (sd1 or OsGA20ox2) gene detected in Dee-geo-woo-gen (DGWG) was used for a HYV(IR8). Later, HYVs released independently in different countries were proved to carry different alleles at the same locus, suggesting its significant role for the seed production (Kikuchi and Futsuhara 1997; Monna et al. 2002; Spielmeyer 2002; Sasaki et al. 2002; Ashikari et al. 2002). Only two naturally occurring variants at this locus are known at present: one native variant is a widely used allele (loss-of-function mutation) in DGWG and the other is a leaky mutant found in Jikkoku (JKK). Our knowledge on the allelic distribution is, however, limited in rice since the expression of sd1 is modified by other genes (Aquino and Jennings 1966). Furthermore, semi-dwarfing traits due to the sd1 gene are slightly deleterious under traditional cultivation in the tropics, giving a question how such a deleterious gene has been preserved in landraces (Jennings and Aquino 1968). The recent molecular information prompted us to investigate the distribution of the semi-dwarf alleles at the OsGA20ox2 locus in wild and cultivated rice strains. We report here the genealogy of a gene which contributed to the “Green revolution” of rice, suggesting that the selected allele by farmers had been preserved in the wild ancestor until the change in the demands of agriculture.

View Details

Fig. 1. Structure of the OsGA20ox2 (or sd1) gene (a) and the detection of the DGWG and JKK alleles. The DGWG allele has a 382-bp deletion in the 1st and 2nd exons while the JKK allele has a replacement substitution of G to T in the 2nd exon. The DGWG allele is distinguished from the others by PCR amplification with the primers of L2 and R1 due to the presence of a large deletion of 382-bp (b). After the PCR amplification, the JKK allele is detectable by the digestion with BbrPI (c). IR36, W1944 and Liu-t’ou-tu have the DGWG allele, and T65 has the wild form. + and – represent the treatments with and without digestion by BbrPI, respectively.

MATERIALS AND METHODS

Plants.

A total of 256 accessions (10 species), which were chosen to include accessions from throughout the native geographic ranges of the species, were surveyed for the allelic distribution at the OsGA20ox2 locus. They included 58 accessions of Oryza rufipogon and 140 of O. sativa. O. rufipogon is the wild progenitor of Asian rice (O. sativa). O. rufipogon is divisible into perennial or annual, and O. sativa is divisible into the Indica and Japonica types (Oka 1988; Morishima et al. 1992). In the present study, the Japonica type included tropical and temperate Japonica lines as indicated in Fig. 2. In addition, 58 accessions of 8 rice species (O. glaberrima, O. barthii, O. longistaminata, O. glumaepatula, O. meridionalis, O. officinalis, O. eichingeri, and O. meyriana) were also surveyed. The materials were obtained from Natl. Inst. Genetics, Mishima, and International Rice Research Institute, Los Banos.

View Details

Fig. 2. Neighbor-joining (NJ) reconstruction of the genealogical relationships among 24 OsGA20ox2 haplotypes in 66 strains of wild and cultivated rice. ME1 and ME2 of O. meridionalis were used as outgroups. Bootstrap values for nodes supported in >50% of 1, 000 bootstrap replicates are shown. I and J show Indica and Japonica types of O. sativa, respectively. P and A show perennial and annual forms of O. rufipogon, respectively.

Allelic identification.

The two different mutant alleles are reported at the OsGA20ox2 locus on chromosome 1 of the two landraces, Dee-geo-woo-gen (DGWG) and Jikkoku (JKK). The DGWG allele was distinguished by PCR with the primers L1 (5’-CTCGTCTTCTCCCCTGTTACAA-3’) and R2 (5’-GTGTCGCCGATGTTGATGAC-3’) and the JKK allele was detected by the digestion with Bbr PI after PCR (Fig. 1).

DNA amplification and sequencing.

DNA fragments were amplified using specific primers that were designed on the basis of the reported sequence of Nipponbare (http://rgp.dna.affrc.go.jp). The 5’ flanking region to the 2nd exon (1, 874 bp) was sequenced in 63 strains including the carriers of the DGWG (IR8 and IR36) and the JKK (Shiranui) alleles. The sequence was determined in both directions using an ABI 377 automatic sequencer (Applied Biosystem, Inc., Foster City, CA) with a Big Dye Terminator Cycle Sequencing Kit (Perkin Elmer). The DNA sequences are available from DDBJ with the accession numbers AB213460 to AB213479. The reported OsGA20ox2 sequences of Nipponbare (AF254556), Milyang 23 (AF465256), IR64 (AY115558) and a strain of O. rufipogon (AY115559) were added for data analysis.

Phylogenetic analysis.

Molecular population genetic analysis was conducted by using DnaSP 3.99 (Rozas and Rozas 1999). Insertion/deletion polymorphisms (indels) were treated as a single mutation. Genetic diversity was estimated by nucleotide diversity (π) and Watterson’ estimator (θ_w) per base pair based on silent sites. For haplotype diversity, the Shannon information measure (H_hap = - Σ p_i ln p_i, where p_i is frequency of haplotypes) was used. Phylogenetic trees were constructed using PAUP*, version 4.0 (Swofford 1998). The NJ (Saitou and Nei 1987) method was conducted with the Kimura’s (1980) two-parameter distances, ignoring indels. The haplotype tree was also constructed using a maximum parsimony analysis (branch and bound search, stepwise addition) according to Caicedo et al. (2004), including indels. The tree was based on 72 polymorphic sites, and the length was 92 steps excluding a character which is highly homoplasious (consistency index < 0.333). Reproductively isolated taxa (O. glaberrima, O. glumaepatula and O. meridionalis ) were used as outgroups.

Polymorphism near the OsGA20ox2 gene.

The three markers (KS3, RM8278 and RM1387) were used to examine polymorphisms near the DGWG allele. KS3 was designed from P0005H10 (Rice Genome Research Project; http://rgp.dna.affc.go.jp/), and RM8278 and RM1387 were based on microsatellite markers from Cornell Univ. (http://www.gramine.org/). The primers designed for KS3 were 5’- TTCGGCCAGCCGCACAATATTT -3’ and 5’- CGATGGTGATAAGCAAATGCTAGCAAA -3’. The map positions of KS3, RM8278 and RM1387 were 108cM, 108cM and 121cM, respectively, while the sd1 was located 114 cM based on RILs of A58 x W107.

Hybridization experiments.

F₅ populations were made from a cross of A58 (Japonica rice) x W1944 (perennial) by the single seed descent method. The plants were grown in Sapporo (43°N) and the length of the main culm was measured after maturity to compare the phenotypic effects due to the parental alleles.

RESULTS AND DISCUSSION

A total of 256 accessions chosen from the world-wide collections were surveyed regarding the allelic distribution at the OsGA20ox2 locus. The surveyed accessions included two cultivated rice species (O. sativa and O. glaberrima from 15 countries) and eight wild rice species (from 17 countries). Polymerase chain reaction (PCR) amplification of the OsGA20ox2 gene revealed that, in addtion to three HYVs with the DGWG allele, an accession of O. sativa and two accessions of O. rufipogon carry the DGWG allele showing a large deletion, while no accession carries the JKK allele except for Shiranui. The DGWG carriers were Liu-t’ou-tu (an Indica landrace) and two O. rufipogon accessions, W1944 and W1718, all of which are from China, showing their limited distribution. The nucleotide sequence of the 1st and 2nd exons and the 5’-flanking region (alignment length of 1,824 bp) was determined and compared in 66 accessions including the DGWG carriers. Based on the nucleotide sequences, 24 haplotypes were discriminated. The 62 sequences of O. sativa and O. rufipogon included 20 distinct haplotypes. The neighbor-joining (NJ) tree divided the 20 haplotypes into two different groups (Fig. 2). Groups A₁ and A₂ included 13 and seven haplotypes, respectively, showing that the O. sativa – O. rufipogon complex contained OsGA20ox2 genes from two different lineages. All of the sequences from O. sativa were found only in the group A₁.

A total of 17 haplotypes were found in 27 accessions of O. rufipogon, however, five haplotypes (S1, S3, RS1, RS2 and RS3) were found in Indica rice (18 accessions), and only haplotype S1 was found in Japonica rice (16 accessions), except in Shiranui, reflecting a lower value of haplotype diversity in Japonica rice of O. sativa (H_hap= 0.224) than in others (Table 1). Among the 13 haplotypes detected in the group A₁, three haplotypes (S1 – S3) and seven haplotypes (R1 – R7) were found only in O. sativa and only in O. rufipogon, respectively, while three haplotypes (RS1 – RS3) were frequently shared between O. rufipogon and Indica rice (Fig. 2).

View Details

Table 1. Features of OsGA20ox2 sequence variation*

Comparisons of DNA sequence variation between closely related species might provide insight into the history of a gene during the domestication process (Eyre-Walker et al. 1998; Wang et al. 1999; Lukens and Doebley 2001). The silent site nucleotide diversities (π and θ_w) were extremely low in Japonica rice, in contrast with those in Indica rice as well as in the wild progenitor (Table 1). Based on Tajima’s D, no indication against the prediction from its neutrality was detected (Table 1). The loss of genetic diversity in Japonica rice appears to result from a genetic bottleneck effect rather than a selective sweep since the OsC1 and Wx genes on a different chromosome (Chr. 6) show a similar trend (Saitoh et al. 2004; Olsen and Purugganan 2002).

The finding that the DGWG carriers have the identical OsGA20ox2 sequence indicates their common origin, showing that the mutant allele is distributed not only in landraces but also in the wild progenitor. Based on the NJ method, genealogical relationships among the 20 haplotypes in the O. rufipogon – O. sativa complex are shown in Fig. 2. These haplotypes are organized into two groups (A₁ and A₂ haplogroups) and all of the cultivated forms are grouped into the A₁ haplogroup. The haplotype network using a maximum parsimony analysis showed that haplotypes RS2 (DGWG allele) and S2 (JKK allele) were recently derived in different lineages; the former from haplotype RS1 and the latter from haplotype S1 (Fig. 3).

View Details

Fig. 3. The OsGA20ox2 haplotype tree based on the maximum-parsimony method. The tree represents one of three most-parsimonious arrangements; relationships among two major haplogroups (A1 and A2) do not change with alternative trees. Mutational steps (substitutions and indels) are shown by small circles. The 20 haplotypes found in O. rufipogon and O. sativa are shown in circles (R) and squares (S), respectively, while the shared haplotypes are shown in hexagons (RS). The dwarf mutations are shown by DGWG and JKK. Numbers near each haplotype indicate its occurrence in the three taxa (O.rufipogon/Indica/Japonica) including 62 rice accessions.

Regarding the origin of the DGWG allele, it was driven by a new mutation that arose during the domestication process or preexisted as a rare variant in the wild progenitor. The fixation of a beneficial allele eliminates DNA variation in the surrounding region (Wang et al. 1999; Innan and Kim 2004). To investigate polymorphisms around the OsGA20ox2 gene, three loci (KS3, RM8278 and RM1387), whose positions are separated from the OsGA20ox2 gene by 6 – 7 cM on chromosome 1, were examined in the RS2 carriers (IR36, IR8, Liu-t’ou-tu, W1718 and W1944). Only W1944 was different from the other carriers in all of the three loci, suggesting that the DGWG allele has been present in the wild progenitor, rather than that the DGWG allele has been introgressed from HYVs to W1944. This assumption agrees with that O. rufipogon incliding W1944 has a quantitative trait locus for seed-shattering on a similar location of chromosome 1 to that of the OsGA20ox2 gene (Xiong et al. 1999; Cai and Morishima 2002).

High-planting-density in productive or fertilized fields increases competition, which selects against semi-dwarfed plants. A negative relationship between competitive ability and seed production is often observed as a trade-off in various plant species in natural fields (Grime 1974). In rice, this is explained as follows: an increase in plant height leads to mutual shading and a decrease in leaf area index, causing low seed production; conversely, short-statured plants minimize mutual shading, giving rise to a high-yielding capacity per unit area (Tanaka et al. 1966). This complex relationship means that the selective value of the dwarfing phenotype differs depending on the neighboring plants or habitat. It was not anticipated before the “Green Revolution” that populations composed of semi-dwarfed plants would have a high potential for grain yield in comparison with those composed of tall plants. Since farmers recognize a higher yield per unit area, a high yielding short-statured plants could be selected for in intensively managed monocultures, where rice plants are less competitive with each other, as has been proposed (Jennings 1966; Jennings and Aquino 1968).

How could such a slightly deleterious gene be preserved in the wild progenitor? Phenotypic effects of the OsGA20ox2 alleles were examined in the F₅ populations of a cross between A58 (a landrace) and W1944 (a wild accession carrying the DGWG allele). W1944 showed a perennial habit and a tall-stature (Fig. 4a). The crossing experiment revealed a continuous and transgressive segregation in culm length, implying that the DGWG allele of W1944 functions in combination with other interacting genes (Fig. 4b). The existence of interacting genes is supported by the fact that backcrosses with A58 generated only distinct dwarfing phenotypes. Therefore, the DGWG variant might have been preserved as a hidden variation in the genetic background of wild rice, without expressing a short-stature.

View Details

Fig. 4. Frequency distribution of culm length observed in the 96 F5 lines of A58 × W1944. The allelic state at the OsGA20ox2 locus was determined by the method described in Fig. 1. Bar is 20 cm.

The present results confirm that farmers have selected the DGWG allele together with its interacting genes to obtain a high yield under various practices in agriculture and that the identical variant was maintained in different landraces. This view is controversial to an assumption that some landraces obtained the agronomic gene from a new mutation during traditional practices. It is, therefore, concluded that gene diversity in the wild relatives was actually indispensable for the “Green Revolution” of rice and it will be more essential to meet future agricultural demands in our changing world (Ruttan 1999; Tilman et al. 2002).

We would like to appreciate H. Morishima for invaluable comments and the Ministry of Education, Sports, Culture, Science and Technology, Japan, for partial financial support.