| Edited by Yoshio Sano. Eiichi Ohtsubo: Corresponding author. E-mail: eohtsubo@iam.u-tokyo.ac.jp. Note: Supplementary materials in this article are at http://wwwsoc.nii.ac.jp/gsj3/sup/82(3)Xu/ |
The rice genus, Oryza, consists of 22 species, including two cultivated rice species Oryza sativa and Oryza glaberrima (Khush, 1997; Ge et al., 1999; Vaughan et al., 2003). O. sativa is now cultivated worldwide, whereas O. glaberrima is grown in a limited area in Africa. Both cultivated rice species are diploid (2n = 24) and have the AA genome. There are five wild rice species with the AA genome, Oryza rufipogon, Oryza barthii, Oryza glumaepatula, Oryza longistaminata and Oryza meridionalis. Previous studies of the cultivated and wild rice strains of the AA-genome species, on the basis of morphology (Oka and Chang, 1962; Oka, 1974), isozymes (Second, 1982), RFLPs (Ishii et al., 1988; Wang et al., 1992), AFLPs (Aggarwal et al., 1999), inter-simple sequence repeat (ISSR) polymorphisms (Joshi et al., 2000), RAPDs (Bautista et al., 2001), and microsatellite polymorphisms (Bautista et al., 2001; Ishii et al., 2001), have indicated with little controversy that the progenitors of the two cultivated rice species O. sativa and O. glaberrima are O. rufipogon and O. barthii, respectively.
O. sativa has been classified into two subspecies, japonica and indica (Kato et al., 1928). A third subspecies, called javanica, has also been reported for certain strains (Matsuo, 1952; Morinaga, 1954). Javanica strains are, however, thought to be tropical components of a single japonica group (Oka, 1958). O. rufipogon has been classified into two ecotypes; perennial and annual (Oka and Morishima, 1967; Oka, 1988; Morishima et al., 1984; 1992). An intermediate type has been noted for some O. rufipogon strains (Morishima et al., 1961; Sano et al., 1980). The particular ecotype of O. rufipogon that is the actual progenitor of O. sativa remains controversial. Based on the results of analyses of morphological and ecological characteristics, many researchers proposed that O. sativa originated monophyletically from O. rufipogon (Oka, 1964; Chang, 1976; Sano et al., 1980; Oka and Sano, 1981). Biochemical and molecular studies, however, suggest that the japonica and indica strains are much closer to different O. rufipogon strains than they are to each other (Second, 1982; Wang et al., 1992; Mochizuki et al., 1993; Hirano et al., 1994; Bautista et al., 2001; Ishii et al., 2001), suggesting that the two subspecies japonica and indica of O. sativa originated diphyletically or polyphyletically from O. rufipogon.
Short interspersed elements (SINEs) are 70–500 bp repetitive DNA sequences that have proliferated via transcription, followed by reverse transcription. SINEs are found in a wide variety of eukaryotes, including animals, fungi and plants (Umeda et al., 1991; Okada, 1991; Kachroo et al., 1995). SINEs have served as useful markers for phylogenetic studies owing to their specific characters: once inserted, SINE remains in that genomic locus; the probability of insertion occurring more than once at any single site has been presumed to be extraordinarily low (Batzer and Deininger, 1991; Batzer et al., 1994; Takahashi et al., 1998; Nikaido et al., 1999; for a review, see Shedlock and Okada, 2000). In particular, the presence or absence of a SINE inserted at a locus is easy to assay by PCR, and thus SINEs have found increasing use as phylogenetic markers to study relationships among species of primates (Bailey and Shen, 1993; 1997; Hamdi et al., 1999; Salem et al., 2003), whales and even-toed ungulates (Shimamura et al. 1997; Nikaido et al. 1999), salmonid fish (Murata et al., 1993; Hamada et al., 1998) and plants (Mochizuki et al., 1993; Tatout et al., 1999; Cheng et al., 2002).
The first plant SINE, named p-SINE1, was identified in the genomes of O. sativa and O. glaberrima (Umeda et al., 1991; Mochizuki et al., 1992). A large number of p-SINE1 members that are present at particular loci were further identified in the strains of O. sativa [Nipponbare (ssp. japonica), IR36 and C5924 (ssp. indica)], as well as in those of wild rice species with the AA genome, such as O. barthii, O. glumaepatula, O. longistaminata and O. meridionalis (Mochizuki et al., 1993; Hirano et al., 1994; Motohashi et al., 1997; Cheng et al., 2002; 2003). Some members were found to show insertion polymorphism in strains of one or more species. Such polymorphic p-SINE1 members were used for the study of the phylogenetic relationships of strains of the species with the AA genome. Recently, Cheng et al. (2003) identified 23 p-SINE1 members from O. sativa, which show insertion polymorphism in strains of O. sativa and O. rufipogon. Most of the members contain three common substitution mutations (T at nucleotide position 7, A at 63, and A at 114), and form a subfamily, named RA (recently- amplified). Phylogenetic analysis based on the insertion polymorphism of the members revealed that the O. sativa ssp. japonica and javanica strains are closely related to a group of O. rufipogon perennial strains, and that the indica strains are related to the O. rufipogon annual strains (Cheng et al., 2003). This result indicates that O. sativa has been derived polyphyletically from O. rufipogon. Note, however, that no p-SINE1 members identified from O. rufipogon were used for the phylogenetic analysis in the previous study. It was therefore difficult to determine detailed relationships between some O. rufipogon strains that are distantly related to O. sativa strains.
In this study, to obtain more detailed information about the origin of O. sativa and O. rufipogon, we identified 26 new p-SINE1 members from several O. rufipogon strains representing three ecotypes that show insertion polymorphism among O. rufipogon strains. A total of 51 p-SINE1 members, including the new members and those previously identified, were then examined for their presence or absence at the respective loci in 103 strains of O. rufipogon and O. sativa, and a phylogenetic tree was constructed on the basis of the p-SINE1 insertion polymorphism. Based on the results of phylogenetic analysis, we discuss the possible origins of the strains of O. rufipogon and O. sativa, each of which shows ecotype or subspecies differentiation.
A total of 103 rice strains (68 O. sativa and 35 O. rufipogon strains) were chosen for the phylogenetic analysis (Table 1). These include strains of two subspecies (japonica and indica) of O. sativa and strains of three ecotypes (perennial, annual and intermediate) of O. rufipogon, all of which were originally collected in Asia or New Guinea. In addition to the 103 strains, another five strains representing each of the other rice species with the AA genome were chosen for the present phylogenetic analysis (see Table 1). Total genomic DNA was isolated from some of these rice strains as described previously (Ohtsubo et al., 1991). Total genomic DNA samples of the rest of the strains have been obtained from elsewhere, as described previously (Cheng et al., 2003).
 View Details | Table 1. The presence or absence of p-SINE1 members at respective loci in various rice strains |
Adaptor-ligation based-PCR (ADL-PCR) (Cheng et al., 2003) was performed to identify new p-SINE1 members present at different loci from those previously identified, as follows. The total DNA of an O. rufipogon strain (W1681, W2007, W1943, W0120, or W0593) was digested with BamHI, EcoRI, HindIII, or XbaI (New England Biolabs), none of which cut the p-SINE1 member. T4 DNA ligase (New England Biolabs) was used to ligate the digested DNA with an oligonucleotide adaptor. First, PCR was performed with ExTaq DNA polymerase (Takara) using a ligated sample as the template and using primers that hybridize to the adaptor and to the p-SINE1 sequence in order to obtain fragments with the proximal portion of p-SINE1 and its flanking sequence. Second, PCR was performed with primers that hybridize to the adaptor and to a different portion of the p-SINE1 sequence. Fragments that included the entire p-SINE1 sequence were obtained by an ADL-PCR with primers that hybridize to the flanking sequence of each of the identified members and to the adaptor. Inverse PCR (IPCR) was also performed to identify new p-SINE1 members, as described elsewhere (Tenzen et al., 1994), with the total DNA from an O. rufipogon strain as the template. Primers used for PCR are listed in Table 2.
 View Details | Table 2. Primers used for Adaptor-ligation PCR and IPCR |
The presence or absence of each p-SINE1 member was determined by identifying one unique PCR fragment with or without a p-SINE1 member after electrophoresis in a 1.8% agarose gel, as described previously (Motohashi et al., 1997). When the fragments differed in size or when two or more bands were present, the presence or absence of p-SINE1 in the fragments was confirmed by Southern hybridization or by direct sequencing of the PCR products (Cheng et al., 2003). Failure of the amplification (no bands) was treated as no data. Primers used for PCR to identify the presence or absence of each of the representative p-SINE1 members are listed in Table 2. Primer sequences useful for the presence or the absence of each of the other p-SINE1 members will be provided upon request.
PCR products were cloned into the pGEM-T Easy Vector System I (Promega), according to the supplier’s instructions. Sequencing was performed on an ABI 377 automated DNA sequencer with a BigDye Terminator Cycle Sequencing Reaction kit (PE Applied Biosystems).
Primary nucleotide sequences were analyzed with the GENETYX-Mac 11.2 system program. Multiple sequences were aligned using the programs, GENETYX-Mac 11.2 and CLUSTAL W (Version 1.7).
The phylogenetic tree for various rice strains was constructed by organizing the presence or absence of the p-SINE1 members at particular loci in the strains into a data matrix (Cheng et al., 2003), such that the presence of a p-SINE1 member at a given locus was coded 1, and its absence at the same locus was coded 0. Some strains generated two PCR amplified fragments with or without a p-SINE1 member, indicative of both the presence and absence of the member. Such cases were coded 1 to indicate the presence of a p-SINE1 member (Cheng et al., 2003). The neighbor-joining (N-J) method or the UPGMA method were used for tree construction with the computer program PAUP* 4.0b10 (Swofford, 2002). Bootstrap values of the N-J tree were calculated with the same program. The structure program (Pritchard et al., 2000) was used to infer population structure with burn-in 10,000, run length of 100,000, and a model with admixture (Garris et al., 2005).
Nucleotide sequence data with information for p-SINE1 members (r401–r406, r411–r414, r421–r423, r431–r433, and r441–r452) from O. rufipogon appear in the DDBJ/EMBL/GenBank International Nucleotide Sequence Databases under the accession numbers AB201718–AB201745.
A total of 26 new p-SINE1 members were isolated from five strains of O. rufipogon (W1681, W2007, W1943, W0120 and W0593), which represent perennial, annual, or intermediate ecotypes (see Table 1). Their nucleotide sequences were aligned together with 25 previously identified p-SINE1 members from O. sativa (see Fig. 1). With the exception of seven members (r1, r69, r441, r444, r445, r447 and r452), each of the p-SINE1 members contains three common substitution mutations (T at nucleotide position 7, A at 63, and A at 114) (Fig. 1), which have also been identified in members forming the RA (recently-amplified) subfamily. Of 44 RA-subfamily members that have been identified to date, 27 members contained three additional common substitution mutations (C at 77, T at 113, and T at 117), and 11 members contained two additional common substitution mutations (T at 10 and T at 117) (Fig. 1). These suggest that RA-subfamily members consist of three groups, of which two newly identified groups are referred to as RAα and RAβ (Fig. 1).
 View Details | Fig. 1. An alignment of nucleotide sequences of p-SINE1 members. Twenty-six p-SINE1 members (named r401–404, r406, r431–433, r411–414, r421–422, and r441–452) were newly isolated from O. rufipogon strains, and the other 25 members had been previously isolated from O. sativa strains. p-SINE1 members marked with an x are present in all the rice strains examined; those with solid circles indicate insertion polymorphism among strains with the AA genome and with the non-AA genome; those with solid triangles show polymorphisms among strains of all species with the AA genome; and those with open circles are specifically present in strains of O. sativa and/or O. rufipogon. In each p-SINE1 member, nucleotide sequences identical to those in the consensus sequence are indicated by dashes; deleted nucleotides are indicated by slashes; and duplicated sequences are indicated by asterisks. Nucleotide sequences underlined are target-site duplication. Note that members of the RA subfamily (RA, RAα or RAβ) have characteristic mutations in common. |
A total of 108 rice strains (68 of O. sativa, 35 of O. rufipogon, and 5 of other rice species with the AA genome; see Table 1) were examined for the presence or absence of the 26 new p-SINE1 members, as well as the 25 previously isolated p-SINE1 members, by PCR with primer pairs that hybridize to the regions flanking each p-SINE1 member. Most of the RA-subfamily members were found to show insertion polymorphisms among the O. sativa and O. rufipogon strains (Table 1). All the RA-subfamily members, except two (r34 and r443), were not present in the rice strains of species other than O. sativa and O. rufipogon (Table 1).
To determine the relationships of the rice strains examined, they were bar-coded on the basis of the presence or absence of p-SINE1 members at the respective loci, and a phylogenetic tree of the strains was constructed using the bar codes assigned to each of the strains with the neighbor-joining (N-J) method (Fig. 2; see Materials and Methods). According to the phylogenetic tree obtained, the strains examined were classified into three groups, named I, II and III (Fig. 2).
 View Details | Fig. 2. A phylogenetic tree showing relationships among the O. sativa and O. rufipogon strains. The tree was constructed on the basis of p-SINE1-insertion polymorphism in all the strains shown in Table 1 by the N-J method. I, II, and III indicate the three major groups. Japonica strains of O. sativa are indicated by letters in blue; and indica strains are in red; tropical japonica (javanica) strains are underlined and those in insular Southeast Asia are circled. Annual, perennial and intermediate strains of O. rufipogon are respectively shown by letters in green, cyan (greenish-blue) and pink. The other five AA genome species strains are shown in black; a hypothetical ancestor is shown with a dashed line. Strains with an asterisk are O. rufipogon from China. |
During the course of the phylogenetic analysis, we noticed that bootstrap values of the three groups were relatively low (Supplementary Fig. 1A). We deduced tree of strains without all the O. rufipogon strains or all the O. sativa strains, to know how the bootstrap values fluctuate. The tree without O. rufipogon strains showed high bootstrap values, whereas the values of the tree without O. sativa strains stayed low (Supplementary Fig. 1B, C). These suggest that the low bootstrap values were caused by the presence of O. rufipogon strains. We then deduced trees by excluding various combination of O. rufipogon strains, and found that the bootstrap values of the three groups (I, II and III) were relatively high when several perennial-type strains were excluded from the analysis (see Supplementary Fig. 1D by way of example). These facts suggest that the O. rufipogon perennial-type strains have more or less mixed features of the three groups, which might be partly the results of natural hybridizations between different groups.
Because the bootstrap values of the N-J tree were low, we also constructed a phylogenetic tree with the UPGMA method using the same dataset as the N-J tree, and found that the strains were also classified into three groups, which contain the same members as were observed in the Groups I, II and III deduced by N-J method, except that several O. rufipogon perennial strains of Group III (W1965, CP25, etc.) were clustered together with those of Group II (Supplementary Fig. 2). Furthermore, we inferred population structure of the strains with the Structure program (Pritchard et al., 2000). When the number of populations (K) was set at three, the three clusters almost corresponded with Groups I, II and III, except that several O. rufipogon perennial strains of Groups I and III (W1954, W1965, etc.) showed somewhat admixed populations (Supplementary Fig. 3). These results suggest that the classification by the N-J method is valid in spite of the low bootstrap values, which were probably caused by several O. rufipogon perennial strains with mixed features of different groups. Based on these results, we will describe features of the three groups more in detail in the following three sections.
Group I of the N-J tree consisted of six O. rufipogon perennial strains (W1943, W1945, etc.) and O. sativa ssp. japonica strains. The six O. rufipogon strains were grouped with japonica strains also in the UPGMA tree (see Supplementary Fig. 2). These results support the previous idea that the O. rufipogon perennial strains and O. sativa ssp. japonica strains originated from a common ancestor (Cheng et al., 2003). Interestingly, five of the six O. rufipogon perennial strains are from China. This suggests that japonica rice strains originated in the O. rufipogon perennial population in China, as has been suggested previously (Cheng et al., 2003), if there were no introgression from cultivated rice into these O. rufipogon strains.
Japonica strains include those from the temperate area of East Asia and those from the tropical area of Southeast Asia (Table 1). Interestingly in the tree, the strains from the tropical area of insular Southeast Asia (such as Indonesia, Philippines, and Malaysia) (OKA318, Ketang Nangka, etc.) were clustered together, forming a branch that is distinct from those of the strains from the temperate area of East Asia and mainland Southeast Asia (such as Myanmar and Thailand) (Fig. 2). This was not previously noted by Cheng et al. (2003). Most of the strains from the tropical area of insular Southeast Asia have been classified as the strains of the third subspecies, javanica (Morishima, H. and Kurata, N., personal communication). Consequently, these findings suggest that javanica strains distributed in insular Southeast Asia can be distinguished from the other japonica strains by our method. Note that these javanica strains formed a branch also in the UPGMA tree (see Supplementary Fig. 2), which supports the findings above. Interestingly, javanica strains forming a subgroup were much closer than the other japonica strains to the O. rufipogon perennial strains in Group I. This observation suggests that japonica and javanica rice strains have originated in the common ancestral population, which may have consisted of javanica-type strains. Note, however, that several other javanica strains from mainland Southeast Asia (Myanmar and Thailand) (Bs48 and OKA446) and China (OKA701 and OKA718) are not clustered with those from insular Southeast Asia (Fig. 2 and Table 1). It is possible that these strains are cultivars, which have been improved through hybridization by breeders or farmers in these areas from the typical javanica strains. It is also possible that these strains have been misclassified as javanica strains.
Group II of the N-J tree consisted of almost all the O. rufipogon annual strains and two perennial strains (W0120 and W1654) (Fig. 2). Note that these two O. rufipogon perennial strains have been previously shown to cluster with the other perennial strains (Cheng et al., 2003), probably because no p-SINE1 members from O. rufipogon have been included in the previous phylogenetic study. This provides further support to the previous idea that the annual O. rufipogon was derived from primitive perennial O. rufipogon (Cheng et al., 2003). Several additional O. rufipogon perennial strains (W1965, CP25, etc.) were grouped together with the Group II strains in the UPGMA tree (see Supplementary Fig. 2). This also supports the above idea.
Group II included O. sativa ssp. indica strains, which are clustered with O. rufipogon annual strains (Fig. 2). This shows that O. rufipogon annual strains and O. sativa ssp. indica strains are closely related to each other. Of indica strains, however, most (Km5, OKA710, etc.) formed a branch clearly distinct from that of the annual strains of O. rufipogon, indicating both types originated from a common ancestor, and this ancestor is most likely O. rufipogon of the perennial type as described above. Several O. sativa ssp. indica strains (C5924, A39, etc.) are clustered together with annual strains (Fig. 2), as also noted previously (Cheng et al., 2003). This cluster was also formed in the UPGMA tree (see Supplementary Fig. 2). These results indicate that these indica strains, which include those from mountainous regions of South Asia, such as Assam (India) and Nepal (see Table 1 and Fig. 2), may have originated in the annual population, as was also discussed previously (Cheng et al., 2003).
Interestingly in the phylogenetic tree, six indica strains, such as IR36, IR24, Milyang 23, etc., are clustered to form one subgroup. These strains formed a subgroup also in the UPGMA tree (see Supplementary Fig. 2). Such a subgroup was not observed previously by Cheng et al. (2003), probably because no p-SINE1 members identified from O. rufipogon were included in the previous phylogenetic analysis. Note that these indica strains are the progenies, which have been made by breeders in different institutes, such as IRRI, in different countries through hybridization using various indica strains, as well as a few japonica strains. These indica strains have some p-SINE1 members, r62 for example, which are not present in the other indica strains at the corresponding loci, but are present in japonica strains. This suggests that these strains have more sequence homology in such loci, to japonica strains than other indica strains.
The SINE insertion analyses suggest that japonica and indica strains are closely related to different groups of O. rufipogon strains, respectively. This supports the previous idea that O. sativa originated polyphyletically from perennial O. rufipogon (Cheng et al., 2003).
Group III consisted of O. rufipogon strains of perennial and intermediate ecotypes (see Fig. 2). Note that it is reasonable that the ancestral state of the species with the AA genome would have had no insertion of any p-SINE1 member, because the insertion of SINE is thought to be irreversible. Such a hypothetical ancestor with no p-SINE1 members at the respective loci was placed with the strains of Group III in the phylogenetic tree (Fig. 2). The representative strains of the other rice species with the AA genome have only a few p-SINE1 members at their respective loci (see Table 1) and thus they appeared to also be close to the hypothetical ancestor in the phylogenetic tree (Fig. 2). Interestingly, O. rufipogon strains of the intermediate type (W0596, W1266, etc.) were clustered together, forming a subgroup that is the closest to the hypothetical ancestor among O. rufipogon strains. These strains were also clustered together in the UPGMA tree (see Supplementary Fig. 2).
Morishima et al. (1961) and Morishima et al (1984) have discussed that if the habitat is always submerged in water, plants once established may propagate asexually, while propagation by seeds may to some extent be hindered. On the other hand, if the habitat is dry once a year, seed propagation may be obligatory unless the plants can survive the drought. Furthermore, differences between populations of the perennial and annual types seem to be regarded as the difference between species of “stable” and “unstable” habitats. In the paper of Morishima et al. (1984), it was suggested that the intermediate plants are ecologically unstable and when disruptive selection is strong enough, they can be differentiated into perennial and annual types. Therefore, it is likely that the ancestral population of O. rufipogon had the intermediate characters between perennial and annual types. The p-SINE1 insertion analysis above showed that O. rufipogon intermediate strains forming a subgroup are closer to the hypothetical ancestor. This leads us to suggest that the perennial and intermediate O. rufipogon strains originated from the common ancestral O. rufipogon population, which may have consisted of intermediate-type strains.
Including the p-SINE1 members identified in this study, about four hundred p-SINE1 members have been identified so far from various species with the AA genome (340 from O. sativa, 26 from O. rufipogon, 4 from O. barthii, 6 from O. glumaepatula, 8 from O. longistaminata, and 8 from O. meridionalis) (Umeda et al., 1991; Mochizuki et al., 1992; 1993; Motohashi et al., 1997; Cheng et al., 2002; 2003; Ohtsubo et al., 2004). There are 112 RA-subfamily members, including 64 RAα and 18 RAβ members. Interestingly, p-SINE1 RA-subfamily members have not been identified from strains of species with non-AA genomes until now, suggesting that RA-subfamily members have been inserted into the loci after divergence of species with the AA genome. Furthermore, 27 RAα and 11 RAβ members investigated were specifically present in strains of O. sativa and/or O. rufipogon, but not in strains of the other species with the AA genome (Table 1; Fig. 1). This observation suggests that RAα and RAβ group members have retroposed after the divergence of O. sativa and O. rufipogon from the other rice species with the AA genome.
From seven non-RA subfamily p-SINE1 members, three (r441, r447 and r452) were present at particular loci in the strains of species with the AA genome (Table 1), which include O. sativa, O. rufipogon, O. glaberrima, O. barthii, O. glumaepatula, O. longistaminata and O. meridionalis, but were absent in a strain of species with non-AA genome (Xu, 2004). This result indicates that these three members have been inserted in a common ancestor of strains of the AA-genome species. This also suggests that rice species with the AA genome are monophyletically derived.
It is worthwhile to note that there are several p-SINE1 members that are very useful for readily distinguishing different subspecies strains of O. sativa and different ecotypes strains of O. rufipogon. The p-SINE1 member r502, which is present only in the strains of O. sativa and O. rufipogon, but not in those of the other species with the AA genome, can be used to distinguish strains of O. sativa and O. rufipogon from those of the other species with the AA genome (Table 3; see also Table 1). As shown in Table 3, the p-SINE1 member r55 is present in the indica strains, but not in the japonica and javanica strains, whereas the p-SINE1 member r507 is present in the japonica strains, but not in the javanica and indica strains (see also Table 2). These two p-SINE1 members are therefore very useful to distinguish three different subspecies strains of O. sativa. As also shown in Table 3, the p-SINE1 member r215 is present in the annual strains, but not in the perennial and intermediate strains, whereas the p-SINE1 member r503 is present in the perennial and annual strains, but not present in the intermediate strains (see also Table 1). These two p-SINE1 members are therefore very useful for distinguishing between three different ecotype strains of O. rufipogon. Further characterization and classification of more rice strains based on the p-SINE1 insertion polymorphism using the p-SINE1 members including the subspecies- or ecotype-specific members described above will lead us to more accurately ascertain the origins of rice strains.
 View Details | Table 3. Useful p-SINE1 members to distinguish different subspecies strains of O. sativa and different ecotypes strains of O. rufipogon |
We deeply acknowledge to Prof. N. Kurata and Ms. T. Miyabayashi for wild rice strains and DNA, and to Dr. Y. Fukuta for DNA used in this work. We also deeply thank Prof. M. Batzer for his valuable discussions regarding phylogenetic analysis. This work was partly supported by the grant from the Ministry of Agriculture, Forestry and Fisheries of Japan (Rice Genome Project GD2007 to H.O.). J-H.X. and C.C. were the recipients of Japanese Government (Monbukagakusho) Scholarship.
|
Index