Breeding Science 52 : 309-317 (2002)

QTL Analysis for Plant and Grain Characters of -brewing Using a Doubled Haploid Population

Shinya Yoshida*1), Masaru Ikegami1), Junko Kuze2), Keiko Sawada2), Zentaro Hashimoto3), Takashige Ishii2), Chiharu Nakamura3) and Osamu Kamijima2)

1) Hyogo Prefectural Research Center for , Forestry and Fisheries, 1533 Minamino-oka, Befu, Kasai, Hyogo 679-0198, 2) Laboratory of Plant Breeding, Faculty of Agriculture, Kobe University, 1-1 Rokkodai, Nada, Kobe, Hyogo 657-8501, Japan 3) Laboratory of Plant Genetics, Faculty of Agriculture, Kobe University, 1-1 Rokkodai, Nada, Kobe, Hyogo 657-8501, Japan

Rice ( L.) varieties used for brewing sake Introduction are commonly characterized by traits such as large grain size with white-core (an opaque structure inside Rice (Oryza sativa L.) is one of the most important sta- the rice grain). A linkage map was constructed using ple food crops in the world. In Japan, rice is also used for the doubled haploid lines derived from the cross of Reiho (a production of wine, sake. Sake is brewed from cooking variety) and Yamada-nishiki (a sake-brewing steamed rice through fermentation by koji (Aspergillus variety). Random amplified polymorphic DNA (RAPD), oryzae) and yeast. Rice varieties used strictly for sake- amplified fragment length polymorphism (AFLP) and brewing are characterized by a larger grain size than that of simple sequence repeat (SSR) marker systems were em- ordinary cooking rice, and by the presence of an opaque ployed in QTL analysis. A total of 145 markers were structure named white-core located in the center of rice identified and mapped on rice chromosomes. QTLs for grains. These characters are considered to be suitable for plant and grain characters were detected by interval sake-brewing processes. Contents of proteins, amylose and mapping and single point analysis. Several QTLs with a fatty acids are also known to markedly affect the taste and significant contribution were identified for important flavor of sake. However, the large grain size tends to be as- sake-brewing characters including grain size, grain sociated with several undesirable characters such as white- shape, white-core grain rate and protein content. Sever- belly and cracking of grain. White-belly resembles white- al QTLs simultaneously affected the grain weight, width core as an opaque structure in rice grain, but differs in that it and thickness, while QTLs for the grain length independ- is located on the belly side of grains. Grain size is usually ently affected the grain size. QTLs for the white-core evaluated by the grain weight, but the grain weight may be grain rate did not affect the grain size, although one correlated with sevral characters including grain length, QTL for the white-belly grain rate simultaneously af- grain width and grain thickness. However, the predominant fected the grain weight, width and thickness. Several components in determining the grain size remain largely un- QTLs were detected for the protein content in both known. Genetic studies on the grain size and/or shape have brown and polished rice. One QTL on chromosome 4 been conducted, and in many cases polygenic gene systems that was effective for the decrease of the protein content were implicated. Takita (1985) reported that grain length in polished rice showed a positive relation with the and width were controlled by four to five genes. Kato (1989) grain length. One QTL with the largest effect on the observed additive effects on grain length and width, and grain length on chromosome 11 did not contribute to considered that both traits were controlled by different ge- the decrease of the protein content in polished rice. netic systems. A case of major gene control was reported in Therefore, it is suggested that the grain length QTL on that an incomplete dominant major gene Lk-ƒ controled the chromosome 4 might control not only the grain shape long kernel of the variety Fusayoshi (Takeda and Saito but also the internal structure related to the milling ef- 1980). The grain protein content is also known to be con- ficiency and/or location of the storage protein. trolled by polygenes, although Okamoto (1994) identified one or two major genes affecting the nitrogen content in the Key Words: QTL analysis, sake-brewing rice, grain endosperm. Therefore, it is necessary for the breeding of quality, RAPD, AFLP, SSR. sake-brewing rice to identify the genes involved in the deter- mination of grain characters that are required for sake- brewing, and to define interactions between plant and grain characters. For studying many important agronomic traits in rice, Communicated by C. Kaneda QTL analyses have been applied mainly using restriction Received March 28, 2002. Accepted July 20, 2002. fragment length polymorphism (RFLP) (for a review, see *Corresponding author (e-mail: [email protected]) Yano and Sasaki 1997). Recently, amplified fragment length 310 Yoshida, Ikegami, Kuze, Sawada, Hashimoto, Ishii, Nakamura and Kamijima

2 2 2 2 polymorphism (AFLP) markers (Vos et al. 1995) and micro- according to the following formula: h = σ G ⁄ ()σ G + σ E , 2 2 satellite or simple sequence repeat (SSR) markers have be- where σ G is the genotypic variance and σ E is the environ- come alternative choices for a variety of genetic studies. In mental variance. rice, an SSR map of genome-wide coverage has been con- Rice flour was prepared from and also from structed and marker information is open to the public 70 % polished rice of each line using CYCLOTEC 1093 (Panaud et al. 1996, Chen et al. 1997, Temnykh et al. 2000). Sample Mill (Tecator, Höganäs, Sweden) with 0.5 mm mesh Here, we constructed a linkage map based on RAPD, or the measurement of the amylose (AM) and protein con- AFLP and SSR markers using anther culture-derived dou- tents (BP for brown rice and PP for polished rice). The amy- bled haploid lines. QTL analysis was conducted to detect lose content was measured using AutoAnalyzer™II significant chromosomal regions controlling plant and grain (BRAN+LUEBBE, Hamburg, Germany) after de- characters important for sake-brewing. granulation of 100 mg polished rice flour by 5 ml of 0.5 N NaOH, according to the method of Inatsu (1988). BP and PP Materials and Methods were measured using InfraAlyzer® 500 (BRAN+LUEBBE).

Plant materials RAPD analysis We developed 91 doubled haploid lines (DHLs) Total DNA samples of the 91 DHLs (A1 generation) through anther culture of F1 plants from the cross between and their parental varieties were prepared from leaves at the Reiho (an ordinary cooking rice variety) and Yamada- seedling stage by the CTAB method (Murray and Thompson nishiki (a sake-brewing rice variety). Anthers with micro- 1980). A total of 520 random 10-mer primers (QIAGEN spores of the uninucleate stage were excised from spikes, and Operon, Alameda, CA, USA) were used for RAPD analysis. plated on N6 agar medium (Chu 1975) containing 2 mg/l For the amplification reaction, 20 ng genomic DNA, 1 × 2,4-D, 0.1 mg/l BA and 50 g/l sucrose. Induced calli were buffer (Applied Biosystems, Foster City, CA, USA), 4 pmole plated on MS basal agar medium (Murashige and Skoog arbitrary primer, 2 nmole each of dNTPs and 0.5 unit 1962) containing 1 mg/l NAA, 4 mg/l BA and 50 g/l sucrose. AmpliTaq GOLD (Applied Biosystems) were mixed in a The cultures were incubated at 27°C under a 16-h photoperi- total volume of 20 µl. DNA fragments were amplified using od with a light intensity of ca. 38 µEm−2s−1 provided by cool a thermal cycler GeneAmp9600 (Applied Biosystems) with white fluorescent lamps. Regenerated plants (A0 generation) the following cycling conditions: 1 cycle at 95°C for 5 min, were planted in pots with soil and grown in a greenhouse. 45 cycles at 95°C for 1 min, 37°C for 1 min, 72°C for 1.5 min, The ploidy level of the regenerants (A0) was estimated by and post-extension at 72°C for 5 min. Amplified products the observation of the morphological characters and fertility were separated by electrophoresis through a 1.5 % agarose (selfed seed set) of the panicles, according to the method of gel and stained with ethidium bromide. Nakamura et al. (1994). One panicle from each self- fertilized A0 plant was harvested and maintained as one- AFLP analysis panicle-to-one line (A1 generation). A1 and A2 generations AFLP analysis was conducted by using a high- were cultivated in 1998 and 2000, respectively, in an exper- efficiency AFLP genome-scanning system (Mano et al. imental field. For each line, twenty-five seedlings were 2001), which is a modification of the original AFLP method planted in a single-row plot with a distance of 20 cm between (Vos et al. 1995). Total DNA (150 ng) was digested with plants and a row space of 30 cm. EcoRI and MseI, and EcoRI and MseI adaptors were ligated to the ends of the restriction fragments. Twenty cycles of Data collection and analysis PCR were performed for pre-amplification: 30 sec at 94°C, Five plants per line were randomly selected from the A1 1 min at 56°C and 1 min at 72°C. EcoRI and MseI primers, and A2 generations of DHLs, and culm length (CL), panicle which did not contain additional nucleotides at the 3′ ends, length (PL) and panicle number (PN) were measured. The were used for the pre-amplification. Selective amplification data on the A1 generation were used to evaluate the level of was performed using EcoRI and MseI primers that contained genetic fixation of each line. Fully mature grains were har- three selective nucleotides at the 3′ ends. Before the start of vested from ca. 20 plants in each line. Sixty-four random the first cycle, a 2-min denaturation step was performed at grain samples per line were used for the determination of the 94°C. The selective amplification consisted of 30 cycles of grain length (GL) and width (GWh), and 20 random samples PCR with the following three phases: the first phase with 1 per line were used for the determination of the grain thick- cycle: 30 sec at 94°C, 30 sec at 68°C, and 1 min at 72°C; the ness (GT). Grain weight (GWt, 1000 grains) was measured second touch-down amplification phase with 16 cycles with twice per line and adjusted to a moisture content of 13.5 % the annealing temperature decreasing stepwise from 67.3°C after the determination of the grain moisture of all the lines. to 58.5°C and the last phase with 22 cycles: 30 sec at 94°C, Percentages of occurrence of white-core (WC), white-belly 30 sec at 56°C and 1 min at 72°C. The amplified products (WB) and cracked grain (CG) were calculated using 100 were fractionated by electrophoresis through 13 % non- grains and a statistical analysis was performed using arcsine- denaturing polyacrylamide gels and stained using silver- transformed values. Heritability (h2) was calculated staining kit, Sil-Best Stain for protein/PAGE (Nacalai QTL analysis for sake-brewing rice characters 311

Tesque, Kyoto, Japan). by environmental factors. Since the high heritability values suggested that the DHLs were genetically fixed, further SSR analysis analyses were performed using the data obtained in 2000. In SSR analysis was conducted using 253 primer sets de- 2000, data on all the traits were collected including WB, CG, veloped by Chen et al. (1997) and Temnykh et al. (2000). AM, BP and PP. All of these traits seemed to show a normal Amplification was performed using 20 ng DNA, 0.2 µM of distribution typical of quantitative traits (Fig. 1). Though the each primer, 100 µM of each dNTP, 10 mM Tris-HCl (pH parents did not show appreciable differences in GT and WB, 8.3), 50 mM KCl, 1.5 mM MgCl2, 0.1 % Triton X-100 and the values of the DHLs displayed a wide range like all the 0.5 unit Taq DNA polymerase (TOYOBO, Osaka, Japan) in other values. a 25 µl reaction mixture. The PCR profile was as follows: 1 CL was highly and positively correlated with PL, CG cycle of 5 min at 94°C, 35 cycles of 1 min at 94°C, 1 min at and GL and negatively correlated with BP and PP (Table 2). 55°C, 2 min at 72°C, and 5 min at 72°C for final extension. PN was negatively correlated with PL, CG, WC and GWh. Amplified products were separated by electrophoresis GWt showed the highest positive correlation with GWh fol- through 4 % polyacrylamide denaturing gels, and stained us- lowed by GT, GL and WB. GL was not correlated with GWh ing a silver-staining kit, Silver Sequence™ DNA Staining and GT, while a high positive correlation was observed be- System (Promega, Madison, WI, USA). tween GWh and GT. WC was positively correlated with GL among the traits of seed size, while WB was correlated with Linkage analysis and QTL analysis GT, GWh and GWt. AM was positively correlated with WC Linkage analysis was performed with MAPMAKER and GL, and negatively correlated with PP, which showed Ver. 2.0 for Macintosh (Lander et al. 1987). Linkage groups the highest positive correlation with BP. were assigned to rice chromosomes according to the pub- lished rice map (Chen et al. 1997, Temnykh et al. 2000). Detection of DNA polymorphisms and linkage analysis QTL analysis was performed using the QGENE program A total of 520 arbitrary primers were screened in the (Nelson 1997). Putative QTLs were detected by interval RAPD analysis using DNA extracted from the parental vari- mapping analysis and statistics were obtained by single- eties. Sixty-two primers (11.9 %) produced polymorphic point analysis. The threshold probability level was less than fragments and 93 bands were scored. There were however 0.005. only 40 distinct segregation patterns in the DHLs, indicating that more than half of the bands were derived from the same Results regions of the rice genome. Six out of 48 AFLP primer sets (12.5 %) amplified 10 polymorphic bands between the par- Plant and grain characters ents, and all the bands showed unique segregation patterns. The parental varieties, Yamada-nishiki and Reiho used Informative selective primer sets and polymorphic bands are for the production of DHLs were selected based on their listed in Table 3. In the SSR analysis, 253 primer sets were similar heading date: all the DHLs and the parents headed tested and 47 (18.6 %) gave polymorphic bands. A signifi- within one week in both 1998 and 2000. Three morphologi- cant segregation distortion was detected in five polymorphic cal characters and five grain characters were measured in the bands by χ2 test of fitness to a 1 : 1 segregation in the DHLs DHLs and the parental varieties. In 2000, CL, PN, GWt, (P < 0.01). A framework map was first constructed according GWh and WC were all larger than in 1998 (Table 1). Herita- to the SSR marker order reported by Temnykh et al. (2000). bility (h2) estimates varied from 70.9 % to 95.6 % for these RAPD and AFLP markers were then integrated into the SSR characters. GWh, GT and WC showed intermediate levels of framework map (Fig. 2). Since three RAPDs and one AFLP heritability. Heritability of PN recorded the lowest values, could not be assigned to any linkage groups, and some re- reflecting the fact that this character is considerably affected gions did not have any polymorphic markers, long and short

Table 1. Variance and heritability for plant and grain characters in the doubled haploid population from the cross of Reiho and Yamada-nishiki CL (cm) PL (cm) PN GWt (g) GL (mm) GWh (mm) GT (mm) WC (%) Means1) 1998 80 20 14 24.2 4.9 2.6 2.0 23.9 2000 93 20 17 28.8 5.0 2.8 2.0 34.6 Variances σ2G 173.2 1.9 8.3 2.17 0.0196 0.0036 0.0029 0.1698 σ2E 7.9 0.7 3.4 0.62 0.0018 0.0005 0.0005 0.0316 σ2P 181.1 2.6 11.7 2.79 0.0214 0.0041 0.0034 0.2014 h2 95.6 74.4 70.9 77.6 91.6 88.0 84.0 84.3 1) Means were averaged for all the DHLs and parental varieties, Reiho and Yamada-nishiki each year. 312 Yoshida, Ikegami, Kuze, Sawada, Hashimoto, Ishii, Nakamura and Kamijima

Fig. 1. Distribution of plant and grain characters in 91 DHLs derived from the cross between Reiho and Yamada- nishiki. Solid and open triangles indicate means of Yamada-nishiki and Reiho, respectively. arms of chromosomes 1 and 2 could not be connected. Chro- detected by the marker F16722 on chromosome 1 accounted mosome 10 was also divided into two short segments. for 55 % of the total phenotypic variance, and this Yamada- nishiki allele increased CL by ca. 10 cm. Each of the other QTLs for plant characters two QTLs accounted for ca. 10 % of the total phenotypic Three QTLs were detected for CL on chromosomes 1, variance. Two QTLs were detected for PL, one of which on 2 and 4 (Fig. 2 and Table 4). The Yamada-nishiki alleles at chromosome 1 was associated with the marker F16722 for these QTL loci contributed to the increase of CL. One QTL CL while the other was on chromosome 5. The percentage of QTL analysis for sake-brewing rice characters 313

Table 2. Correlation coefficient among plant and grain characters in the doubled haploid population from the cross of Reiho and Yamada- nishiki CL PL PN GWt GL GWh GT WC WB CG AM BP PL 0.650** PN −0.265* −0.381** GWt 0.245* 0.161 −0.184 GL 0.329** 0.334** −0.187 0.481** GWh 0.121 0.071 −0.296** 0.722** 0.086 GT 0.044 −0.025 −0.217* 0.666** 0.048 0.758** WC 0.081 0.201 −0.320** 0.154 0.335** 0.139 0.104 WB −0.143 −0.197 0.017 0.339** 0.004 0.450** 0.491** −0.202 CG 0.450** 0.214* −0.360** 0.194 −0.001 0.215* 0.239* −0.224* −0.036 AM 0.224* 0.274* −0.210 0.255* 0.391** −0.044 0.113 0.394** −0.143 −0.035 BP −0.302** −0.151 −0.226* −0.243* −0.238* 0.016 −0.019 −0.003 0.066 −0.119 −0.118 PP −0.379** −0.172 −0.072 −0.236* −0.372** 0.004 −0.070 −0.181 0.130 −0.131 −0.278** 0.767** *, ** significant at 5 % and 1 % levels, respectively. phenotypic variance explained by these QTLs was about 10 % mosome 5 showed the largest positive effect on both GWh each. One QTL on chromosome 11 was associated with PN, and GT. explaining 31 % of the total phenotypic variance. Only one QTL for WC near F16737 on chromosome 12 was detected by interval mapping, while another QTL near QTLs for grain characters Me59801a (chromosome location unknown) was detected Yamada-nishiki showed six QTLs for the increase of by single-point analysis. The Yamada-nishiki allele at the GWt on five chromosomes (Fig. 2 and Table 4). Each of former QTL on chromosome 12 accounted for 14 % of the these QTLs accounted for 10 to 18 % of the phenotypic vari- total phenotypic variance, while the Yamada-nishiki allele at ance, and the cumulative effect was about 75 %. Four QTLs the latter QTL accounted for 17 % of the variance. Three were detected for GL. One Yamada-nishiki allele at this QTLs for WB were detected, all on chromosome 5, of which QTL which was detected by the marker B01482 on chromo- one Yamada-nishiki allele at RM31 accounted for 21 % of some 11 displayed the largest positive effect, explaining the variance. This QTL with the largest effect coincided with 23 % of the total phenotypic variance. Four and three QTLs the QTL for GWt, GWh and GT. Of these QTLs, the Yamada- were detected for GWh and GT, respectively, of which two nishiki alleles contributed to the increase of WB as well as detected by the markers T082023 and RM31 were common the above three grain characters, while the Reiho alleles at and mapped on chromosomes 4 and 5, respectively. One the remaining two QTLs contributed to the increase of only Yamada-nishiki allele at the QTL detected by RM31 on chro- WB. Four QTLs were detected for CG. The largest effect was associated with the QTL detected by F16722 on chro- mosome 1, which was a marker for the QTL for CL and PL. Table 3. List of informative AFLP primer sets between Reiho and Each of the two QTLs of Yamada-nishiki and Reiho alleles Yamada-nishiki and polymorphic bands contributed to the increase of CG. Selective Polymorphic For AM in polished rice, although one QTL was detect- Primer set 1) sequence bands (kb) ed by the marker Me5907b on chromosome 8, this QTL Me0238 M-AAG/E-TGG 244 a accounted for only 10 % of the total phenotypic variance. 117 b Four QTLs were detected for BP, of which one QTL with 91 c the largest effect detected by the marker RM206 on chromo- Me5801 M-CTG/E-AAA 909 a some 11 was also a marker for PN. Five QTLs were detected 150 a for PP. Although the QTLs detected by RM255 on chromo- Me5907 M-CTT/E-AGT 278 a some 4 and F16722 on chromosome 1 showed the large effect 231 b (18 % and 14 %, respectively) on PP, no significant effects on 99 b BP were detected at these QTL loci. Me6433 M-CCC/E-TAA 487 a 98 b Discussion Me6455 M-CCC/E-CGT 272 a 103 a Yamada-nishiki has long been a leading rice variety for Me6460 M-CCC/E-CTC 581 a sake-brewing in Japan. This variety was bred through hy- 128 a bridization between Yamada-ho and Tankan-wataribune, 1) Polymorphic bands with the same letters amplified by each both of which are domestic and local sake-brewing varieties primer set indicate the same segregation pattern. with desirable traits for sake-brewing such as large grain 314 Yoshida, Ikegami, Kuze, Sawada, Hashimoto, Ishii, Nakamura and Kamijima

Fig. 2. Molecular map of the DHLs from F1 between Reiho and Yamada-nishiki showing QTLs identified for plant and grain characters. Abbreviations of the characters are described in the Results. QTLs with enclosed characters show positive effects from Yamada-nishiki alleles. RAPD markers are denoted by the primer names and approximate mo- lecular size. AFLP markers are denoted by the primer set names and alphabet letters of the polymorphic fragments described in Table 4. SSR markers are designated by RM. Markers with black-back show a significant segregation distortion (p < 0.01). size, high white-core rate and low protein content. We con- Japanese varieties (Kubo et al. 1998). A higher level of ducted a QTL analysis to detect the chromosomal regions polymorphism (18.6 %) was detected by SSR analysis. controlling plant and grain characters including those re- DHLs are suitable for QTL analysis because they are homo- quired for sake-brewing, using RAPD, AFLP and SSR zygous recombinant lines. However, segregation distortion markers, and DHLs derived from F1 plants of the cross be- is often observed in the DHLs derived from anther culture tween Reiho and Yamada-nishiki. In the RAPD analysis, we (Yamagishi et al. 1996, Xu et al. 1997). In our study, several detected polymorphic bands generated by 62 (11.9 %) out of markers on chromosomes 4 and 10 showed a significant 520 primer sets. This polymorphism frequency is almost the segregation distortion. We omitted these distorted markers same as that estimated by RFLP and RAPD analyses in in our QTL analysis. QTL analysis for sake-brewing rice characters 315

Table 4. Molecular markers associated with QTLs for plant and grain characters Marker Chr Direction R2 LOD P AE CL F16722 1 Y 0.549 15.89 <0.0001 −10.1 J12738 2 Y 0.100 2.13 0.0020 −4.4 P06724 4 Y 0.106 2.26 0.0014 −4.7 PL F16722 1 Y 0.097 2.04 0.0025 −0.5 I041663 5 Y 0.096 2.04 0.0025 −0.5 PN RM206 11 Y 0.307 7 <0.0001 −2.1 GWt RM145 2 Y 0.180 3.92 <0.0001 −0.08 X03617 3 Y 0.106 2.23 0.0016 −0.06 RM255 4 Y 0.110 2.24 0.0015 −0.06 T082023 4 Y 0.104 2.19 0.0017 −0.06 RM31 5 Y 0.139 2.99 0.0003 −0.07 RM332 11 Y 0.107 2.27 0.0014 −0.06 GL G031003 1 Y 0.119 2.54 0.0007 −0.05 RM255 4 Y 0.164 3.47 0.0001 −0.06 RM204 6 Y 0.130 2.48 0.0009 −0.05 B01482 11 Y 0.234 5.31 <0.0001 −0.07 GWh X03617 3 Y 0.087 1.82 0.0043 −0.02 T082023 4 Y 0.160 3.49 0.0001 −0.02 RM31 5 Y 0.271 6.31 0.0000 −0.03 Me0238b 6 Y 0.091 1.87 0.0038 −0.02 GT W131132 2 Y 0.099 2.04 0.0025 −0.01 T082023 4 Y 0.097 2.03 0.0025 −0.01 RM31 5 Y 0.196 4.36 <0.0001 −0.02 WC F16737 12 Y 0.142 3.02 0.0002 −0.16 Me5801a unknown Y 0.173 3.8 0.0000 −0.17 WB RM249 5 R 0.087 1.82 0.0043 0.14 RM163 5 R 0.089 1.83 0.0042 0.15 RM31 5 Y 0.207 4.64 <0.0001 −0.22 CG F16722 1 Y 0.143 3.09 0.0002 −0.16 RM232 3 Y 0.124 2.61 0.0006 −0.15 RM21 11 R 0.117 2.08 0.0024 0.15 A03994 12 R 0.097 2.06 0.0024 0.13 AM Me5907b 8 R 0.099 1.92 0.0034 0.37 BP J12738 2 R 0.121 2.58 0.0007 0.10 RM168 3 Y 0.110 2.3 0.0013 −0.10 RM206 11 R 0.151 3.09 0.0002 0.12 U03579 12 R 0.114 2.37 0.0011 0.10 PP F16722 1 R 0.143 2.92 0.0003 0.11 L19251 3 Y 0.109 2.19 0.0018 −0.10 RM255 4 R 0.182 3.66 0.0001 0.12 RM206 11 R 0.124 2.41 0.0010 0.10 E151353 12 R 0.099 1.96 0.0031 0.09 Direction: Parental variety with positive effect (R: Reiho, Y: Yamada-nishiki) R2: Percentage of total phenotypic variance by the QTL AE: Additive effect of the Reiho allele (AE of WC, WB and CG indicate the arcsin-transformed values: 0 < AE < π)

It is generally recognized that grain characters are sig- Shinbashi (1982) reported that the sd-1 gene reduced the nificantly correlated with growth characters. Reiho harbours grain size. Our results using the DHLs were in agreement the semidwarf gene (sd-1) derived from a local variety with the conclusion of Ogi et al. (1993). We detected a QTL Jikkoku that contributed significantly to Japanese rice breed- with the highest effect on culm length on rice chromosome ing (Kikuchi et al. 1985). Ogi et al. (1993) reported that the 1. It is considered that the sd-1 is located in this QTL region sd-1 gene did not reduce the panicle length but rather in- (Cho et al. 1994, Yu et al. 1995, Maeda et al. 1997). A com- creased the panicle number, while grain length and grain parison between a SSR map and a RFLP map (Chen et al. width were not affected. On the other hand, Kinoshita and 1997) suggests that the QTL for culm length on 316 Yoshida, Ikegami, Kuze, Sawada, Hashimoto, Ishii, Nakamura and Kamijima chromosome 1 detected in our study corresponds to the mo- Grain protein content has been found to be affected by lecular markers linked to the sd-1 locus. polygenes (Higashi and Kushibuchi 1976, Kataoka 1978, Grain length is known to be dominated by the shape of Kambayashi et al. 1984). Particularly the protein content in husk, whereas grain length and grain width are controlled by brown rice is markedly affected by cultural and environmen- different genes (Takita 1981). In our study, the QTLs for tal factors. Higashi and Kushibuchi (1976) reported that grain length were not correlated with grain weight, width broad-sense heritability of the protein content was 0.45 in a and thickness, while the QTLs of these three grain characters japonica F2 population. Kataoka (1978) reported that broad- were significantly correlated. Only one QTL for grain length sense heritability ranged from 0.39 to 0.78 in several other on chromosome 4 was weakly associated with grain weight. F2 populations and that there was a negative correlation be- This QTL might be the one identified for grain length by tween the protein content and other characters (grain weight, Xiao et al. (1996), Lin et al. (1996) and Redona and Mackill culm length and grain yield). Our observation showed that (1998). Redona and Mackill (1998) reported that QTLs on the grain protein content in brown and polished rice was sig- chromosomes 3 and 7 controlled both grain length and nificantly correlated with the culm length. However, a QTL width. It is unlikely that these indica QTLs would be detect- with the highest effect on the grain protein content of brown ed among japonica varieties, because japonica varieties do rice was effective in the panicle number, and a QTL for the not display the long grain character. protein content, F16722, was detected near the putative Takeda and Saito (1983) who observed a high correla- semidwarf gene (sd-1) loci in polished rice, but not in brown tion between the grain weight and the white-belly rate, sug- rice. The grain protein content may be controlled either by gested that the major gene Lk-ƒ for large grain pleiotropical- direct and independent genetic effects or by secondary ef- ly affected the white-belly rate. Kamijima (1980) reported fects of sink-source balance. It is suggested that the QTL on that the weight of the white-core grain was greater than that chromosome 11 may regulate the grain protein content of the non-white-core grain even in the same variety. There- through the latter mechanism. The relationship between the fore, it has been considered that the white-core rate and the grain protein content and the characters for grain yield white-belly rate are both affected by the grain size. How- should be further analyzed in detail. One QTL for the protein ever, QTLs for the white-core rate did not show any significant content in polished grain on chromosome 4 showed an asso- association with the QTLs for the grain size in our study. On ciation with one QTL for grain length. However, the QTL the other hand, a QTL with the largest effect on the white- with the largest effect on grain length on chromosome 11 did belly rate showed a significant association with the QTLs for not contribute to the decrease of the protein content in pol- grain weight, width and thickness. It is thus suggested that ished rice. Therefore, the QTL for grain length on chromo- white-core and white-belly are formed through different some 4 might control not only the grain shape but also the pathways during grain development. The white-belly rate internal structure related to the milling efficiency and/or was not affected by the QTLs for grain width and thickness, location of the storage protein. except for a QTL on chromosome 5. Several QTLs with dif- In conclusion, we detected several significant QTLs for ferent actions thus might be involved in the determination of the grain characters of sake-brewing rice that are related to grain width and thickness. QTLs for the white-core rate did growth characters, i.e. culm length and panicle number. not show any association with the amylose content that was Multiple QTLs which were found for each grain character, significantly correlated with the white-core rate. It is possi- are regulated by different mechanisms. Many QTLs showed ble that another putative QTLs for white-core rate associated pleiotropic effects on several different characters. For exam- with amylose content may have been undetected in our ple, the grain size increased without any increase in the study. white-belly rate in individual plants harboring the Yamada- Grain thickness affects the cracked grain rate both in nishiki allele for the grain weight QTL near RM145 (Table 4 japonica and indica varieties and varietal differences in the and Fig. 2). These observations suggest that the reduction of cracked grain rate are more stable with a higher heritability the white-belly rate and cracked grain rate is possible in a in varieties (Takita 1992). However, none of program for sake-brewing rice breeding aimed at obtaining a the QTLs for grain cracking were associated with grain larger grain size. A molecular map with a higher density is thickness. Especially, three QTLs on chromosomes 3, 11 required to evaluate the suitability of these QTLs for breed- and 12 did not show any association with any of the QTLs ing for grain quality in sake-brewing rice. detected for the grain characters. Therefore, these QTLs may control the resistance to cracking of the rice grains. On the Acknowledgements other hand, grain cracking is caused by physical stresses that occur through repeated process of hydration and dehydration This work was supported in part by a grant-in-aid from of rice grains (Nagato et al. 1964). The QTL with the largest the Japanese Ministry of Education, Culture, Sports, Science effect on the cracked grain rate was associated with the QTL and Technology (to CN, no. 11794003). for culm length, suggesting that this QTL causes grain cracking due to the rehydration process accompanied with lodging. QTL analysis for sake-brewing rice characters 317

Literature Cited Nagato, K., M. Ebata and M. Ishikawa (1964) On the formation of cracks in rice kernels during wetting and drying of paddies. Chen, X., S. Temnykh, Y. Xu, Y.G. Cho and S.R. McCouch (1997) Jpn. J. Crop Sci. 33: 82- 89. Development of a microsatellite framework map providing Nakamura, K., H. Suzuki, K. Hattori and Y. Futuhara (1994) Identifica- genome-wide coverage in rice (Oryza sativa L.). Theor. Appl. tion of ploidy level of the regenerated plants by anther culture Genet. 95: 553-567. in rice. Breed. Sci. 44: 19-22. Cho, Y.G., M.Y. Eun, S.R. McCouch and Y.A. Chae (1994) The semi- Nelson, J.C. (1997) QGENE: software for marker-based genomic anal- dwarf gene, sd-1, of rice (Oryza sativa L.). II. Molecular map- ysis and breeding. Mol. Breed. 3: 239-245. ping and marker-assisted selection. Theor. Appl. Genet. 89: 54- Ogi, Y., H. Kato, K. Maruyama, A. Saito and F. Kikuchi (1993) Identifi- 59. cation of RFLP markers closely linked to the semidwarfing Chu, C.C. (1975) Establishment of an efficient medium for anther gene at the sd-1 locus in rice. Jpn. J. Breed. 43: 141-146. culture of rice through comparative experiments on nitrogen Okamoto, M. (1994) Studies on effect of chemical components on sources. Sci. Sin. 18: 659-668. stickiness of cooked rice and their selection method for breed- Higashi, T. and K. Kushibuchi (1976) Studies on breeding for high ing. Bull. Chugoku Agr. Exp. Stn. 14: 1-68 (in Japanese). protein rice. II. Inheritance of the high protein property in a Panaud, O., X. Chen and S.R. McCouch (1996) Development of micro- Japonica rice cross. Jpn. J. Breed. 26: 17-24 (in Japanese). satellite markers and characterization of simple sequence length Inatsu, O. (1988) Analysis method of eating quality. In “Studies on im- polymorphism (SSLP) in rice (Oryza sativa L.). Mol. Gen. proving the eating quality of Hokkaidou rice” Rep. Hokkaido Genet. 252: 597-607. Pref. Agr. Exp. Stn. No. 66: 3-6 (in Japanese). Redona, E.D. and D.J. Mackill (1998) Quantitative trait locus analysis Kambayashi, K., I. Tsurumi and T. Sasahara (1984) Genetic studies on for rice panicle and grain characteristics. Theor. Appl. Genet. improvement of protein content in rice grain. Jpn. J. Breed. 34: 96: 957-963. 356-363 (in Japanese). Takeda, K. and K. Saito (1980) Major genes controlling grain size of Kamijima, O. (1980) Studies on breeding rice for sake brewery. Sci. rice. Jpn. J. Breed. 30: 280-282 (in Japanese). Rep. Fac. Agr. Kobe Univ. 14: 25-30 (in Japanese). Takeda, K. and K. Saito (1983) Heritability and genetic correlation of Kataoka, K. (1978) Inheritance of grain protein content in rice. Jpn. J. kernel weight and white belly frequency in rice. Jpn. J. Breed. Breed. 28: 263-268 (in Japanese). 33: 468-480 (in Japanese). Kato, T. (1989) Diallel analysis of grain size of rice (Oryza sativa L.) Takita, T. (1981) Breeding for large grain in rice and varieties in hull Jpn. J. Breed. 39: 39-45. development. Annu. Rep. Natl. Agric. Sci. Jap. Div. Genetics. Kikuchi, F., N. Itakura, H. Ikehashi, M. Yokoo, A. Nakane and K. p. 26-27. Maruyama (1985) Genetic analysis of semidwarfism in high- Takita, T. (1985) Inheritance of grain size and the relationship between yielding rice varieties in Japan. Bull. Nat. Inst. Agri. Sci. D 36: grain size and other characters in rice. Bull. Natl. Agric. Res. 125-146 (in Japanese). Center 3: 55-71 (in Japanese). Kinoshita, T. and N. Shinbashi (1982) Identification of dwarf genes and Takita, T. (1992) Varietal differences in cracked formation in Japonica their character expression in the isogenic background. Jpn. J. and Indica rice. Jpn. J. Breed. 42: 397-402 (in Japanese). Breed. 32: 219-231. Temnykh, S., W.D., Park, N. Ayres, S. Cartinhour, N. Hauck, L. Kubo, T., A. Yoshimura and N. Iwata (1998) Trial construction of mo- Lipovich, Y.G. Cho, T. Ishii and S.R. McCouch (2000) Mapping lecular linkage map of Japonica rice. J. Fac. Agr. Kyushu Univ. and genome organization of microsatellite sequences in rice 43: 95-101. (Oryza sativa L.). Theor. Appl. Genet. 100: 697-712. Lander, E.S., P. Green, J. Abrahamson, A. Barlow, M.J. Daly, S.E. Vos, P., R. Hogers, M. Bleeker, M. Rejians, T. van de Lee, M. Hornes, Lincoln and L. Newburg (1987) MAPMAKER: an interactive A. Frijers, J. Pot, J. Peleman, M. Kuiper and M. Zabeau (1995) computer package for primary genetic linkage map of experi- AFLP: a new concept for DNA fingerprinting. Nucleic Acids mental and natural populations. Genomics 1: 174-181. Res. 23: 4407-4414. Lin, H.X., H.R. Qian, J.Y. Zhuang, J. Lu, S.K. Min, Z.M. Xiong, N. Xiao, J., J. Li, L. Yuan and S.D. Tanksley (1996) Identification of QTLs Huang and K.L. Zheng (1996) RFLP mapping of QTLs for affecting traits of agronomic importance in a recombinant in- yield and related characters in rice (Oryza sativa L.). Theor. bred population derived from a subspecific rice cross. Theor. Appl. Genet. 92: 920-927. Appl. Genet. 92: 230-244. Maeda, H., T. Ishii, J. Kuroda, M. Horimoto, I. Takamure, T. Kinoshita Xu, Y., L. Zhu, N. Huang and S.R. McCouch (1997) Chromosomal re- and O. Kamijima (1997) High density molecular map of semi- gions associated with segregation distortion of molecular mark- dwarfing gene, sd-1, in rice (Oryza sativa L.). Breed. Sci. 47: ers in F2, backcross, doubled haploid, and recombinant inbred 317-320. population in rice (Oryza sativa L.). Mol. Gen. Genet. 253: Mano, Y., S. Kawasaki, F. Takaiwa and T. Komatsuda (2001) Construc- 535-545. tion of a genetic map of (Hordeum vulgare L.) cross Yamagishi, M., M. Yano, Y. Fukuta, K. Fukui, M. Otani and T. Shimada ‘Azumamugi’ × ‘Kanto Nakate Gold’ using a simple and effi- (1996) Distorted segregation of RFLP derived from anther cul- cient amplified fragment length polymorphism system. Genome ture of an F1 of rice. Genes Genet. Syst. 71: 37-41. 44: 284-292. Yano, M. and T. Sasaki (1997) Genetic and molecular dissection of Murashige, T. and F. Skoog (1962) A revised medium for rapid growth quantitative traits in rice. Plant Mol. Biol. 35: 145-153. and bioassay with tobacco tissue cultures. Physiol. Plant. 15: Yu, Z.H., S.R. McCouch, T. Kinoshita, S. Sato and S.D. Tanksley (1995) 473-493. Association of morphological and RFLP markers in rice Murray, M.G. and W.F. Thompson (1980) Rapid isolation of high mo- (Oryza sativa L.). Genome 38: 566-574. lecular weight plant DNA. Nucleic Acids Res. 8: 4321-4325.