Semidwarf Gene d60 Affected by Ubiquitous Gamete Lethal Gene gal Produced Rare Double Dwarf with d30 via Recombination Breaking Repulsion-Phase Linkage on Rice Chromosome 2

The genotype of gal and d60 were investigated in 33 rice varieties chosen from representative semidwarf and dwarf rice varieties. These were crossed with three tester lines, the d60Gal line (genotype d60d60GalGal), the D60gal line (Koshihikari, D60D60galgal), and the D60Gal line (D60D60GalGal). Each F1 plant was measured for culm length, and seed fertility. As a result, all F1 lines with the d60Gal line showed tallness and partial sterility, reduced by 25% in average from those with the D60gal line (Koshihikari) and the D60Gal line. These data indicated that the genotype of the 33 varieties is D60D60galgal and that the d60 locus is not allelic to those of sd1, d1, d2, d6, d18k, d29, d30, d35, d49, d50, and qCL1 involved in the 33 varieties. In addition, the gal gene is not complementarily activated with the semidwarf and dwarf genes described above, other than d60. The Gal gene will be ubiquitously distributed in rice. It is emphasized that Gal is a rare and valuable mutant gene essential to the transmission of d60. The double dwarf genotype of homozygous d30d60 was rarely gained in the F3 of the d30 line × d60 line by breaking their repulsion d60-D30 linkage on chromosome 2.


Introduction
The breeding program that has made the greatest contribution in the history of mankind is the 'green revolution' in which the production of grain was dramatically increased in the 1960s with the development of dwarf varieties of rice and wheat [1]. Dwarfing prevents plants from lodging at their full-ripe stage, which makes them lodging-resistant to wind and rain, and has enhanced their adaptability for heavy maturing, which has dramatically improved (up to double) rice yields, and so has contributed to the stabilization of yields all over the world. Surprisingly, semidwarf rice varieties developed independently using different native varieties or artificially induced mutant lines as mother plants, which are controlled by a single dwarf gene sd1. This is a defective C20-oxidase gene present in a late step in the gibberellin (GA) biosynthesis pathway [2], making options for dwarf breeding limited.
In order to find a novel dwarf gene to replace sd1, the first author conducted gene analyses focusing on Hokuriku 100, a mutant line with culms approximately 15 cm shorter than those of the Koshihikari variety. A novel dwarf gene, d60, was discovered, which gives rise to a good plant type with erect leaves by shortening culms by approximately 20%. Furthermore, d60 complements the gametic lethal gene, gal, to cause gametic lethality [3]. For example, in the F 1 hybrid (genotype D60d60Galgal) of Koshihikari (D60D60galgal) × Hokuriku 100 (d60d60GalGal), male and female gametes having both gal and d60 become gametic lethal, and the pollen and seed fertility decrease to 75%. As a result, the F 2 progeny show a unique mode of inheritance that is segregated into a ratio of 6 fertile long-culm (4D60D60:2D60d60GalGal):2 partially fertile long-culm (D60d60Galgal = F 1 type):1 dwarf (d60d60GalGal). Moreover, the isogenic line that was introduced with both d60 and sd1 derived from Jukkoku [4,5] into Koshihikari by backcrossing [3], viz. the d60sd1 line, and became the extreme-dwarf, indicating that d60 is functionally independent from sd1 and not related to the GA1 biosynthesis pathway [3]. Above all, d60 is expected to diversify semidwarf breeding as a novel alternative of sd1. However, in the process of cross breeding, d60 may cause gamete sterility if the counter parent has gal, and would result in an abnormal F 2 segregation in an 8:1 ratio. Moreover, d60 may affect the segregation of linked genes in the process of heredity. In this study we show: (1) the distribution of gal and d60 were investigated in 33 representative semidwarf or dwarf varieties; and (2) double dwarfness of d60 and linked d30 was rarely gained from the F 3 generation derived from the cross d30 and d60 line.

Genotyping Using the Test Crossed F 1 Lines
Each F 1 plant was measured for culm length and seed fertility. Three tester lines were isogenic lines in the genetic background of Koshishikari, which has a different single allele for D60/d60 and Gal/gal loci, namely d60Gal, D60gal, and D60Gal. Taking into account the expectation that if the test subject has gal, F 1 with the d60Gal line shows partial sterility, and both the F 1 with the D60Gal line and the D60gal line shows fertility. On the other hand, if the test subject has a d60 allele, the F 1 with the d60Gal line shows dwarfness, the F 1 with the D60gal line shows partial sterility, and the F 1 with the D60Gal line shows fertility. Each of the F 1 plants were scored with heading time and culm length in the field. The length between the ground surface and the panicle base of the main culm was measured as the culm length for all plants. The time when the tip of the panicle first emerged from the flag leaf sheath was recorded as the heading time for all plants. Three panicles were harvested from each F 1 plant, and the number of filled and unfilled spikelets was counted for each panicle. The percent of seed fertility was calculated as the number of filled spikelets divided by the total number of spikelets multiplied by 100. The genotype was determined by seed fertility. Aceto-carmine squash mounts to stain the pollens of several F 1 lines were conducted for Olympus BX40 microscopic examination.

Linkage Analysis for d60
Firstly, 318 F 2 plants of a marker gene line FL212 [7] that has d30 and gh2 on chromosome 2(D60D60galgal) and the d60 line (d60d60GalGal) was used for segregation analysis for the marker genes and d60. The result showed that the segregation ratio of wild type to d30 homozygote at the d30 locus was 195:123, and wild type to gh2 homozygote at the gh2 locus was 218:100, and that both deviated significantly from 3:1 (Figure 1). When a recessive marker gene is fully linked to D60, the F 2 segregation ratio of wild type to recessive marker gene homozygotes will be 5:4 (Supplementary File 1). This was closer to the expected ratio of 5:4 for the cases where d30 is fully linked to D60. Then, F 3 lines (50 individuals/lines) from 56 gh2 homozygous F 2 individuals from the cross between FL212 (d30gh2) and the Koshihikari d60 line were developed, and the genotype of the F 2 was subsequently determined.
Genes 2019, 10, x FOR PEER REVIEW 3 of 12 the d60Gal line shows dwarfness, the F1 with the D60gal line shows partial sterility, and the F1 with the D60Gal line shows fertility. Each of the F1 plants were scored with heading time and culm length in the field. The length between the ground surface and the panicle base of the main culm was measured as the culm length for all plants. The time when the tip of the panicle first emerged from the flag leaf sheath was recorded as the heading time for all plants. Three panicles were harvested from each F1 plant, and the number of filled and unfilled spikelets was counted for each panicle. The percent of seed fertility was calculated as the number of filled spikelets divided by the total number of spikelets multiplied by 100. The genotype was determined by seed fertility. Aceto-carmine squash mounts to stain the pollens of several F1 lines were conducted for Olympus BX40 microscopic examination.

Linkage Analysis for d60
Firstly, 318 F2 plants of a marker gene line FL212 [7] that has d30 and gh2 on chromosome 2(D60D60galgal) and the d60 line (d60d60GalGal) was used for segregation analysis for the marker genes and d60. The result showed that the segregation ratio of wild type to d30 homozygote at the d30 locus was 195:123, and wild type to gh2 homozygote at the gh2 locus was 218:100, and that both deviated significantly from 3:1 (Figure 1). When a recessive marker gene is fully linked to D60, the F2 segregation ratio of wild type to recessive marker gene homozygotes will be 5:4 (Supplementary File 1). This was closer to the expected ratio of 5:4 for the cases where d30 is fully linked to D60. Then, F3 lines (50 individuals/lines) from 56 gh2 homozygous F2 individuals from the cross between FL212 (d30gh2) and the Koshihikari d60 line were developed, and the genotype of the F2 was subsequently determined. According to the dwarf trait, we visually discriminated the d30 homozygote and wild type. As a result, the segregation ratio of wild type to d30 homozygote at the d30 locus was 195:123. In the correlation diagram with culm length and days to heading, red plots mean d30 homozygous plants, whereas vacant plots mean wild type plants. (B) Genotyping for the Gh2/gh2 locus. gh2 homozygous plants showed characteristic goldcoloring of unhulled grain. gh2 mutant is a lignin-deficient mutant, and Gh2 encodes a cinnamyl- Figure 1. Excessive segregation of recessive homozygotes, according to the ratio of 5:4, considerably deviated from the Mendelian 3:1 ratio in the F 2 between the recessive marker gene line FL212 and the d60Gal line (d60d60GalGal). (A) Genotyping for the D30/d30 locus. d30 homozygous plants showed characteristic dwarf phenotypes with short panicles and small grains. According to the dwarf trait, we visually discriminated the d30 homozygote and wild type. As a result, the segregation ratio of wild type to d30 homozygote at the d30 locus was 195:123. In the correlation diagram with culm length and days to heading, red plots mean d30 homozygous plants, whereas vacant plots mean wild type plants.
(B) Genotyping for the Gh2/gh2 locus. gh2 homozygous plants showed characteristic gold-coloring of unhulled grain. gh2 mutant is a lignin-deficient mutant, and Gh2 encodes a cinnamyl-alcohol dehydrogenase [8] According to the gold color of the matured hull, we visually discriminated the gh2 homozygote and wild type. As a result, the segregation ratio of wild type to gh2 homozygote at the gh2 locus was 218:100. In the diagram, green plots showed gh2 homozygous plants, whereas vacant plots mean wild type plants. Above all, the recessive morphological gene d30 and gh2 on the chromosome 2, segregated in the characteristic ratio of 5 wild type:4 recessive homozygotes, suggesting their linkage with D60 locus.

Universal Distribution of Gal and D60 Except for the d60 Donor Hokuriku100
All F 1 lines when crossed with the d60Gal line showed tallness and partial sterility, being reduced by an average of 25% from those with the D60gal line (Koshihikari) and the D60Gal line (Table 1, Figure 2). Regarding the dwarf varieties with d1, d2, d6, d18, d29, d30, d35, and d50, F 1 lines with both of the D60gal line and the D60Gal line showed normal seed fertility over 90%, and there was statistically no significant difference between them. On the other hand, F 1 lines with the d60gal line showed partial seed sterilities in the lower 70% level, which were reduced by approximately 25% from the F 1 s with the other two testers, namely the D60gal line and the D60Gal line, and the differences were statistically significant (5% level). Regarding to culm length, each of the F 1 lines between the dwarf variety and the three testers showed almost the same normal length and there was statistically no significant differences between them. The above observations revealed that the all the dwarf varieties had gal, because the seed fertilities of F 1 s with the d60Gal lines were significantly reduced by 25% in accordance to the frequency of the genotype d60gal gametes. Furthermore, all the dwarf varieties did not have d60, because the seed fertilities of F 1 s with D60gal lines were at the normal 90% level, and culm length of the F 1 s with the three testers showed statistically the same level. Therefore, the genotype of the representative dwarf varieties for D60Gal loci were determined as D60gal homozygous.
These data gave the following facts. The genotype of the 33 varieties is D60D60galgal, and the d60 locus is not allelic to those of sd1, d1, d2, d6, d18, d29, d30, d35, d49, d50, qCL1, and unknown genes involved in the 33 varieties. In addition, the gal gene does not cause complementarily gamete lethality together with the semidwarf and dwarf genes other than the d60 described above. Based on the above facts it is suggested that the gal gene will likely be distributed universally in rice. Therefore, it is emphasized that the Gal is rare, and is a valuable mutant gene essential to the transmission of d60. .78 ** 0.14 * and **: Significant at 5% and 1% levels, respectively. All F 1 lines when crossed with the d60Gal line showed tallness and partial sterility, being reduced by an average of 25% from those with the D60gal line (Koshihikari) and the D60Gal line. These data gave the following facts. The genotype of the 33 varieties is D60D60galgal and the d60 locus is not allelic to those of sd1, d1, d2, d6, d18, d29, d30, d35, d49, d50, qCL1, and unknown genes involved in the 33 varieties. In addition, the gal gene does not cause complementarily gamete lethality together with the semidwarf and dwarf genes other than the d60 described above.

Double Dwarfness with d30 and d60 Broken by Their Repulsion Linkage on Chromosome 2
Each F3 line (50 individuals/line) was developed from 56 gh2 homozygous F2 plants in the cross between d30gh2 line and the Koshihikari d60 line, and determined F2s' genotypes (Table 2). First, 32 lines of gh2d30 homozygous F2 plants were classified into three genotypes (Figure 3). Thirty lines were homozygous of non-recombinant gametes d30-D60, because the F3 progenies were fixed in the d30 homozygous dwarf phenotype. The single line has the recombinant gametes d30-d60 and the nonrecombinant gametes d30-D60 in the heterozygous plant, because d30d60 double recessive phenotypes appeared with approximately one fourth of the whole, namely, indicating a 3:1 segregation at the d60 locus in the d30 homozygous background. Only one single line was a d30d60 double recessive dwarf, having the recombinant gametes d30-d60 in the homozygous plant, due to its apparently shorter phenotype than the d30 homozygous plant. (Figures 3 and 4).

Double Dwarfness with d30 and d60 Broken by Their Repulsion Linkage on Chromosome 2
Each F 3 line (50 individuals/line) was developed from 56 gh2 homozygous F 2 plants in the cross between d30gh2 line and the Koshihikari d60 line, and determined F 2 s' genotypes (Table 2). First, 32 lines of gh2d30 homozygous F 2 plants were classified into three genotypes (Figure 4). Thirty lines were homozygous of non-recombinant gametes d30-D60, because the F 3 progenies were fixed in the d30 homozygous dwarf phenotype. The single line has the recombinant gametes d30-d60 and the non-recombinant gametes d30-D60 in the heterozygous plant, because d30d60 double recessive phenotypes appeared with approximately one fourth of the whole, namely, indicating a 3:1 segregation at the d60 locus in the d30 homozygous background. Only one single line was a d30d60 double recessive dwarf, having the recombinant gametes d30-d60 in the homozygous plant, due to its apparently shorter phenotype than the d30 homozygous plant (Figures 4 and 5).
Secondly, twenty lines having homozygous gh2 and heterozygous D30d30 were classified into three genotypes (Figure 3). Nine lines were heterozygous of the non-recombinant gametes D60-d30, d60-D30 and also for heterozygous Galgal, because these lines exhibited an excess segregation of the non-Mendelian 5:4 ratio at the d30 locus together with partial sterility, which is the same as in F 2 (214:143, χ 2 = 2.786, 0.05 ≤ p ≤ 0.10). On the other hand, 10 lines were heterozygous for the non-recombinant gametes D60-d30, d60-D30, and homozygous for Gal, because these lines segregated at the d30 locus in the Mendelian 3:1 ratio (301:87, χ 2 = 1.375, 0.10 ≤ p ≤ 0.90). The single line has the recombinant gametes D60-D30 and the non-recombinant gametes D60-d30 in heterozygous, because the line segregated in a ratio of 3:1 at the d30 locus. Three lines were non-recombinant d60-D30 homozygous and one line was heterozygous for recombinant gametes d60-D30 and non-recombinant D60-D30 gametes, and also for heterozygous Galgal (Table 2). These results indicated that d60 is linked to d30 with the recombination value calculated as 3.57% (= 4 recombinant gametes/112 total gametes × 100) on chromosome 2.

Genotype of F 1 Gamete and F 2 genotype : gh2gh2D30d30D60d60Galgal
Genotype of F 1 Gamete and F 2 genotype : gh2gh2D30d30D60d60GalGal Genotype of F 1 Gamete and F 2 (gh2gh2D30d30D60D60GalGal)

Discussion
The threat of strong typhoons due to global warming is increasing [10]. This is a serious problem in rice production, because strong winds cause stem lodging and consequent yield losses and deterioration in crop quality [11]. Extensive damage from the lodging of rice due to frequent typhoons has become a social problem in recent years, and developing new varieties of typhoon-resistant rice by introducing dwarf genes is an imperative task. Hence, there is a pressing need to develop new short-culm rice cultivars resistant to strong winds [12]. So far, sd1 is the world's only short-culm gene source in practical rice breeding. However, in the consideration of maintaining/expanding the genetic diversity of varieties, one should not rely only on sd1, which is a GA biosynthesis enzyme-defective gene, and should develop more new dwarf genes and promote their use in lodging-resistant breeding.
The excellent semidwarf quality of the rice mutant Hokuriku 100 is controlled by the single semidwarf gene d60 [3,6]. It is desirable to generate lodging-resistant rice cultivars that carry a novel short-culm gene, d60, as an alternative to sd1. However, d60 causes complementally gamete sterility, together with the gametic lethal gene gal. F 2 progenies between d60Gal line (Hokuriku 100) and the original tall variety D60gal Line (Koshihikari) segregate distortedly into 1 semidwarf (d60d60GalGal):8 tall (2D60d60Galgal:2D60d60GalGal:4D60D60) ratio, because of the deterioration of the F 1 male-and female-gametes having both gal and d60 [6] (Supplementary File 1). In this study, the author developed F 1 lines between 33 representative dwarf or semidwarf lines and three isogenic tester lines, the d60Gal line instead of Hokuriku 100 [13], the D60Gal line, and the D60gal line. Three tester lines were isogenic lines in the genetic background of Koshishikari, which were different in only a single allele for the D60/d60 and Gal/gal loci, namely d60Gal, D60gal, and D60Gal. Therefore, when the test subject has gal, F 1 with the d60Gal line shows partial sterility, and both the F 1 with the D60Gal line and the D60gal line show fertility. On the other hand, when the test subject has d60, F 1 with the d60Gal line shows as a semidwarf, F 1 with the D60gal line shows partial sterility, and F 1 with the D60Gal line shows fertility. As a result, all F 1 lines with the d60Gal line showed tallness and partial sterility lower than a 70% level, whereas the F 1 lines with the D60gal line (Koshihikari) and the D60Gal line showed a 90% level. In conclusion, the genotype of the 33 varieties was determined as D60D60galgal, and d60 was different from sd1, d1, d2, d6, d18, d29, d30, d35, d49, d50, qCL1, and unknown genes involved in the 33 varieties. Moreover, there were no dwarf or semidwarf genes which were complementary with gal, except for d60.
The findings above suggest that the gal gene will likely be distributed universally in rice. Therefore, d60 is a dwarf gene that could not have been obtained by chance without Gal's simultaneous mutation. The d60 gene could not have been transmitted without Gal. This means that the Gal gene is absolutely necessary to transmit d60, and d60 is very unique in that it always makes a pair with Gal and segregates according to an 8:1 ratio. In other words, d60 is a valuable gene because, without the gal to Gal mutation, d60 would not exist in a normal environment.
The d35 gene of Tanginbouzu, which became the best rice breed in Japan between 1955 and 1964, was kaurenoic acid oxidase-or 3-β hydroxylase-defective in the same GA biosynthesis pathway [14]. The Daikoku type dwarf gene d1 in rice is defective in the α subunit of the heterotrimeric G protein, affecting GA signal transduction [15]. Both genes did not show complementary effects between d60 and gal.
A progeny test was conducted in the F 3 of the cross between the Koshihikari d60 line and a line carrying a gene marker d30 on chromosome 2, which when segregated in a ratio of wild-type to d30 homozygote was 200:118, close to the theoretical segregation ratio of 5:4 at the d30 locus when completely linked to the D60 locus. This resulted in the genetic linkage between d30 and d60 loci on chromosome 2.
Here, we discuss the relationship between the complementary gamete sterility caused by gal, d60, and the previously reported hybrid gamete sterile genes in rice. Firstly, Oka [16] proposed that the duplicate S gene loci, which work as developmental factors in gametes, cause hybrid sterility when the F 1 gametes receive both recessive S genes on each duplicate locus. For example, if parents A and B have genotypes s1/s1 +2/+2 and +1/+1s2/s2, respectively, in which at least one + gene is necessary for normal development of the gamete, then 25% of their F 1 hybrids will be sterile. This is because those gametes carrying the double recessive combination s1s2 deteriorate due to deficiencies during gamete development. This hybrid sterility is similar to that caused by gal and d60 in that two genes are responsible for both systems. However, gal and d60 cause both sex sterilities, whereas Oka [17] suggests that the duplicate S gene model can only explain male gamete sterility.
On the other hand, Kitamura [18] explained female sterility in indica/japonica hybrids by the one locus sporo-gametophytic interaction hypothesis, that is, disharmony between one allele in the gamete and another in the surrounding sporophytic tissues. This model assumes parent genotypes of S/S and S a /S a creating the hybrid S/S a , in which the allele S present in the maternal tissue induces abortion of gametes carrying the opposite allele, Sa. Thus, 50% of S/Sa plants are sterile and produce gametes carrying the S allele only; selfed progenies are all fertile. Ikehashi et al. [19][20][21][22] showed that this one locus model was a more likely explanation for indica/japonica hybrid sterility than the two loci model [16]. The allelic interaction model [22] has been accepted as the genetic basis of hybrid sterility and the allelic interaction S 5 locus has been cloned [23].
In subsequent studies based on analyses of the fertility of a number of indica × japonica hybrids, over 30 female gametes sterility loci-including major genes-were identified and mapped [24][25][26][27][28][29][30][31][32][33], or male gametes sterility were identified [31]. So far, indica/japonica hybrid sterility loci were identified on chromosomes 4, 6, 7, 12, and 1 that lead to female gamete abortion through allelic interactions: S 7 [24], S 8 [25], S 9 and S 15 [27], and S 16 [26], etc. Among them, the Sa locus has been successfully cloned [34]. One-locus allelic interactions for male sterility were also recognized in hybrids between two cultivated rice species Oryza sativa and Oryza glaberrima Steud. [35][36][37], O. sativa, and Oryza rufipogon [38], and O. sativa and Oryza. glumaepatula [39], and a series of S 1 [37,40], S 18 [40], S 20 and S 21 [35,36], S 22A and S 22B [39], were identified. Above all, hybrid sterilities in rice can be explained by a single locus allelic interaction. Therefore, hybrid sterility caused by the two genes d60 and gal is an extremely rare case in rice. Moreover, gamete breakdowns of both sexes, for gal and d60, are particularly rare and its ubiquities distribution is quite novel discovery. Funding: This work is funded by the Adaptable and Seamless Technology Transfer Program (A-STEP) through Target-driven R&D (high-risk challenge type) by the Japan Science and Technology Agency (JST) to Motonori Tomita, whose project ID14529973 was entitled "NGS genome-wide analysis-based development of rice cultivars with super high-yield, large-grains, and early/late flowering, suitable for the globalized world and global warming" since 2014 to 2018.