Hd1 Allele Types and Their Associations with Major Agronomic Traits in Korean Rice Cultivars

Optimizing flowering time in crop plants is critical for maximizing yield and quality under target environments. While there is a wide range of heading date variation in Korean rice cultivars, the underlying gene mechanisms are unclear. Here, we sequenced the protein coding regions of Hd1, the major rice heading date gene, from 293 Korean rice cultivars and investigated the associations between Hd1 allele types and major agronomic traits under four different environments. There were four functional Hd1 and five nonfunctional hd1 alleles distributed among the 293 Korean rice cultivars. The effects of the Hd1 allele types were highly significant for days to heading in all four environments, explaining 51.4–65.8% of the phenotypic variation. On average, cultivars carrying nonfunctional hd1 headed 13.7 days earlier than those carrying functional Hd1. While the Hd1 allele types exhibited highly significant effects on culm length and protein content under all four environments, the differences between cultivars carrying Hd1 and hd1 were minimal. The effects of the Hd1 allele types on amylose content were significant in only one of the four environments. Our results provide useful information for fine-tuning rice heading dates by utilizing different Hd1 alleles in rice breeding programs.


Introduction
The transition from the vegetative to reproductive phase is a critical developmental event in plants. To ensure offspring survival under favorable environments, the onset of reproductive development is regulated by intricate genetic mechanisms sensing various endogenous cues and environmental conditions such as daylength and temperature [1,2]. In agriculture, flowering time control is important as it determines the regional adaptability of different crop species and affects a range of agronomic traits relevant to crop yield and quality [3]. In rice, the vegetative-to-reproductive transition is induced under short day condition when the photoperiod becomes shorter than the critical daylength [4]. As photoperiod sensitivity shows wide variation among different rice genotypes, the use of germplasm with varying degrees of photoperiod sensitivity enabled the expansion of rice cultivation from the tropics to high latitude regions [5][6][7][8][9].
The main genetic factors involved in rice flowering time (aka heading date) control are the Hd3a and RFT1 genes, the orthologs of Arabidopsis Flowering Locus T (FT) [10,11]. These genes are upregulated under short day conditions and produce florigen, the mobile flowering signal that moves from leaves to the shoot apical meristem through the phloem, and induces floral development. The expression level of Hd3a shows a strong correlation with the rice heading date, with the rice genotypes exhibiting higher Hd3a expression in the leaves heading earlier [12]. While the amino acid sequences of Hd3a are highly conserved ). Compared with the other traits, the AC in each environment exhibited a greater number of outliers due to the inclusion of waxy rice cultivars (e.g., cultivars with "chal", meaning "glutinous", in their names in Table 1) with the AC range of 4.5-7.1% and high amylose cultivars (e.g., cultivars with "goami", meaning "high amylose", in their names in Table 1) with the AC range of 25.1-41.6%. Similar to DTH (0.94 < r < 0.98), the AC across four different environments showed very strong positive correlations (0.97 < r < 0.98) (Figure 2), indicating that both DTH and AC are largely determined by genetic factors.
Correlations between different traits: The DTH-CL correlation was significant and weakly positive in all four environments (0.25 < r < 0.35) ( Figure 2). The DTH-AC correlation was significant in three of the four environments and was weakly positive (0.18 < r < 0.23). The DTH-PC correlation was significant and negative (−0.64 < r < −0.25) in three of the four environments. There was no significant correlation between CL and AC in all four environments. Significant but weak correlations were observed between CL and PC (0.24 < r < 0.35) in two environments, and between AC and PC (r = −0.18) in one environment.

Hd1 Allele Types of 293 Korean Rice Cultivars
To study the allelic distributions of Hd1 in 293 Korean rice cultivars, we sequenced the protein coding regions of Hd1 and determined the allele type of each cultivar (Figure 3). This revealed nine different previously reported Hd1 alleles distributed in the 293 Korean rice cultivars. According to previous reports (Figure 3), four alleles (i.e., GBZ, Type 1, Type 6, and Type 11) were classified as functional Hd1 alleles, and five alleles (i.e., se1, T65, Type 7, Type 14, and AK) were classified as nonfunctional hd1 alleles. Out of the 293 Ko-  Figure 1c). Compared with the other traits, the AC in each environment exhibited a greater number of outliers due to the inclusion of waxy rice cultivars (e.g., cultivars with "chal", meaning "glutinous", in their names in Table 1) with the AC range of 4.5-7.1% and high amylose cultivars (e.g., cultivars with "goami", meaning "high amylose", in their names in Table 1) with the AC range of 25.1-41.6%. Similar to DTH (0.94 < r < 0.98), the AC across four different environments showed very strong positive correlations (0.97 < r < 0.98) (Figure 2), indicating that both DTH and AC are largely determined by genetic factors.
Correlations between different traits: The DTH-CL correlation was significant and weakly positive in all four environments (0.25 < r < 0.35) (Figure 2). The DTH-AC correlation was significant in three of the four environments and was weakly positive (0.18 < r < 0.23). The DTH-PC correlation was significant and negative (−0.64 < r < −0.25) in three of the four environments. There was no significant correlation between CL and AC in all four environments. Significant but weak correlations were observed between CL and PC (0.24 < r < 0.35) in two environments, and between AC and PC (r = −0.18) in one environment.
Among the five nonfunctional hd1 allele types distributed in Korean rice cultivars, Type 14, due to the 2-bp deletion in exon 2, was the most common, carried by 52 japonica and 4 Tongil-type cultivars ( Figure 3 and Table 1) [12]. The second most common hd1 allele was Type 7 due to the 4-bp deletion in exon 2 [12], carried by three japonica and 28 Tongil-type cultivars. The se1 type hd1 due to the 43-bp deletion in exon 1, which was initially identified in the gamma irradiation mutant HS66 [16], was carried by 20 japonica cultivars and 1 Tongil-type cultivar, Mogyang. The T65 type hd1, due to the 1901 bp retrotransposon insertion in exon 2, which was initially identified in the japonica cultivar Taichung 65 [22], was carried by two Korean japonica cultivars, Jungmo1043 and Naepungbyeo 2. The AK type hd1 due to the 312-bp insertion in exon 1 was reported in the Japanese rice cultivar Akage (AB300058.1) [8], and the same 312-bp insertion was also reported as the nonfunctional hap11 and hd1-3 [13,23]. Only one Korean cultivar, Jinbuolbyeo, carried the AK type hd1.

Association of Hd1 Allele Types with Major Agronomic Traits
To study the effects of Hd1 allele types on DTH, CL, AC, and PC, one-way ANOVAs were conducted for each trait under four different environments, with the nine Hd1 allele types as a single factor ( Table 2). The effect of Hd1 allele types was highly significant (p < 0.0001) for DTH, CL, and PC in all four environments, and explained 51.4-65.8% of the DTH variation, 13.2-24.2% of the CL variation, and 17.0-53.0% of the PC variation. However, the effect of Hd1 allele types on AC was significant (p < 0.05) in only one (Suwon 2018) of the four environments, explaining 5.5% of the phenotypic variation, indicating the small effect of Hd1 on AC. This observation was consistent with the weak correlations (0.18 < r < 0.23) between DTH and AC (Figure 2), and is likely because AC is mainly determined by the starch-synthesis related genes such as GBSSI and SSIIa in rice [24]. 68.2 **** 65.8 6.9 **** 16.2 1.4 NS -33.9 **** 42.2 z Four environments composed of two locations, Wanju (WJ) and Suwon (SW), and two consecutive years, 2018 and 2019. One-way ANOVAs were conducted in each environment for each trait, with nine Hd1 allele types as a single factor. DTH-days to heading; CL-culm length; AC-amylose content; PC-protein content; PVE-phenotypic variance explained by the Hd1 allele types. **** p < 0.0001, * p < 0.05, NS not significant.
For mean comparisons of DTH, CL, and PC, which showed significant differences according to Hd1 allele types in all four environments, additional ANOVAs were conducted with environments as blocks (a random variable) and Hd1 allele types as a fixed variable (Table 3). On average, cultivars carrying nonfunctional hd1 alleles headed 71.8 days after transplanting, which was 13.7 days earlier (p < 0.0001) than those carrying functional Hd1 alleles that headed 85.5 days after transplanting. There were significant DTH differences within the Hd1 and hd1 allele groups as well, e.g., Type 6 (84.5 days) exhibiting significantly earlier DTH than GBZ (86.9 days) in the functional Hd1 group, and Type 14 (68.9 days) exhibiting significantly earlier DTH than Type 7 (80.6 days) in the nonfunctional hd1 group. The difference in CL between cultivars carrying Hd1 (76.3 cm) and hd1 (75.3 cm) was significant (p < 0.05) but small (Table 3). Among the nine Hd1 allele types, remarkable differences in CL were observed between Type 11 Hd1 (101.1 cm) and the AK type hd1 (58.6 cm), while the CL range of the other seven allele types was 72.0-79.3 cm. Similar to CL, a significant (p < 0.0001) but small difference was observed in PC between cultivars carrying Hd1 (6.1%) and hd1 (6.7%) ( Table 3). Among the nine Hd1 allele types, Type 7 hd1 (6.8%) and the AK type hd1 (7.1%) exhibited a significantly higher PC than the GBZ type Hd1 (6.0%), while there was no significant difference in PC among the rest of the allele types.

Discussion
Heading date optimization is an important goal in rice breeding for enhancing land use efficiency by enabling diverse double cropping systems between rice and other crops, as well as ensuring stable yield and grain quality under target environments. Developing early heading rice cultivars is especially important in high latitude and mountainous areas where early heading characteristics are required for rice plants to complete grain filling before the risk of cold winter [3]. Early heading rice cultivars are also useful for avoiding extreme whether events due to climate change and mitigating methane emissions from rice paddies by minimizing the duration of rice cultivation [25,26]. To exploit such advantages, tremendous efforts have been made in Korean rice breeding programs to develop a number of early heading rice cultivars [19,20]. However, the gene mechanisms underlying heading date variation of Korean rice cultivars are unclear. In this study, we determined the allele types of Hd1, the major rice heading date gene, in 293 Korean rice cultivars, and investigated their associations with major agronomic traits to provide useful information for rice breeding programs.
A total of nine Hd1 allele types, four functional and five nonfunctional alleles, were distributed in 293 Korean rice cultivars (258 japonica and 35 Tongil-type derived from indica-japonica crosses). The two most frequent functional Hd1 alleles, Type 6 (n = 96) and the GBZ type (n = 83), were carried only by japonica cultivars (Table 1). These two alleles were present only in temperate japonica accessions among diverse rice genotypes, as well as those evaluated at IRRI [15], indicating that they are japonica specific alleles. The other two functional allele types, Type 1 and Type 11, were rare in Korean rice cultivars, carried by only one (Mimyeon) and two (Sangnambatbyeo and Nokwoo), respectively.
Among the five nonfunctional hd1 allele types, the most frequent was Type 14, due to the 2 bp deletion in exon 2, which was carried by 52 japonica and four Tongil-type cultivars. Type 14 hd1 was initially reported in the indica cultivar Kasalath [16], and five different alleles (Types 13, 14, 15, 16, and17) with the same 2 bp deletion were subsequently identified and classified as nonfunctional hd1 alleles [12]. This mutation has been identified over 700 rice accessions belonging to different subgroups encompassing japonica, indica, aus, intermediate, and landraces among IRRI's 3 K rice accessions [18], indicating that it has been present before the indica-japonica divergence and is widely utilized to expand the adaptability of different rice subgroups. The second most frequent hd1 allele was Type 7, due to the 4 bp deletion in exon 2, which was carried by 28 Tongil-type and three japonica cultivars. This mutation was also frequent among IRRI's 3 K rice accessions, carried by over 400 accessions, of which 99% were indica [18], indicating that it is likely to have occurred after the indica-japonica divergence and has been popularly utilized in developing Tongil-type cultivars in Korea. The se1 type hd1 due to the 43 bp deletion in exon 1 has been widely utilized in rice breeding in Japan, China, and Europe [6,8,17], and also for developing early heading Korean rice cultivars, as seen in 20 japonica and 1 Tongil-type cultivars in this study. The T65 type hd1 due to the 1.9 kb retrotransposon insertion in exon 2 was initially reported in the japonica cultivar Taichung65, and was subsequently identified in rice accessions from China, Cambodia, and Nepal [18,22]. Only two Korean cultivars, Jungmo1043 and Naepungbyeo, carried the T65 type hd1. The AK type hd1 due to the 312 bp insertion in exon 1 is frequently found in rice cultivars developed in the Hokkaido (41 • 2 -45 • 3 N latitude) region of Japan, which is considered as one of the northern limits of rice cultivation [8]. This mutation was carried by only one Korean cultivar, Jinbuolbyeo, which showed the earliest heading date among 293 Korean rice cultivars investigated in this study. Further work is required to determine if the earliness of Jinbuolbyeo is mainly due to the AK type hd1 or if there are other genetic factors affecting the early heading characteristics of this cultivar.
Nine Hd1 allele types explained 51.4-65.8% of the DTH variation of 293 Korean rice cultivars under four different environments (Table 2), demonstrating the substantial role of Hd1 as a single gene determining the heading dates of Korean rice cultivars. On average, 111 cultivars carrying nonfunctional hd1 headed 14 days earlier than 182 cultivars carrying functional Hd1. While many previous studies evaluated the effects of different Hd1 alleles similarly by classifying them into two categories, functional Hd1 and nonfunctional hd1 [12,13,15], our study showed that large DTH variation exists also within each category (Table 3), demonstrating the need for evaluating the effects of each individual allele precisely. Studies using near isogenic lines (NILs) carrying different Hd1 alleles showed that missense mutations in exon 1 (i.e., N165K and R205Q) induce functional differences in the Hd1 protein and indicated that Hd1 sequence variation other than loss-of-function mutations can also affect rice heading dates [7]. Therefore, for fine-tuning heading dates in rice molecular breeding programs, it is important to characterize the function of different alleles within both Hd1 and hd1 categories under an isogenic background to precisely evaluate their effects on DTH. In addition, as genes other than Hd1 also affect rice heading dates, cataloging the allele types of other major rice heading date genes such as Ghd7, DTH8, and Hd3a in different rice cultivars, and evaluating their effects on DTH would facilitate rice breeding programs aimed at precisely controlling heading dates under target environments.
The heading date is known to be associated with a range of agronomic traits affecting yield, such as plant height, panicle number, grain number, and weight, and those affecting eating and cooking quality, such as AC and PC [3,[27][28][29][30]. Therefore, to facilitate the use of different Hd1 alleles in rice molecular breeding, it is important to evaluate the pleiotropic effects of Hd1 and the effects of different Hd1 alleles on major agronomic traits. Our study using 293 Korean rice cultivars showed that nine Hd1 allele types explain 13.2-24.2% of CL variation and 17.0-53.0% of PC variation under four different environments, while explaining only 5.5% of AC variation under only one environment ( Table 2). In addition, the differences in CL and PC between the cultivars carrying functional Hd1 and nonfunctional hd1 were significant but small (Table 3). Our results indicate that it would be possible to use different Hd1 alleles to modulate DTH, while affecting CL, AC, and PC to a mild extent. The replacement of the GBZ type (aka Haplotype 8) Hd1 carried by the japonica rice cultivar Chunjiang06 with Type 7 (aka Haplotype 16) resulted in a yield improvement without affecting the grain quality traits, such as gel consistency and gelatinization temperature [17], supporting that the use of different Hd1 alleles can optimize DTH effectively without negatively influencing other agronomic traits. As rice cultivars with different genetic background used in this study provide a limited capability for precisely evaluating the effects of different Hd1 alleles on agronomic traits, further studies using isogenic lines would be necessary to generate more precise information that can be utilized in rice breeding programs for selecting optimal Hd1 alleles, meeting different breeding objectives.
In conclusion, we cataloged the allele types of Hd1 distributed in Korean elite rice cultivars and evaluated their associations with DTH, CL, AC, and PC under four different field environments. While the Hd1 allele types explained a large proportion (51.4-65.8%) of DTH variation, their effects on CL, AC, and PC were limited. Our study and further work to characterize the precise allelic effects of Hd1 and other heading date genes will advance rice molecular breeding tools for fine-tuning DTH and other important agronomic traits.

Plant Materials and Phenotyping
A total of 293 Korean rice cultivars released by the National Institute of Crop Science (NICS), of the Rural Development Administration (RDA) of South Korea between 1979 and 2017 were used in this study. The list of the cultivars initially described in Lee et al. [20] contains 297 cultivars, but we excluded four with potential seed contamination in this study and used only 293 cultivars.
The rice plants were cultivated at the experimental field stations of NICS located in Wanju (35 •  15 cm, and the rows were spaced by 30 cm. The plants were grown and managed according to the standard rice cultivation methods of NICS, RDA [31]. Days to heading (DTH) were determined by counting the days from transplanting to heading, when 40% of the plants of each cultivar exhibited emerged panicles. Culm length (CL) was determined by measuring the length from the ground to the panicle node of the main culm from 10 random plants, and averaging them to represent each cultivar. Upon maturity, the rice grains were harvested, dried at 15% grain moisture content naturally under shaded condition, dehulled with a roller husking machine (SY88-TH; Ssangyong Ltd., Incheon, Korea), and polished with a laboratory polishing machine (MC-90A; Toyo Co., Wakayama, Japan) as previously described [32]. The amylose content (AC) was evaluated according to Juliano [33] using the iodine colorimetric method with a UV/visible spectrophotometer (Evolution 600; Thermo Fisher Scientific, Waltham, MA, USA). Protein content (PC) was evaluated using the Micro Kjeldahl method according to AOAC [34] with automated Kjeldahl analyzers (Foss Digester 2020 and Foss Kjeltec 2400; Foss Tecator, Huddinge, Sweden).

Hd1 Genotyping
Leaf tissues were collected from 3-week-old rice seedlings and genomic DNA was extracted using the CTAB method [35]. To determine the Hd1 allele types of the 293 Korean rice cultivars by sequencing the protein coding regions of Hd1, each of the two exons of Hd1 was amplified by a pair of primers (Hd1_exon1_F and Hd1_exon1_R for the first exon, and Hd1_P4_F and Hd1_P4_R for the second exon) described in Table 4. For the first exon, three additional primers (Hd1_P1_R, Hd1_exon1_R1, and Hd1_P2_F) were used as the inner primers for sequencing. For sequencing the 312 bp and 1901 bp insertions identified in the first exon of several cultivars, two (Hd1_P1_F and Hd1_P2_R) and four (Hd1_exon1_F1, Hd1_exon1_F2, Hd1_exon1_R2, and Hd1_exon1_R3) additional primers were used as the inner primers. For the second exon, the primers used for PCR were also used for sequencing. The Sanger sequencing reactions were conducted using the Capillary Electrophoresis Sequencing (CES) service at Macrogen, South Korea (https://dna.macrogen.com, accessed on 13 October 2021).

Data Availability Statement:
The data generated in this study are available from the corresponding author upon reasonable request.