Advertisement

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Impact of rice GENERAL REGULATORY FACTOR14h (GF14h) on low-temperature seed germination and its application to breeding

Abstract

Direct seeding is employed to circumvent the labor-intensive process of rice (Oryza sativa) transplantation, but this approach requires varieties with vigorous low-temperature germination (LTG) when sown in cold climates. To investigate the genetic basis of LTG, we identified the quantitative trait locus (QTL) qLTG11 from rice variety Arroz da Terra, which shows rapid seed germination at lower temperatures, using QTL-seq. We delineated the candidate region to a 52-kb interval containing GENERAL REGULATORY FACTOR14h (GF14h) gene, which is expressed during seed germination. The Arroz da Terra GF14h allele encodes functional GF14h, whereas Japanese rice variety Hitomebore harbors a 4-bp deletion in the coding region. Knocking out functional GF14h in a near-isogenic line (NIL) carrying the Arroz da Terra allele decreased LTG, whereas overexpressing functional GF14h in Hitomebore increased LTG, indicating that GF14h is the causal gene behind qLTG11. Analysis of numerous Japanese rice accessions revealed that the functional GF14h allele was lost from popular varieties during modern breeding. We generated a NIL in the Hitomebore background carrying a 172-kb genomic fragment from Arroz da Terra including GF14h. The NIL showed superior LTG compared to Hitomebore, with otherwise comparable agronomic traits. The functional GF14h allele from Arroz da Terra represents a valuable resource for direct seeding in cold regions.

Author summary

Rice serves as a fundamental crop sustaining over half of the global population. With the rapid growth of the world’s population, it will become increasingly important to improve rice productivity. On the other hand, the aging of rice farmers in Japan has resulted in a constant labor shortage. To address this, direct seeding, in which seeds are sown directly in rice fields without going through the most labor-intensive part of the rice cultivation process, i.e., seedling production and transplanting, has been recommended. However, prevalent elite rice varieties are known to be unsuitable for direct seeding due to their poor seed germination ability under low-temperature conditions. In this study, we show for the first time that GF14h gene from the Portuguese variety Arroz da Terra improves seed germination at low temperatures (LTG). In addition, a novel cross-bred line was generated by introducing the GF14h-containing genomic segment from Arroz da Terra into Hitomebore, a widely cultivated variety in northern Japan. This line is expected to be used as a pre-breeding material to enhance LTG. This study will provide a genetic basis for LTG and contribute to basic and applied research progress.

Citation: Sugimura Y, Oikawa K, Sugihara Y, Utsushi H, Kanzaki E, Ito K, et al. (2024) Impact of rice GENERAL REGULATORY FACTOR14h (GF14h) on low-temperature seed germination and its application to breeding. PLoS Genet 20(8): e1011369. https://doi.org/10.1371/journal.pgen.1011369

Editor: Ortrun Mittelsten Scheid, Gregor Mendel Institute of Molecular Plant Biology, AUSTRIA

Received: February 25, 2024; Accepted: July 12, 2024; Published: August 7, 2024

Copyright: © 2024 Sugimura et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The NGS raw reads were deposited in the DNA Data Bank of Japan (DDBJ) under the BioProject accession numbers PRJDB13449, PRJDB13450, PRJDB13864, and PRJDB17450. Supplementary Tables S2 and S5 include the SRA accession numbers. The genome sequences assembled in this study were deposited at Zenodo (https://doi.org/10.5281/zenodo.10460309). All relevant data are within the paper and its Supporting Information files.

Funding: This work was partly supported by a grant from the Ministry of Agriculture, Forestry, and Fisheries of Japan (Genomics-based Technology for Agricultural Improvement, IVG3007), received by HT, RT and AA. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Low-temperature seed germination (LTG) is a pivotal agronomic trait in rice (Oryza sativa). As rice originated from tropical and subtropical regions, it is highly susceptible to low-temperature conditions compared to other cereal crops such as wheat (Triticum aestivum) and barley (Hordeum vulgare) [1]. Nevertheless, rice is produced in temperate and high-altitude regions, where it frequently experiences temperatures below 20°C. In Japan, rice is abundantly cultivated in relatively cold areas such as Tohoku and Hokkaido. In recent years, there has been an increasing demand to shift from conventional transplantation-based rice cultivation to direct seeding to reduce labor and costs. However, direct seeding raises the risk of exposure to low temperatures during seed germination [2]. Therefore, to expand the use of direct seeding, it is crucial to breed rice cultivars with enhanced LTG.

LTG is a quantitative trait regulated by complex molecular mechanisms. Linkage mapping and genome-wide association studies (GWAS) have identified over 30 LTG-related quantitative trait loci (QTLs) or genomic regions associated with this trait, located on all 12 rice chromosomes [3–20]. However, only a few genes involved in LTG have been described, such as qLTG3-1 [3] and STRESS-ASSOCIATED PROTEIN16 (OsSAP16) [15]. The qLTG3-1 gene, encoding a protein of unknown function, has a substantial influence on LTG [3]. During seed germination, qLTG3-1 expression is strongly induced in embryos, which leads to the loosening of the tissues covering the embryo by promoting vacuolation [3]. OsSAP16 encodes a stress-associated protein with two AN1-C2H2 zinc finger domains [15]. OsSAP16 presumably acts as a regulator of LTG.

14-3-3 proteins are regulatory proteins that are widely conserved in eukaryotes. These proteins bind to phosphorylated serine and tyrosine residues in their target proteins that participate in signal transduction and the regulation of gene expression [21, 22], thus altering their enzymatic activity, subcellular localization, stability, or protein–protein interactions [23–25]. The rice genome encodes eight 14-3-3 proteins, named GF14a–h for GENERAL REGULATORY FACTOR14 [26]. GF14h is involved in rice seed germination under optimal temperature conditions [27,28]. In addition, GF14h contributes to phytohormone signaling, including abscisic acid and gibberellin signaling [27,28]. However, it remains unclear whether GF14h promotes seed germination under low-temperature conditions [28].

QTL pyramiding has been proposed as a breeding concept [29] for bringing together several QTLs (or genes) related to agronomically important traits in the genetic background of locally adapted elite cultivars. In practice, it is essential to generate pre-breeding materials for QTL pyramiding, i.e., near-isogenic lines (NILs) that harbor one or a few genomic segments introgressed from the donor parent into the genome of the recipient parent through a combination of continuous backcrossing and selfing via marker-assisted selection [30]. In this study, we determined that GF14h is responsible for an LTG-related QTL in Portuguese rice variety Arroz da Terra. We generated a NIL in the background of rice cultivar Hitomebore, which is adapted for growth in northern Japan, by replacing its GF14h genomic fragment with that from Arroz da Terra and tested its LTG performance.

Results

Evaluation of QTLs associated with low-temperature germination using the Portuguese rice variety Arroz da Terra

We investigated seed germination characteristics of a Portuguese rice variety Arroz da Terra and a Japanese varieties Iwatekko and Hitomebore. Under low-temperature conditions (15°C), Arroz da Terra exhibited significantly higher germination rates compared to Iwatekko and Hitomebore during days 7−15, particularly showing a 30−40% increase in germination rate 10−11 days after imbibition (Fig 1A and 1B). Similarly, at normal temperature conditions (25°C), Arroz da Terra showed superior germination rates 2−4 days after imbibition, especially with a 30% difference observed 3 days after imbibition (Fig 1A and 1C). These findings indicate that Arroz da Terra exhibits more vigorous germination under normal and low-temperature conditions than the Japanese cultivars Iwatekko and Hitomebore. To identify the genes responsible for this difference, we searched for QTLs involved in the high LTG of Arroz da Terra. We previously generated a set of 200 RILs at the F7 generation derived from a cross between Arroz da Terra and Iwatekko (S1 Fig) [31]. We phenotyped all RILs for LTG at 13°C and selected the 20 RILs with the highest LTG and the 20 RILs with the lowest LTG. We assembled two pools of seedlings with low or high LTG and extracted their genomic DNA for whole-genome sequencing on the Illumina platform (S1 Fig) [31]. We mapped the resulting sequencing reads to the Nipponbare rice reference genome (IRGSP-1.0) and performed QTL-seq analysis using our new high-performance pipeline [32]. Based on the Δ(SNP-index), we identified three QTLs related to LTG on chromosome 3 (qLTG3-1 and qLTG3-2) and chromosome 11 (qLTG11) (Fig 1D), which is consistent with the results of a previous study [31].

thumbnail
Download:
Fig 1. Effects of quantitative trait loci (QTLs) on low-temperature seed germinability.

(A) Representative photographs showing the germination of seeds from the Iwatekko, Hitomebore, and Arroz da Terra varieties 11 days (15°C) or 3 days (25°C) after the onset of seed imbibition. Scale bar, 1 cm. (B–C) Germination time courses of Iwatekko, Hitomebore, and Arroz da Terra at 15°C (B) or 25°C (C). Values are means ± standard deviation (SD) from biologically independent samples (n = 8). Dunnett’s test shows significant differences in germination for Hitomebore (upper) and Iwatekko (lower) compared with Arroz da Terra at each time point (*P < 0.05, **P < 0.01 and ***P < 0.001). (D) Map positions of QTLs for low-temperature germination, as determined by QTL-seq. The Δ (SNP-index) values (red lines) were plotted for chromosomes 3 and 11, with statistical confidence intervals under the null hypothesis of no QTL (green, P < 0.05; orange, P < 0.01). (E) Diagram showing the genotype of qLTG11-NIL. qLTG11-NIL harbors the Arroz da Terra allele at qLTG11 on chromosome 11. Light blue indicates genomic fragments from Hitomebore; red indicates genomic fragments from Arroz da Terra; dark blue indicates heterozygous regions. (F) Germination time courses of Hitomebore and qLTG11-NIL at 15°C. Values are means ± SD from biologically independent samples (n = 10). Two-tailed t-test was used between qLTG11-NIL and Hitomebore for each time point (*P < 0.05 and ***P < 0.001).

https://doi.org/10.1371/journal.pgen.1011369.g001

The qLTG3-1 region contained the gene Os03g0103300, which was reported to be involved in LTG in a study using rice cultivar Italica Livorno, which has high LTG, and Hayamasari, which has low LTG [3]. An examination of its coding sequences in Arroz da Terra, as well as Iwatekko and Hitomebore, revealed that they were identical to those found in Italica Livorno and Hayamasari, respectively (S2 Fig). While Italica Livorno harbored a functional haplotype for this gene, Hayamasari carried a loss-of-function haplotype due to a 71-bp deletion (S2 Fig) [3]. Therefore, we propose that the causal gene for the QTL LTG3-1 is Os03g0103300.

To evaluate the contribution of the two other QTLs to LTG, we generated NILs harboring a segment from the Arroz da Terra genome for each QTL (approximately 5 Mb) in the Hitomebore background (Figs 1E, S3A and S4A). We detected no clear effect of qLTG3-2 on LTG, as qLTG3-2-NIL and Hitomebore showed similar seed germination rates at 15°C (S4B Fig). By contrast, qLTG11-NIL showed a significantly higher rate of germination than Hitomebore 7−12 days after imbibition at 15°C, particularly after 8 and 9 days, showing a difference of more than 40% (Fig 1E and 1F), indicating that qLTG11 enhances LTG. qLTG11-NIL seeds also germinated more rapidly than Hitomebore seeds under normal conditions (25°C), although with a smaller difference between the two genotypes than at low temperature (S5 Fig). We therefore focused our analysis on qLTG11.

Identification of GF14h as the candidate gene for qLTG11

To delineate the qLTG11 region, we carried out map-based cloning using a segregating population derived from a cross between BC2F3 line qLTG11-NIL and Japanese elite cultivar Hitomebore (S3A Fig). For mapping, we conducted germination tests at 15°C. We narrowed down the genomic region containing the QTL to a 52-kb segment (from 23.512 Mp to 23.564 Mb) on chromosome 11 based on the Nipponbare reference genome (IRGSP-1.0) (Fig 2A). This interval contains two annotated genes based on the Nipponbare genome sequence (Fig 2B). We compared the genomic sequence of Hitomebore and Arroz da Terra across the candidate region using de novo genome assembly obtained from Nanopore long reads. The cultivars Hitomebore and Nipponbare had an identical genomic sequence over the entire candidate region (S6A Fig). By contrast, the genome sequence from Arroz da Terra was substantially different from that of Nipponbare, with the equivalent candidate region spanning approximately 94 kb (Figs 2B and S6B).

thumbnail
Download:
Fig 2. Positional cloning of qLTG11.

(A) Fine mapping of qLTG11 to a 52-kb region between markers E and J. The chromosomal positions are based on the Nipponbare reference genome (Os-Nipponbare-Reference-IRGSP-1.0). Germination percentage was determined at 9 days of incubation at 15°C. Red and blue rectangles indicate chromosomal segments homozygous for Arroz da Terra or Hitomebore, respectively. Different lowercase letters indicate significant differences (n = 3 biologically independent samples, P < 0.001, Tukey’s HSD test). (B) Genomic structure of the candidate genomic region in Arroz da Terra and Hitomebore. Os11g0609600 (shown in red), encoding GF14h, is expressed in germinating seeds. (C) Diagram of the GF14h gene structure and sequence polymorphisms between Arroz da Terra and Hitomebore. The chromosomal positions are based on the Nipponbare reference genome. The coding region of GF14h in Hitomebore is identical to that in Nipponbare. The 4-bp deletion in Hitomebore causes a frameshift and the introduction of a premature stop codon. (D) Relative GF14h expression levels in germinating seeds of qLTG11-NIL. This expression analysis was conducted by RT-qPCR. In the boxplots, the box edges represent the upper and lower quantiles, the horizontal line in the middle of the box represents the median value, whiskers represent the lowest quantile to the top quantile, and the black squares show the mean. Five biological replicates were measured independently. Different lowercase letters indicate significant differences based on Tukey’s HSD test (P < 0.05). OsActin1 (Os03g0718100) was used for normalization.

https://doi.org/10.1371/journal.pgen.1011369.g002

As the causal gene behind the variation in LTG is likely expressed in seeds, we performed transcriptome deep sequencing (RNA-seq) during seed germination in Hitomebore and qLTG11-NIL (S1 Table). Within the candidate region, the gene Os11g0609600, corresponding to the 14-3-3 gene GF14h, was expressed in both Hitomebore and qLTG11-NIL, whereas Os11g0609500 (Jacalin-like lectin domain containing protein) was not expressed in seeds (S7 Fig), thus suggesting that GF14h is a strong candidate gene for LTG. The GF14h gene structure and haplotypes in Arroz da Terra and Hitomebore are shown in Fig 2C. We detected a 4-bp deletion in the GF14h coding region in Hitomebore, causing a frameshift mutation predicted to introduce a premature stop codon (Figs 2C and S8). These results suggest that Hitomebore carries a loss-of-function allele of GF14h. To assess the role of GF14h in LTG, we examined the expression pattern of the putative functional GF14h (GF14hArroz) allele during seed germination at low temperature (15°C) using qLTG11-NIL. RT-qPCR analysis of GF14h expression levels showed that they were comparable in the embryo and endosperm at 1 and 3 days after the onset of seed imbibition (Fig 2D). At the beginning of germination, when a white coleoptile was visible (5 and 7 days after seed imbibition), GF14h expression levels rose in the endosperm, but not in the embryo (Fig 2D). By nine days of imbibition, when most seeds had germinated, GF14h expression in the endosperm returned to basal levels (Fig 2D). These results support the notion that GF14h plays a role in seed germination at low temperature.

GF14h plays a vital role in LTG

To investigate the contribution of GF14h to LTG, we knocked out the functional GF14h copy present in qLTG11-NIL by clustered regularly interspersed short palindromic repeat (CRISPR)/CRISPR-associated nuclease 9 (Cas9)-mediated gene editing and evaluated LTG. Specifically, we introduced two single guide RNA (sgRNA) constructs targeting the exons of GF14h individually into qLTG11-NIL by Agrobacterium-mediated transformation. We chose to knock out GF14h in the qLTG11-NIL background rather than Arroz da Terra to evaluate the specific contribution of GF14h to LTG without the influence of qLTG3-1, which would be present in the Arroz da Terra background. To accurately evaluate the phenotypes of the edited plants, we selected heterozygous plants in the T0 generation and isolated homozygous mutant lines and their unedited homozygous siblings in the T1 generation. We obtained four knockout lines (gf14h-1, gf14h-2, gf14h-3, and gf14h-4) and their wild-type sibling (WTArroz) (S9 Fig). We detected significant drops in the germination percentage in all four knockout lines compared to WTArroz (Fig 3A and 3B). We also found that the knockout lines tended to have lower germination rates than WTArroz under normal temperature conditions (25°C) (S10 Fig), which is consistent with the previous report [27]. Furthermore, we generated transgenic lines in the Hitomebore background overexpressing the functional GF14h allele from Arroz da Terra under the control of the CaMV 35S promoter. In these overexpression lines, GF14h expression increased approximately 1,000-fold compared to the wild-type sibling (WTHitomebore) (S11 Fig). Importantly, the overexpression lines showed higher LTG than WTHitomebore when tested at 15°C (Fig 3C and 3D). Taken together, these data indicate that GF14h is a key gene involved in LTG.

thumbnail
Download:
Fig 3. Effect of GF14h mutation and overexpression on low-temperature germination.

(A) Representative photographs showing seed germination in wild-type harboring Arroz-type GF14h (WTArroz) and CRISPR/Cas9 knockout lines (gf14h-1) at 8 days after the onset of seed imbibition. Scale bar, 1 cm. (B) Seed germination rate of WTArroz and its CRISPR/Cas9 knockout lines at 7 days of seed imbibition at 15°C. The two target constructs (S9 Fig) were introduced into the qLTG11-NIL line. Data are means ± standard error (SE, n = 3). Different lowercase letters indicate significant differences based on Tukey’s HSD test (P < 0.01). (C) Representative photographs showing seed germination of wild-type (WTHitomebore) and OsGF14hArroz overexpression lines (OsGF14hArroz-Ox #2) at 7 days of seed imbibition at 15°C. Scale bar, 1 cm. (D) Seed germination rate of WTHitomebore and GF14hArroz overexpression lines in the Hitomebore background at 7 days of seed imbibition at 15°C. Data are means ± SE (n = 3). Different lowercase letters indicate significant differences based on Tukey’s HSD test (P < 0.05).

https://doi.org/10.1371/journal.pgen.1011369.g003

Loss-of-function alleles GF14h and qLTG3-1 increased in frequency during rice breeding in Japan

We reconstructed the GF14h haplotype network using genotype data obtained from whole-genome resequencing of 492 O. sativa accessions from various collections, including the World Rice Core Collection [33], the Rice Core Collection of Japanese Landraces [34], and a set of Japanese landraces and modern varieties [35], in addition to 11 wild rice (O. rufipogon) accessions [36] (S2 Table). We distinguished 10 haplotypes for the GF14h coding region based on 11 polymorphic sites comprising one frameshift mutation caused by a 4-bp deletion, four nonsynonymous single nucleotide polymorphisms (SNPs), and six synonymous SNPs (S3 Table). The conversion of the functional allele Hap2 to its nonfunctional allele Hap1 required only a single step: a 4-bp deletion (S12 Fig).

We analyzed the haplotype frequencies of GF14h and qLTG3-1 in Japanese landraces and cultivars, which we grouped according to their time of release. More than half of all Japanese landraces carried functional alleles of both GF14h and LTG3-1 (Fig 4A). The next most common allele combination among the Japanese landraces was a nonfunctional GF14h allele with a functional LTG3-1 allele (Fig 4A). The percentage of lines with loss-of-function alleles at both GF14h and LTG3-1 has increased since the beginning of crossbreeding in Japan in the early 20th century, with more than 80% of varieties released after 2001 carrying loss-of-function alleles for both genes (Fig 4A and 4B). Looking at each gene separately in landraces, only a few lines carried a loss-of-function allele for LTG3-1, whereas roughly half of all lines already harbored a loss-of-function allele for GF14h (Fig 4B). Modern breeding thus appears to have increased the proportion of loss-of-function alleles for these two genes, with a substantial increase in LTG3-1, reaching almost 90% among lines bred after 2001 (Fig 4B).

thumbnail
Download:
Fig 4. Gradual selection of loss-of-function alleles in GF14h and qLTG3-1 during rice breeding in Japan.

A total of 350 Japanese varieties were examined, including the World Rice Core Collection [33], the Rice Core Collection of Japanese Landraces [34], and the collection of Japanese core cultivars [35]. The allele type at each gene was determined to be functional or nonfunctional by the k-mer method using Illumina short reads for each variety. (A) Proportion of allele type combinations at GF14h and qLTG3-1 sorted by breeding year. (B) Proportion of nonfunctional allele types at GF14h and qLTG3-1 sorted by breeding year.

https://doi.org/10.1371/journal.pgen.1011369.g004

The Arroz da Terra GF14h allele could be valuable for rice breeding

To assess how useful the above findings might be to practical breeding programs, we developed new breeding materials. QTL pyramiding, a strategy for introducing multiple QTLs for desired traits into a single genetic background, is a key strategy employed in current breeding. An essential step in QTL pyramiding is the generation of NILs containing the desired QTLs. Therefore, we developed a NIL, termed NIL-GF14hArroz, using the elite cultivar Hitomebore as the genetic background into which we introgressed a 172-kb region from the Arroz da Terra genome containing GF14h (S3 Fig). This NIL showed a higher seed germination rate under low-temperature conditions compared to Hitomebore (Fig 5A and 5B). In addition, although no significant difference was observed, it is likely that NIL-GF14hArroz tends to be more susceptible to pre-harvest sprouting than Hitomebore (S13 Fig). Importantly, we observed no substantial differences in five agronomic traits (culm length, panicle length, panicle number, grain number, and grain weight) between NIL-GF14hArroz and Hitomebore (Fig 5C–5G). Brown rice yield was slightly lower in NIL-GF14hArroz compared to Hitomebore, but a sufficient yield was guaranteed (Fig 5H and 5I). These results suggest that NIL-GF14hArroz could be a valuable parental line for breeding via QTL pyramiding.

thumbnail
Download:
Fig 5. Phenotypic analysis of a near-isogenic line homozygous for Arroz-type GF14h in the Hitomebore genetic background (NIL-GF14hArroz).

(A) Gross morphology of Hitomebore and NIL-GF14hArroz at 87 days after transplanting (top row), and seed germination at 9 days after the onset of seed imbibition at 15°C (bottom row). Scale bars, 20 cm (top), 1 cm (bottom). (B) Low-temperature germination ability of NIL-GF14hArroz at 8 days after the onset of seed imbibition at 15°C. Values are means ± SE of biologically independent samples (n = 3). Asterisks indicate significant differences, as determined by two-tailed t-test. (C–G) Agronomic traits in Hitomebore and NIL-GF14hArroz: culm length (C), panicle length (D), panicle number (E), grain number per plant (F), and grain weight per plant (G). Values are means ± SE of biologically independent plants (n = 40). (H) Side view of Hitomebore and NIL-GF14hArroz growing in the field under typical rice-growing conditions at the maturity stage. (I) Brown rice yield per unit area. Values are means ± SE of biologically independent plots (n = 3). P-values calculated by t-tests are listed throughout the figure.

https://doi.org/10.1371/journal.pgen.1011369.g005

Discussion

Here, we demonstrated that the functional GF14h allele present in Portuguese rice variety Arroz da Terra plays a pivotal role in supporting seed germinability under low-temperature conditions. Although the regulation of seed germination by GF14h was recently documented, its activity under low-temperature conditions remained unclear [27,28]. While many genomic regions associated with LTG have been detected through QTL mapping and GWAS, only a few studies have identified the causal genes [3–20]. Indeed, LTG is a quantitative trait involving the cumulative effects of multiple genes and their epistatic relationships, making it difficult to assess the specific effect of a single genomic region on LTG. To eliminate the influence of other chromosomal regions on LTG, we first generated qLTG11-NIL containing only one of three QTLs detected in the Arroz da Terra background for genetic analysis. The analysis of qLTG11-NIL revealed that GF14h participates in LTG. Significantly, the NIL harboring the functional GF14h allele from Arroz da Terra in the Hitomebore background provides valuable pre-breeding materials for QTL pyramiding. These findings provide a genetic understanding of low-temperature germinability as well as new resources for rice breeding.

Influence of GF14h on low-temperature germination

In this study, we identified GF14h as being implicated in LTG. In a previous study, a genetic complementation assay with a functional GF14h allele introduced into the rice cultivar Nipponbare background increased the germination rate at 30°C, but only to a limited extent at 15°C [28]. This result is not consistent with our finding that introducing functional GF14h into Hitomebore resulted in a significant improvement in germination at low temperature. Perhaps this discrepancy is due to differences in the rice varieties used in the germination assays. Notably, the haplotypes of qLTG3-1, a major QTL behind LTG [3], are different between Nipponbare and Hitomebore: whereas Nipponbare, which was released in 1961, harbors the functional allele of qLTG3-1, Hitomebore carries a loss-of-function allele with a deletion of 71 bp (S2 Fig) [37]. Moreover, the cultivars Koshihikari and Hayamasari, which carry the same loss-of-function qLTG3-1 allele as the Hitomebore variety, were reported to exhibit lower germination rates at low temperatures than Nipponbare [37]. LTG tests using chromosome segment substitution lines derived from a cross between Koshihikari and Nipponbare indicated that qLTG3-1 contributes to the difference in LTG between the two varieties [37]. Based on these observations, it is likely that Nipponbare has a better LTG ability than Hitomebore, which may have masked the effect of functional GF14h on LTG in the previous study [28]. Our study confirmed the involvement of GF14h in LTG through map-based cloning and analysis of knockout and overexpression lines. It is also worth mentioning that our experiments were performed in the qLTG11-NIL background, which allowed us to isolate the contribution of GF14h to LTG without any influence from qLTG3-1 or other genes in the Arroz da Terra background. In summary, we provided multiple lines of evidence that GF14h contributes to LTG.

The expression pattern of functional GF14h during seed germination was previously unclear. While GF14h has been shown to be expressed in the aleurone layer surrounding the embryo [28], it is also highly expressed in the endosperm [27]. In the current study, we showed that GF14h was expressed throughout the seeds, but with a transient induction in expression in the endosperm at roughly the time of initiation of seed germination. GF14h was reported to regulate seed germination by interacting with the abscisic acid and gibberellin signaling pathways at optimal temperatures [27, 28]. We therefore suggest that GF14h controls LTG by interacting with various phytohormone signaling pathways.

Low-temperature germinability in rice was lost due to selection in modern Japanese breeding

In this study, we performed haplotype network analysis of GF14h using many Japanese rice varieties. We identified ten distinct haplotypes based on 11 polymorphic sites in the GF14h coding region. Of these, Hap4, encompassing the aus, indica, tropical japonica, temperate japonica, and O. rufipogon accessions, was defined at the center of the haplotype network. Furthermore, we determined that a 4-bp deletion converted the functional haplotype Hap2, which was derived from Hap4, into the nonfunctional haplotype Hap1. This finding is consistent with the relationship between Hap6 and Hap1 (which we defined as Hap2 and Hap1, respectively) observed by [27]. These results suggest that the nonfunctional allele represented by Hap1 was introduced into temperate japonica varieties from tropical japonica varieties carrying Hap2 and then spread to Japanese cultivars.

We studied the haplotype frequencies of GF14h and qLTG3-1 in various Japanese landraces and cultivars, considering their time of release from breeding programs into the field. More than half of the Japanese landraces analyzed carried both functional GF14h and qLTG3-1 alleles. However, the frequency of varieties carrying loss-of-function alleles for both GF14h and qLTG3-1 has increased since crossbreeding began in the early 20th century. This trend has continued to the present, perhaps as a result of artificial selection to improve resistance to pre-harvest sprouting, with more than 80% of all varieties bred since 2001 carrying these loss-of-function alleles.

Our study provides a historical perspective on allelic shifts in Japanese rice breeding while highlighting the influence of modern breeding on genetic diversity. Further research is needed to elucidate the potential effects of higher frequencies of loss-of-function alleles on the overall phenotypic characteristics and ecological adaptability of Japanese rice varieties. In addition, as mentioned in previous reports [27,28], we believe that the reintroduction of functional alleles should be considered in order to develop cultivars suitable for labor-saving cultivation techniques such as direct seeding.

Application to direct seeding for rice cultivation

Cultivation stability under direct seeding conditions is important for managing rice production costs and reducing labor. However, since rice is sensitive to low temperatures, improving seed germination and seedling establishment at low temperatures is a desirable breeding trait in high-latitude rice production areas such as Japan. In the current study, we developed a potentially useful NIL (NIL-GF14hArroz) by introducing the functional GF14h allele into the Hitomebore background. Overexpressing this functional GF14h allele was previously shown to improve anaerobic germination and tolerance to seedling establishment under anaerobic conditions in laboratory experiments [27]. However, whether NIL-GF14hArroz exhibits strong seedling vigor at low temperatures in rice fields remains to be determined. We previously identified a QTL associated with seedling vigor, qPHS3-2 (QTL for plant height of seedling 3–2) [38]. qPHS3-2 most likely corresponds to the gibberellin biosynthesis gene GA20 oxidase1 (OsGA20ox1), a paralog of Semi Dwarf1 (sd-1, corresponding to OsGA20ox2) [38]. Therefore, the pyramiding of qPHS3-2 in NIL-GF14hArroz by marker-assisted selection represents a promising approach for further improving seedling vigor in NIL-GF14hArroz. On the other hand, enhancing LTG may conversely increase the risk of pre-harvest sprouting. Should the level of pre-harvest sprouting in NIL-GF14hArroz pose practical issues, the pyramiding of QTLs for pre-harvest sprouting resistance, such as Seed Dormancy 4 [39], may offer a solution.

In recent years, “early-winter direct seeding” has been experimentally tested as a new system of direct seeding for rice production in Japan [40]. In this system, seeds are directly sown in the early winter of the previous year instead of the spring. The sown seeds thus overwinter in snow-covered soil and germinate the following spring. The major advantage of this approach is that it can significantly decrease the amount of labor required by farmers during the busy spring season. However, it is challenging to overwinter the seeds of modern rice varieties in the soil and achieve good seedling establishment [40–42]. We expect that reintroducing beneficial alleles like GF14hArroz that were lost during modern breeding into future rice varieties will enable the implementation of new cultivation practices and increase productivity.

Materials and methods

Plant materials

Rice was cultivated in a paddy field at Iwate Agricultural Research Center (39°35’N, 141°11’E). A recombinant inbred line (RIL) population of 200 F7 lines was generated from a cross between Japanese variety Iwatekko and the high-LTG variety Arroz da Terra (S1 Fig) [31]. To develop NILs, rice cultivar Hitomebore was used as the recipient parent to generate NILs harboring the target genomic region from Arroz da Terra (S3A Fig), thereby establishing qLTG3-2-NIL, qLTG11-NIL, and NIL-GF14hArroz (Figs 1E, S3B and S4A).

Evaluation of germination rate

Seeds for each line were harvested 45 days after heading, air-dried at 30°C for two days, and stored at 4°C until use. The seeds were air-dried at 50°C for seven days in the dark to break dormancy. For germination tests, 50 seeds per replicate were incubated in a Petri dish filled with distilled water in the dark at 15°C (low-temperature conditions) or 25°C (optimal temperature conditions). For QTL-seq analysis, germination tests were conducted at 13°C [31]. The germination rate was calculated as the total number of germinated seeds at each time point divided by the number of seeds tested. Seeds were considered to have germinated when the white coleoptile was visible.

To evaluate resistance to pre-harvest sprouting, panicles were harvested from NIL-GF14hArroz and Hitomebore 30 days after heading. The panicles were incubated in dark, wet conditions (by covering them with filter paper moistened with water) at 28°C for 10 days, and seed germination was scored.

QTL-seq analysis

LTG data for the 200 RILs derived from a cross between Iwatekko and Arroz da Terra (S1 Fig) were analyzed [31]. The top 20 RILs showing high-LTG and the bottom 20 RILs showing low-LTG phenotypes were selected to assemble the two bulk samples with contrasting LTG phenotypes. All seedlings with high or low LTG were pooled, and DNA was extracted from each bulk as previously described [31]. The genomic DNA of the two bulks was used to generate DNA-seq libraries and sequenced on a GAIIx sequencer (Illumina, CA, USA). QTL-seq was performed to identify QTLs related to LTG [31,32].

Map-based cloning of qLTG11

To narrow down the qLTG11 region, genotyping was performed using a cross population of qLTG11-NIL (BC2F3) backcrossed to Hitomebore. The germination rate under low-temperature conditions (15°C) was measured to characterize LTG activity. High-resolution fine mapping with ten markers (markers B−K) between 23.466 Mb and 23.609 Mb on chromosome 11 identified six informative recombinants in the target region. Primers used for mapping are listed in S4 Table.

De novo assembly of the Hitomebore and Arroz da Terra genomes

To reconstruct the qLTG11 regions in Hitomebore and Arroz da Terra, de novo assembly was performed for each genome using Nanopore long reads and Illumina short reads according to a published method [43]. To extract high-molecular-weight genomic DNA from leaf tissue for Nanopore sequencing, a NucleoBond high-molecular-weight DNA kit (MACHEREY-NAGEL, Germany) was used. Following DNA extraction, low-molecular-weight DNA was eliminated using a Short Read Eliminator Kit XL (Circulomics, MD, USA). Library preparation was then performed using a Ligation Sequencing Kit (SQK-LSK-109; Oxford Nanopore Technologies [ONT], United Kingdom) according to the manufacturer’s instructions, and sequencing was performed using MinION (ONT, UK) for Arroz da Terra. For Hitomebore, Nanopore long reads sequenced by [43] were used. Base-calling of the Nanopore long reads was performed using Guppy 4.4.2 (ONT, UK). Sequences derived from the lambda phage genome were removed from the raw reads with NanoLyse v1.1.0 [44]. The first 50 bp of each read were then removed, as were reads with an average read quality score below 7 and reads shorter than 3,000 bases, using NanoFilt v2.7.1 [44]. The clean Nanopore long reads were assembled using NECAT v0.0.1 [45], setting the genome size to 380 Mb. To improve the accuracy of assembly, Racon v1.4.20 [46] was used twice for error correction using Nanopore reads, and Medaka v1.4.1 (https://github.com/nanoporetech/medaka) was subsequently used to correct mis-assembly. Two rounds of consensus correction were then performed using bwa-mem v0.7.17 [47] and HyPo v1.0.3 [48] with the Illumina short reads. Redundant contigs were removed using purge-haplotigs v1.1.1 [49], resulting in a de novo assembly of 374.8 Mb comprising 82 contigs for Hitomebore and 376.3 Mb consisting of 82 contigs for Arroz da Terra. The resulting genome sequences have been deposited in Zenodo (https://doi.org/10.5281/zenodo.10460309).

Plant transformation

To generate GF14h knockout mutants, two single guide RNAs (sgRNAs) targeting exon 4 or exon 5 of GF14h were designed using the web-based service CRISPRdirect (crispr.dbcls.jp) [50] and cloned individually into the pZH::OsU6gRNA::MMCas9 vector [51]. The resulting vectors were introduced into Agrobacterium (Agrobacterium tumefaciens) strain EHA105 for transformation into qLTG11-NIL plants [52]. The target sites in the positive transformants were sequenced by Sanger sequencing to detect mutations. To obtain overexpression constructs, the full-length coding sequence of functional GF14h was amplified from total RNA extracted from qLTG11-NIL and cloned into the plant binary vector pCAMBIA1300 under the control of the cauliflower mosaic virus (CaMV) 35S promoter. The overexpression plasmid was introduced into Agrobacterium strain EHA105 for transformation of rice variety Hitomebore [52]. All primers used are listed in S4 Table.

Expression analysis

Total RNA was extracted from germinating seeds using an RNA-suisui S kit (Rizo). Total RNA was treated with RNase-free DNase I (Nippon Gene). The resulting samples were reverse transcribed into first-strand cDNA using a PrimeScript RT Reagent Kit (Takara Bio). Quantitative PCR (qPCR) was conducted using a QuantStudio 3 system (Thermo Fisher Scientific) with Luna Universal qPCR Master Mix (New England Biolabs). The cycling parameters were 1 min at 95°C, followed by 40 cycles of amplification (95°C for 15 sec and 60°C for 30 sec). The Actin gene (Os03g0718100) served as an internal control, and the Delta CT method was used to calculate the relative expression levels. The primer sets are listed in S4 Table.

RNA-seq

Total RNA was extracted from Hitomebore and qLTG11-NIL seeds at 0, 1, 2, and 3 days after the onset of seed hydration under low (15°C) or optimum (25°C) temperature conditions using an RNA-suisui S kit (Rizo, Ibaraki, Japan). Sequencing libraries were prepared using an NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs Japan, Tokyo, Japan) following the manufacturer’s protocol. The libraries were sequenced in paired-end mode using an Illumina HiSeq X instrument (Illumina, CA, USA). The raw reads have been deposited in the DNA Databank of Japan (BioProject accession No. PRJDB17450; S5 Table). For quality control, reads shorter than 50 bases and those with an average read quality below 20 were discarded using Trimmomatic v0.36 [53], and poly(A) sequences were trimmed using PRINSEQ++ v1.2 [54]. The resulting clean reads were aligned to the de novo assembled Hitomebore and Arroz da Terra genomes with HISAT2 v2.1 [55]. BAM files were sorted and indexed with SAMtools v1.10 [56], and aligned reads were assembled into transcripts with StringTie [57] by combining bam files for each variety. In a similar manner, the expression data were generated using the Nipponbare reference genome downloaded from IRGSP-1.0 (https://rapdb.dna.affrc.go.jp/download/irgsp1.html).

Haplotype network analysis

Sequencing datasets were obtained for 503 rice accessions. Of these, 379 were FASTQ files downloaded from the DNA Data Bank of Japan Sequence Read Archive (DRA) [33–35,58] and 124 were sequenced in this study (S2 Table). Details about DNA extraction, whole-genome sequencing techniques, and construction of the genotype datasets in VCF format are provided in a previous report [35]. This study specifically focused on the coding region of GF14h. Genotype information related to the coding region of GF14h was extracted from the VCF dataset. In addition, the k-mer analysis program (https://github.com/taitoh1970/kmer) [59] was used with Illumina short reads to detect the 4-bp deletion with high sensitivity. Genotype information for the presence of the 4-bp deletion was added to the VCF file, and 81 samples with heterozygous genotypes were discarded. A haplotype network was then constructed using the median-joining network algorithm [60] implemented in Popart v1.7 [61].

Evaluation of agronomic traits and yield performance of NIL-GF14hArroz

The grain yields of Hitomebore and NIL-GF14hArroz were investigated in experimental paddy fields in 2023. Field experiments were conducted at the Iwate Agricultural Research Center (39°35’N, 141°11’E) in Kitakami, Iwate, Japan. A fertilization regime of N:P2O5:K2O = 6:6:6 g m−2 was applied as a basal dressing, and N:K2O = 2:2 g m-2 was applied as a top dressing. Seeds were sown in a seedling nursery box on 21 April, and seedlings were transplanted to the paddy field on 18 May. To evaluate agronomic traits, the seedlings were transplanted at a rate of one plant per hill, with a planting density of 22.2 hills m−2. Culm length, panicle length, panicle number, grain number per plant, and grain weight per plant were measured at maturity. To evaluate yield performance, seedlings were transplanted with three plants per hill at a planting density of 16.7 hills m−2. The 0.9 × 5.0 m experimental plots in the paddy fields were arranged in a randomized complete block design with three replicates. At maturity, 50 hills were harvested from each plot to measure brown rice yield. The hulls were removed using a rice huller (Model 25MC, Ohya Tanzo Factory Co., Ltd., Japan), and the hulled grains were screened with a grain sorter (1.9-mm sieve size). Brown rice yields were adjusted to 15% moisture content and converted to weight per hectare.

Supporting information

S1 Fig. Frequency distribution of germination rates in the RIL population and germination rates of selected RILs for QTL-seq analysis.

(A) Frequency distribution of germination rates at 13°C after eight days from seed imbibition in 200 F7 RILs derived from a cross between Iwatekko and Arroz da Terra [31]. (B) Selection of RILs with low cold germination rates. The 62 RILs with the lower germination rates from the first test (A) were tested for the second, and then the 20 RILs with the lowest germination rates were selected as a bulk sample for QTL-seq analysis. The bar graph shows the mean values of the two tests. (C) Selection of RILs with high cold germination rates. The 37 RILs with the higher germination rates from the first test (A) were tested for the second, and then the 20 RILs with the highest germination rates were selected for a bulk sample for QTL-seq analysis. The bar graph shows the average values of the two tests.

https://doi.org/10.1371/journal.pgen.1011369.s001

(PDF)

S2 Fig. Multiple DNA sequence alignment of qLTG3-1 variants.

Arroz da Terra and Italica Livorno harbor a functional qLTG3-1 variant. Nipponbare carries another functional qLTG3-1 variant due to the nonsynonymous substitution (*). Iwatekko, Hitomebore, and Hayamasari contain a loss-of-function variant for qLTG3-1 due to a 71-bp deletion.

https://doi.org/10.1371/journal.pgen.1011369.s002

(PDF)

S3 Fig. Generation of a near-isogenic line with high LTG in the Hitomebore background.

(A) Strategy for the development of qLTG3-2-NIL, qLTG11-NIL, and NIL-GF14hArroz. Molecular markers were used for foreground and background selection. (B) Diagram showing the genotype of NIL-GF14hArroz. NIL-GF14hArroz contains a 172-kb region on chromosome 11 harboring the Arroz da Terra allele of GF14h. Light blue bars indicate genomic fragments from Hitomebore; red bars indicate genomic fragments from Arroz da Terra.

https://doi.org/10.1371/journal.pgen.1011369.s003

(PDF)

S4 Fig. Summary of qLTG3-2.

(A) Diagram showing the genotype of qLTG3-2-NIL containing the Arroz da Terra allele at qLTG3-2 on chromosome 3. Light blue bars indicate genomic fragments from Hitomebore; red bars indicate genomic fragments from Arroz da Terra; dark blue bars indicate heterozygous regions. (B) Germination time courses for seeds of Hitomebore, the qLTG3-2-NIL, and Arroz da Terra at 15°C. Values are means ± SD of biologically independent samples (Hitomebore and NIL n = 10, Arroz da Terra n = 5).

https://doi.org/10.1371/journal.pgen.1011369.s004

(PDF)

S5 Fig. Seed germination of qLTG11-NIL under optimal temperature conditions.

Germination time courses of seeds from Hitomebore and qLTG11-NIL at 25°C. Values are means ± SD of biologically independent samples (n = 3). Two-tailed t-test was used between qLTG11-NIL and Hitomebore for each time point (*P < 0.05 and **P < 0.01).

https://doi.org/10.1371/journal.pgen.1011369.s005

(PDF)

S6 Fig. Comparison of the qLTG11 genomic region in Hitomebore, Arroz da Terra, and Nipponbare.

Dot blot analyses of the genomic sequence in the qLTG11 candidate region between (A) Hitomebore and Nipponbare and (B) Hitomebore and Arroz da Terra, using D-GENIES [62]. Based on the Nipponbare genome (IRGSP-1.0), the genomic region containing the causative gene is located at 23.512–23.564 Mb (approximately 52 kb) on chromosome 11. The genome sequence of Hitomebore is identical to that of Nipponbare. The candidate region corresponds to a fragment of approximately 94 kb in the Arroz da Terra genome.

https://doi.org/10.1371/journal.pgen.1011369.s006

(PDF)

S7 Fig. Expression levels of the two annotated genes in the candidate genomic region of qLTG11 based on RNA-seq data.

Total RNA was extracted from Hitomebore and qLTG11-NIL seeds at 0, 1, 2, and 3 days after the onset of seed imbibition under 15 or 25°C temperature conditions, followed by RNA-seq. The sequence reads were mapped to the Nipponbare genome (IRGSP-1.0), and expression data were obtained. (A–B) The expression levels of Os11g0609600 (GF14h) during seed germination under 15°C (A) and 25°C (B) are shown. Data are presented as means ± SE. n = 3 biologically independent samples. (C–D) The expression levels of Os11g0609500 (Jacalin-like lectin domain containing protein) during seed germination under 15°C (C) and 25°C (D) are shown. Data are presented as means ± SE. n = 3 biologically independent samples.

https://doi.org/10.1371/journal.pgen.1011369.s007

(PDF)

S8 Fig. Multiple DNA sequence alignment of GF14h variants.

Arroz da Terra carries a functional GF14h variant. Hitomebore and Nipponbare harbors a loss-of-function variant of GF14h due to a 4-bp deletion (black line).

https://doi.org/10.1371/journal.pgen.1011369.s008

(PDF)

S9 Fig. CRISPR/Cas9-mediated genome editing of GF14h.

Top, diagram showing the GF14h locus, with the locations of the two sgRNA target sites marked by inverted red triangles. Bottom, sequencing results of putative gf14h mutants. The sgRNA target sites are underlined, and the PAMs are highlighted. The mutation sites in GF14hArroz for the four mutants (gf14h-1, gf14h-2, gf14h-3, and gf14h-4) are indicated.

https://doi.org/10.1371/journal.pgen.1011369.s009

(PDF)

S10 Fig. Effect of GF14h mutation on optimal-temperature germination.

(A) Representative photographs showing seed germination in wild-type harboring Arroz-type GF14h (WTArroz) and CRISPR/Cas9 knockout lines (gf14-1) at 3 days after the onset of seed imbibition. Scale bar, 1 cm. (B) Seed germination rate of WTArroz and its CRISPR/Cas9 knockout lines at 2 days of seed imbibition at 25°C. The two target constructs (S9 Fig) were introduced into the qLTG11-NIL line. Data are means ± standard error (WTArroz, gf14h-1, gf14h-2 and gf14h-4, n = 3; gf14h-3, n = 2). Different lowercase letters indicate significant differences based on Tukey’s HSD test (P < 0.05).

https://doi.org/10.1371/journal.pgen.1011369.s010

(PDF)

S11 Fig. Relative GF14h expression levels in germinating seeds of GF14hArroz overexpression lines and the parental line.

The GF14hArroz overexpression construct was introduced into Hitomebore. OsActin1 (Os03g0718100) was used for normalization. Values are means ± SE (n = 3 or 4). Different lowercase letters indicate significant differences based on Tukey’s HSD test (P < 0.001).

https://doi.org/10.1371/journal.pgen.1011369.s011

(PDF)

S12 Fig. Haplotype network of GF14h.

The GF14h genomic sequences obtained from 411 O. sativa varieties and 11 O. rufipogon accessions were used for analysis (S1 Table). The haplotype network was reconstructed by the median joining network algorithm [60] implemented in Popart v1.7 [61]. The haplotype Hap1 evolved from Hap2 by acquiring the 4-bp sequence, resulting in a nonfunctional GF14h gene. The Hitomebore cultivar contains Hap1 (nonfunctional), and the Arroz da Terra cultivar contains Hap9 (functional).

https://doi.org/10.1371/journal.pgen.1011369.s012

(PDF)

S13 Fig. Pre-harvest sprouting of NIL-GF14hArroz.

Germination time courses of seeds from Hitomebore (blue circles) and the NIL-GF14hArroz (pink triangles) under wet conditions at 28°C. Seeds were harvested from tagged panicles 30 days after heading. Values are means ± SE of biologically independent samples (n = 3). The P-values calculated from t-tests at each time point are shown in the figure.

https://doi.org/10.1371/journal.pgen.1011369.s013

(PDF)

S14 Fig. Development of a functional marker based on the 4-bp deletion in GF14h.

(A) Diagram of the sequence around the 4-bp InDel of GF14h. In the Hitomebore (loss-of-function) allele, the 4-bp deletion creates a SmlI restriction site. (B) Genotyping of the 4-bp deletion in GF14h. A genomic fragment containing the 4-bp InDel of GF14h was amplified by PCR and digested with SmlI. The products were separated on a 3% (w/v) agarose gel and stained with Midori Green. The PCR product from GF14hArroz (approximately 500 bp) was not cleaved, whereas the PCR product from GF14hHitomebore was cleaved, producing two fragments of approximately 250 bp each. Both bands were detected in heterozygous plants.

https://doi.org/10.1371/journal.pgen.1011369.s014

(PDF)

S1 Table. Expression profile (TPM) during seed germination by RNA-seq.

https://doi.org/10.1371/journal.pgen.1011369.s015

(CSV)

S2 Table. List of rice varieties used in this study and their sequence read archive (SRA) IDs.

https://doi.org/10.1371/journal.pgen.1011369.s016

(XLSX)

S3 Table. Haplotypes of the GF14h gene analyzed in S12 Fig.

https://doi.org/10.1371/journal.pgen.1011369.s017

(XLSX)

S4 Table. Primers used in this study.

https://doi.org/10.1371/journal.pgen.1011369.s018

(XLSX)

S5 Table. List of RNA-seq samples used in this study and their sequence read archive (SRA) IDs.

https://doi.org/10.1371/journal.pgen.1011369.s019

(XLSX)

S6 Table. Numerical data used for the graphs.

https://doi.org/10.1371/journal.pgen.1011369.s020

(XLSX)

Acknowledgments

We thank the National Agriculture and Food Research Organization (NARO) gene bank, Japan, for providing rice seeds. This work was performed using the National Institute of Genetics (NIG) supercomputer at the Research Organization of Information and Systems (ROIS) National Institute of Genetics and the Academic Center for Computing and Media Studies (ACCMS) supercomputer at Kyoto University.

References

  1. 1. Okuno K. Molecular Mechanisms of Cold Tolerance in Rice and Wheat. Thermal Medicine(Japanese Journal of Hyperthermic Oncology). 2004;20(2):51–60.
  2. 2. Iwata N, Shinada H, Kiuchi H, Sato T, Fujino K. Mapping of QTLs controlling seedling establishment using a direct seeding method in rice. Breeding Science. 2010;60(4):353–60.
  3. 3. Fujino K, Sekiguchi H, Matsuda Y, Sugimoto K, Ono K, Yano M. Molecular identification of a major quantitative trait locus, qLTG3-1, controlling low-temperature germinability in rice. Proc Natl Acad Sci U S A. 2008;105(34):12623–8. Epub 20080821. pmid:18719107; PubMed Central PMCID: PMC2527961.
  4. 4. Miura K, Lin SY, Yano M, Nagamine T. Mapping Quantitative Trait Loci Controlling Low Temperature Germinability in Rice (Oryza sativa L.). Breeding Science. 2001;51(4):293–9.
  5. 5. Fujino K, Sekiguchi H, Sato T, Kiuchi H, Nonoue Y, Takeuchi Y, et al. Mapping of quantitative trait loci controlling low-temperature germinability in rice (Oryza sativa L.). Theor Appl Genet. 2004;108(5):794–9. Epub 20031118. pmid:14624339.
  6. 6. Han LZ, Zhang YY, Qiao YL, Cao GL, Zhang SY, Kim JH, et al. Genetic and QTL analysis for low-temperature vigor of germination in rice. Yi Chuan Xue Bao. 2006;33(11):998–1006. Epub 2006/11/23. pmid:17112971.
  7. 7. Jiang L, Liu S, Hou M, Tang J, Chen L, Zhai H, et al. Analysis of QTLs for seed low temperature germinability and anoxia germinability in rice (Oryza sativa L.). Field Crops Research. 2006;98(1):68–75.
  8. 8. Nguyen HN, Park I-K, Yeo S-M, Yun Y-T, Ahn S-N. Mapping quantitative trait loci controlling low-temperature germinability in rice. Korean Journal of Agricultural Science. 2012;39(4):477–82.
  9. 9. Li L, Liu X, Xie K, Wang Y, Liu F, Lin Q, et al. qLTG-9, a stable quantitative trait locus for low-temperature germination in rice (Oryza sativa L.). Theor Appl Genet. 2013;126(9):2313–22. Epub 20130608. pmid:23748708.
  10. 10. Fujino K, Obara M, Shimizu T, Koyanagi KO, Ikegaya T. Genome-wide association mapping focusing on a rice population derived from rice breeding programs in a region. Breed Sci. 2015;65(5):403–10. Epub 20151201. pmid:26719743; PubMed Central PMCID: PMC4671701.
  11. 11. Pan Y, Zhang H, Zhang D, Li J, Xiong H, Yu J, et al. Genetic analysis of cold tolerance at the germination and booting stages in rice by association mapping. PLoS One. 2015;10(3):e0120590. Epub 20150319. pmid:25790128; PubMed Central PMCID: PMC4366098.
  12. 12. Satoh T, Tezuka K, Kawamoto T, Matsumoto S, Satoh-Nagasawa N, Ueda K, et al. Identification of QTLs controlling low-temperature germination of the East European rice (Oryza sativa L.) variety Maratteli. Euphytica. 2015;207(2):245–54.
  13. 13. Jiang N, Shi S, Shi H, Khanzada H, Wassan GM, Zhu C, et al. Mapping QTL for Seed Germinability under Low Temperature Using a New High-Density Genetic Map of Rice. Front Plant Sci. 2017;8:1223. Epub 20170712. pmid:28747923; PubMed Central PMCID: PMC5506081.
  14. 14. Shakiba E, Edwards JD, Jodari F, Duke SE, Baldo AM, Korniliev P, et al. Genetic architecture of cold tolerance in rice (Oryza sativa) determined through high resolution genome-wide analysis. PLoS One. 2017;12(3):e0172133. Epub 20170310. pmid:28282385; PubMed Central PMCID: PMC5345765.
  15. 15. Wang X, Zou B, Shao Q, Cui Y, Lu S, Zhang Y, et al. Natural variation reveals that OsSAP16 controls low-temperature germination in rice. J Exp Bot. 2018;69(3):413–21. Epub 2017/12/14. pmid:29237030; PubMed Central PMCID: PMC5853544.
  16. 16. Jiang S, Yang C, Xu Q, Wang L, Yang X, Song X, et al. Genetic Dissection of Germinability under Low Temperature by Building a Resequencing Linkage Map in japonica Rice. Int J Mol Sci. 2020;21(4). Epub 20200214. pmid:32074988; PubMed Central PMCID: PMC7072905.
  17. 17. Shim KC, Kim SH, Lee HS, Adeva C, Jeon YA, Luong NH, et al. Characterization of a New qLTG3-1 Allele for Low-temperature Germinability in Rice from the Wild Species Oryza rufipogon. Rice (N Y). 2020;13(1):10. Epub 20200205. pmid:32025935; PubMed Central PMCID: PMC7002630.
  18. 18. Yang T, Zhou L, Zhao J, Dong J, Liu Q, Fu H, et al. The Candidate Genes Underlying a Stably Expressed QTL for Low Temperature Germinability in Rice (Oryza sativa L.). Rice (N Y). 2020;13(1):74. Epub 20201019. pmid:33074410; PubMed Central PMCID: PMC7573065.
  19. 19. Pan Z, Tan B, Cao G, Zheng R, Liu M, Zeng R, et al. Integrative QTL Identification, Fine Mapping and Candidate Gene Analysis of a Major Locus qLTG3a for Seed Low-Temperature Germinability in Rice. Rice (N Y). 2021;14(1):103. Epub 20211215. pmid:34910270; PubMed Central PMCID: PMC8674402.
  20. 20. Mao F, Wu D, Lu F, Yi X, Gu Y, Liu B, et al. QTL mapping and candidate gene analysis of low temperature germination in rice (Oryza sativa L.) using a genome wide association study. PeerJ. 2022;10:e13407. Epub 20220511. pmid:35578671; PubMed Central PMCID: PMC9107303.
  21. 21. Morrison DK. The 14-3-3 proteins: integrators of diverse signaling cues that impact cell fate and cancer development. Trends Cell Biol. 2009;19(1):16–23. Epub 20081120. pmid:19027299; PubMed Central PMCID: PMC3073487.
  22. 22. Madeira F, Tinti M, Murugesan G, Berrett E, Stafford M, Toth R, et al. 14-3-3-Pred: improved methods to predict 14-3-3-binding phosphopeptides. Bioinformatics. 2015;31(14):2276–83. Epub 20150303. pmid:25735772; PubMed Central PMCID: PMC4495292.
  23. 23. The Hermeking H. 14-3-3 cancer connection. Nat Rev Cancer. 2003;3(12):931–43. pmid:14737123.
  24. 24. Wilson RS, Swatek KN, Thelen JJ. Regulation of the Regulators: Post-Translational Modifications, Subcellular, and Spatiotemporal Distribution of Plant 14-3-3 Proteins. Front Plant Sci. 2016;7:611. Epub 20160509. pmid:27242818; PubMed Central PMCID: PMC4860396.
  25. 25. Camoni L, Visconti S, Aducci P, Marra M. 14-3-3 Proteins in Plant Hormone Signaling: Doing Several Things at Once. Front Plant Sci. 2018;9:297. Epub 20180313. pmid:29593761; PubMed Central PMCID: PMC5859350.
  26. 26. Chen F, Li Q, Sun L, He Z. The rice 14-3-3 gene family and its involvement in responses to biotic and abiotic stress. DNA Res. 2006;13(2):53–63. Epub 20060421. pmid:16766513.
  27. 27. Sun J, Zhang G, Cui Z, Kong X, Yu X, Gui R, et al. Regain flood adaptation in rice through a 14-3-3 protein OsGF14h. Nat Commun. 2022;13(1):5664. Epub 20220929. pmid:36175427; PubMed Central PMCID: PMC9522936.
  28. 28. Yoshida H, Hirano K, Yano K, Wang F, Mori M, Kawamura M, et al. Genome-wide association study identifies a gene responsible for temperature-dependent rice germination. Nat Commun. 2022;13(1):5665. Epub 20220929. pmid:36175401; PubMed Central PMCID: PMC9523024.
  29. 29. Ashikari M, Matsuoka M. Identification, isolation and pyramiding of quantitative trait loci for rice breeding. Trends Plant Sci. 2006;11(7):344–50. pmid:16769240.
  30. 30. Zhang B, Ma L, Wu B, Xing Y, Qiu X. Introgression Lines: Valuable Resources for Functional Genomics Research and Breeding in Rice (Oryza sativa L.). Front Plant Sci. 2022;13:863789. Epub 20220426. pmid:35557720; PubMed Central PMCID: PMC9087921.
  31. 31. Takagi H, Abe A, Yoshida K, Kosugi S, Natsume S, Mitsuoka C, et al. QTL-seq: rapid mapping of quantitative trait loci in rice by whole genome resequencing of DNA from two bulked populations. Plant J. 2013;74(1):174–83. Epub 20130218. pmid:23289725.
  32. 32. Sugihara Y, Young L, Yaegashi H, Natsume S, Shea DJ, Takagi H, et al. High-performance pipeline for MutMap and QTL-seq. PeerJ. 2022;10:e13170. Epub 20220318. pmid:35321412; PubMed Central PMCID: PMC8935991.
  33. 33. Tanaka N, Shenton M, Kawahara Y, Kumagai M, Sakai H, Kanamori H, et al. Whole-Genome Sequencing of the NARO World Rice Core Collection (WRC) as the Basis for Diversity and Association Studies. Plant Cell Physiol. 2020;61(5):922–32. Epub 2020/02/27. pmid:32101292; PubMed Central PMCID: PMC7426033.
  34. 34. Tanaka N, Shenton M, Kawahara Y, Kumagai M, Sakai H, Kanamori H, et al. Investigation of the Genetic Diversity of a Rice Core Collection of Japanese Landraces using Whole-Genome Sequencing. Plant Cell Physiol. 2021;61(12):2087–96. pmid:33539537; PubMed Central PMCID: PMC7861467.
  35. 35. Shimono H, Abe A, Kim CH, Sato C, Iwata H. Upcycling rice yield trial data using a weather-driven crop growth model. Commun Biol. 2023;6(1):764. Epub 20230721. pmid:37479731; PubMed Central PMCID: PMC10362053
  36. 36. Zhao Q, Feng Q, Lu H, Li Y, Wang A, Tian Q, et al. Pan-genome analysis highlights the extent of genomic variation in cultivated and wild rice. Nat Genet. 2018;50(2):278–84. Epub 20180115. pmid:29335547.
  37. 37. Hori K, Sugimoto K, Nonoue Y, Ono N, Matsubara K, Yamanouchi U, et al. Detection of quantitative trait loci controlling pre-harvest sprouting resistance by using backcrossed populations of japonica rice cultivars. Theor Appl Genet. 2010;120(8):1547–57. Epub 20100210. pmid:20145904; PubMed Central PMCID: PMC2859223.
  38. 38. Abe A, Takagi H, Fujibe T, Aya K, Kojima M, Sakakibara H, et al. OsGA20ox1, a candidate gene for a major QTL controlling seedling vigor in rice. Theor Appl Genet. 2012;125(4):647–57. Epub 20120406. pmid:22481119.
  39. 39. Sugimoto K, Takeuchi Y, Ebana K, Miyao A, Hirochika H, Hara N, et al. Molecular cloning of Sdr4, a regulator involved in seed dormancy and domestication of rice. Proc Natl Acad Sci U S A. 2010;107(13):5792–7. Epub 20100310. pmid:20220098; PubMed Central PMCID: PMC2851884.
  40. 40. Shimono H. New technique: Early- winter direct- sowing cultivation in rice under cool climate. Regulation of Plant Growth & Development. 2020;55(1):63–6.
  41. 41. Shimono H, Tamai M, Hamasaki T, Sagawa R, Ohtani R. Effects of Autumn Direct-seeding on Rice Growth and Yield under Cool Climates. Japanese journal of crop science. 2012;81(1):93–8.
  42. 42. Oikawa S, Nishi M, Yui S, Kashiwagi J, Nakashima T, Ichikawa S, et al. Improvement of Seedling Establishment by Seed Coating with Iron in Early-winter Direct-sowing Rice. Japanese Journal of Crop Science. 2019;88(4):259–67.
  43. 43. Sugihara Y, Abe Y, Takagi H, Abe A, Shimizu M, Ito K, et al. Disentangling the complex gene interaction networks between rice and the blast fungus identifies a new pathogen effector. PLoS Biol. 2023;21(1):e3001945. Epub 20230119. pmid:36656825; PubMed Central PMCID: PMC9851567.
  44. 44. De Coster W, D’Hert S, Schultz DT, Cruts M, Van Broeckhoven C. NanoPack: visualizing and processing long-read sequencing data. Bioinformatics. 2018;34(15):2666–9. pmid:29547981; PubMed Central PMCID: PMC6061794.
  45. 45. Chen Y, Nie F, Xie SQ, Zheng YF, Dai Q, Bray T, et al. Efficient assembly of nanopore reads via highly accurate and intact error correction. Nat Commun. 2021;12(1):60. Epub 20210104. pmid:33397900; PubMed Central PMCID: PMC7782737.
  46. 46. Vaser R, Sovic I, Nagarajan N, Sikic M. Fast and accurate de novo genome assembly from long uncorrected reads. Genome Res. 2017;27(5):737–46. Epub 20170118. pmid:28100585; PubMed Central PMCID: PMC5411768.
  47. 47. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60. Epub 20090518. pmid:19451168; PubMed Central PMCID: PMC2705234.
  48. 48. Kundu R, Casey J, Sung W-K. HyPo: Super Fast & Accurate Polisher for Long Read Genome Assemblies. bioRxiv. 2019:2019.12.19.882506.
  49. 49. Roach MJ, Schmidt SA, Borneman AR. Purge Haplotigs: allelic contig reassignment for third-gen diploid genome assemblies. BMC Bioinformatics. 2018;19(1):460. Epub 20181129. pmid:30497373; PubMed Central PMCID: PMC6267036.
  50. 50. Naito Y, Hino K, Bono H, Ui-Tei K. CRISPRdirect: software for designing CRISPR/Cas guide RNA with reduced off-target sites. Bioinformatics. 2015;31(7):1120–3. Epub 20141120. pmid:25414360; PubMed Central PMCID: PMC4382898.
  51. 51. Mikami M, Toki S, Endo M. Comparison of CRISPR/Cas9 expression constructs for efficient targeted mutagenesis in rice. Plant Mol Biol. 2015;88(6):561–72. Epub 20150719. pmid:26188471; PubMed Central PMCID: PMC4523696.
  52. 52. Toki S, Hara N, Ono K, Onodera H, Tagiri A, Oka S, et al. Early infection of scutellum tissue with Agrobacterium allows high-speed transformation of rice. Plant J. 2006;47(6):969–76. pmid:16961734.
  53. 53. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114–20. Epub 20140401. pmid:24695404; PubMed Central PMCID: PMC4103590.
  54. 54. Cantu VA, Sadural J, Edwards R. PRINSEQ++, a multi-threaded tool for fast and efficient quality control and preprocessing of sequencing datasets. PeerJ Preprints. 2019;7:e27553v1.
  55. 55. Kim D, Paggi JM, Park C, Bennett C, Salzberg SL. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat Biotechnol. 2019;37(8):907–15. Epub 20190802. pmid:31375807; PubMed Central PMCID: PMC7605509.
  56. 56. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25(16):2078–9. Epub 20090608. pmid:19505943; PubMed Central PMCID: PMC2723002.
  57. 57. Pertea M, Pertea GM, Antonescu CM, Chang TC, Mendell JT, Salzberg SL. StringTie enables improved reconstruction of a transcriptome from RNA-seq reads. Nat Biotechnol. 2015;33(3):290–5. Epub 20150218. pmid:25690850; PubMed Central PMCID: PMC4643835.
  58. 58. Yabe S, Yoshida H, Kajiya-Kanegae H, Yamasaki M, Iwata H, Ebana K, et al. Description of grain weight distribution leading to genomic selection for grain-filling characteristics in rice. PLoS One. 2018;13(11):e0207627. Epub 20181120. pmid:30458025; PubMed Central PMCID: PMC6245794.
  59. 59. Itoh T, Onuki R, Tsuda M, Oshima M, Endo M, Sakai H, et al. Foreign DNA detection by high-throughput sequencing to regulate genome-edited agricultural products. Sci Rep. 2020;10(1):4914. Epub 20200318. pmid:32188926; PubMed Central PMCID: PMC7080720.
  60. 60. Bandelt HJ, Forster P, Rohl A. Median-joining networks for inferring intraspecific phylogenies. Mol Biol Evol. 1999;16(1):37–48. pmid:10331250.
  61. 61. Leigh JW, Bryant D, Nakagawa S. popart: full-feature software for haplotype network construction. Methods in Ecology and Evolution. 2015;6(9):1110–6.
  62. 62. Cabanettes F, Klopp C. D-GENIES: dot plot large genomes in an interactive, efficient and simple way. PeerJ. 2018;6:e4958. Epub 20180604. pmid:29888139; PubMed Central PMCID: PMC5991294.