Keywords

10.1 Introduction

Flowering is the dramatic transition from the vegetative phase to reproductive development and is predominantly regulated by genetic control pathways that integrate internal and external signals. The ability of plant species to initiate flowering at the most favourable time for reproduction depends primarily on their accurate measurement of seasonal changes in day length and temperature (Thomas and Vince-Pure 1997; Song et al. 2015).

The flowering time (often termed the heading date) is important for regional adaptability and is easy to observe; therefore, its variations among rice varieties (Oryza sativa L.) have been known for a long time. Studies investigating the inheritance of rice flowering time date back to the 1910s. Hoshino (1915) suggested that multiple loci were involved in the inheritance of the flowering time based on the segregation patterns in progeny from experimental crosses between early- and late-flowering varieties. Since the 1920s, due to the development of the chromosome theory of inheritance and the discovery of genetic linkage, the linkage relationship between the flowering time genes and known loci (for other phenotypic traits) has been investigated (Chao 1928; Jodon 1940; Chandraratna 1953).

After the discovery of photoperiodism in plants (i.e. the response of an organism to the relative length of the day and night) by Garner and Allard (1920, 1923), many researchers have measured the flowering responses in rice varieties by day length. These studies revealed that rice is a facultative short-day (SD) plant as follows: its flowering is promoted under SD conditions, and the difference in the photoperiod response among rice varieties results in extensive variations in the flowering time (Vergara and Chang 1985).

The development of DNA markers in the 1990s dramatically enhanced the determination of the chromosomal location of genes/quantitative trait loci (QTLs) that are involved in the flowering time, and after decoding the genome sequences, map-based cloning strategies have facilitated the molecular cloning of these genes (Hori et al. 2016) (see also Chaps. 8 and 9 and those from Chaps. 11, 12, 13, 14, 15, 16, and 17).

In this chapter, we summarise the findings of many studies investigating rice flowering time using forward- and reverse-genetic approaches based on genomic information from rice. The studies performed over the last two decades have clearly shown that the use of a combination of these approaches has enhanced our understanding of the genetic control pathways of flowering in rice.

10.2 Mapping of the QTLs Responsible for Flowering Time

In the 1990s, the development of DNA markers allowed researchers to clarify the number and effects of the genes underlying the flowering time using QTL analyses (Li et al. 1995; Xiao et al. 1996; Yano et al. 1997). In particular, using several types of progeny derived from a single cross between the rice varieties ‘Nipponbare’ (ssp. japonica) and ‘Kasalath’ (ssp. indica), our group detected several QTLs responsible for the flowering time. Five of these QTLs (Hd1–Hd5) were mapped by performing a QTL analysis of an F2 population (Yano et al. 1997), and Hd7, Hd8, and Hd11 were detected in the BC1F5 lines (Lin et al. 1998). Other loci were detected only when advanced backcross progeny, such as BC3F2 or BC4F2, was used (Yamamoto et al. 2000; Lin et al. 2002). The results of these QTL mapping studies are summarised by Yano et al. (2001). Since then, many research groups have detected the QTLs for flowering time using different cross combinations. More comprehensive genetic analyses revealed that more than 100 loci, including major and minor effects, might be involved in flowering time control (Hori et al. 2016).

10.3 Molecular Cloning of Flowering Time Genes

Efforts detecting QTL above mentioned have led to the map-based cloning of the genes responsible for flowering time and improved our understanding of the function of these genes at the molecular level and the genetic pathways controlling flowering in rice.

Hd1 was the first rice flowering time gene cloned using natural variation (Table 10.1; Yano et al. 2000). Using more than 9000 recombinants, we defined the Hd1 region within 12 kb on chromosome 6. This region contained one candidate gene with a high similarity to Arabidopsis CONSTANS (CO). Comparison of the candidate gene in ‘Nipponbare’ and ‘Kasalath’ revealed many sequence variations, including a 36-bp insertion and a 33-bp deletion (in exon 1) and a 2-bp deletion (in exon 2) in ‘Kasalath’. A small genomic fragment of ‘Nipponbare’ containing the Hd1 candidate gene was transferred into the near isogenic line of ‘Nipponbare’ carrying Hd1 from ‘Kasalath’ and was found to promote flowering under SD conditions. These results clearly indicated that the candidate gene homologous to Arabidopsis CO was Hd1.

Table 10.1 Molecularly cloned genes underlying the flowering time of rice
Full size table

At least 14 flowering time QTLs have been isolated using map-based cloning strategies assessing natural variation (Yano et al. 2000; Takahashi et al. 2001; Kojima et al. 2002; Doi et al. 2004; Xue et al. 2008; Wei et al. 2010; Bian et al. 2011; Matsubara et al. 2012; Gao et al. 2013; Hori et al. 2013; Koo et al. 2013; Ogiso-Tanaka et al. 2013; Wu et al. 2013; Shibaya et al. 2016). Rice mutants have also been used to isolate flowering time genes and investigate their functions (Izawa et al. 2000; Lee et al. 2004; Matsubara et al. 2008, 2011; Saito et al. 2012; Dai and Xue 2010; Yang et al. 2013; Yokoo et al. 2014; Yoshitake et al. 2015). Other flowering time genes have been identified, and their functions have been investigated using forward- and reverse-genetic approaches, such as overexpression or knockdown of a target gene. For example, the functions of RFT1 and OsTrx1 were revealed by knocking down these genes, whereas the functions of OsMADS50, OsMADS56, and OsMADS15 were verified by overexpressing these genes (Komiya et al. 2008; Ryu et al. 2009; Lu et al. 2012; Choi et al. 2014).

10.4 Genetic Pathways Controlling Flowering Revealed by Molecular Cloning

Under SD conditions, rice flowering is promoted by the expression of Hd3a, which is activated by Hd1 and Ehd1 (Table 10.1; Yano et al. 2000; Kojima et al. 2002; Doi et al. 2004; Tamaki et al. 2007) (Fig. 10.1). Hd3a acts as a mobile flowering signal (florigen) (Tamaki et al. 2007). The expression of Ehd1 is upregulated by DTH3, Ehd2, Ehd3, Ehd4, and OsMADS51 (Kim et al. 2007; Matsubara et al. 2008, 2011; Bian et al. 2011; Gao et al. 2013) and downregulated by OsCOL4 and OsCOL10 (Lee et al. 2010; Tan et al. 2016).

Fig. 10.1
figure 1

A schematic representation of the genetic pathways controlling flowering in rice. The clocks at the top show the circadian clock. Genes with no obvious Arabidopsis counterparts are shown in red. Orange ovals show physical interactions between genes. Genes with a natural allelic variation are underlined. X indicates an unknown gene. SD short-day conditions, LD long-day conditions. Arrows upregulation, bars downregulation. Phys phytochromes

Full size image

The transcriptional activation of Hd3a is lower under long-day (LD) conditions than under SD conditions; consequently, flowering is suppressed (Fig. 10.1). Although Hd1 activates the expression of Hd3a under SD conditions, Hd1 represses the expression of Hd3a under LD conditions (Fig. 10.1). This functional conversion of Hd1 is caused by phytochrome-mediated signalling (Hayama and Coupland 2004; Izawa 2007). The Hd1 repressor function under LD conditions is enhanced by the kinase activity of Hd6 and is mediated by unknown genes (Takahashi et al. 2001; Ogiso et al. 2010). In addition to Hd1, Ghd7 also represses the expression of Hd3a by repressing Ehd1 under LD conditions (Xue et al. 2008). Based on genetic analyses, it was originally believed that rice photoperiodic flowering is controlled by the following two independent signalling pathways: the Hd1–Hd3a pathway, which is evolutionarily related to the Arabidopsis CO–FT pathway, and the Ghd7-Ehd1–Hd3a pathway, which has no Arabidopsis counterpart (Doi et al. 2004; Xue et al. 2008). However, a physical interaction was recently demonstrated between Hd1 and Ghd7 in vivo (Nemoto et al. 2016). The protein complex of Hd1 and Ghd7 specifically binds to a cis-regulatory region in Ehd1 and represses its expression, suggesting that the two pathways are integrated into Ehd1 and repress flowering under LD conditions (Fig. 10.1).

Due to the progress in our understanding of the core pathways, many genes underlying rice flowering under LD conditions have been discovered during the last decade (Fig. 10.1). RFT1, which is located within 11.5 kb of Hd3a, is an Hd3a paralog (Kojima et al. 2002; Komiya et al. 2008). The expression of RFT1 increases under LD conditions, and RFT1 moves from the leaves to the shoot apical meristem, indicating that the control of the flowering time in rice involves two florigen genes, Hd3a and RFT1, under LD conditions (Komiya et al. 2008, 2009). The expression of RFT1 is promoted by Ehd1 and DTH2 but is repressed by Se14 (Doi et al. 2004; Wu et al. 2013; Yokoo et al. 2014). The expression of Ehd1 is induced by DTH3, OsMADS51, OsMADS56, Ehd2, and Ehd4 (Kim et al. 2007; Matsubara et al. 2008; Ryu et al. 2009; Bian et al. 2011; Gao et al. 2013) but is repressed by DTH8, OsCOL4, OsCOL10, OsLFL1, and OsPRR37 (Peng et al. 2008; Lee et al. 2010; Wei et al. 2010; Yan et al. 2011; Gao et al. 2014; Tan et al. 2016). Recently, DTH8 has been reported to form a complex with Hd1 to control flowering (Chen et al. 2014; Zhu et al. 2017). DTH3 and OsMADS56 form a complex that regulates Ehd1 (Ryu et al. 2009). OsMADS51 is upregulated by Hd18 and induces the expression of Ehd1 (Kim et al. 2007; Shibaya et al. 2016). OsLFL1 is induced by OsVIL2 and has been proposed to downregulate the expression of Ehd1 (Peng et al. 2008; Yang et al. 2013). Ehd2 is downregulated by Se13 and induces the expression of Ehd1 (Matsubara et al. 2008; Yoshitake et al. 2015). OsCOL4 is a constitutive repressor upstream of Ehd1 (Lee et al. 2010). OsCOL10 downregulates the expression of Ehd1 and is upregulated by Ghd7 (Tan et al. 2016). The expression of Ghd7 is induced by Hd17 and is repressed by the Ehd3 and OsTrx1 complex (Choi et al. 2014). Ghd7 activity is increased by phosphorylation by Hd16 (Hori et al. 2013). Hd16 also phosphorylates OsPRR37, which represses the expression of Hd3a either directly or through Ehd1 (Hori et al. 2013; Koo et al. 2013; Gao et al. 2014).

Thus, under LD conditions, most flowering time gene signals (by both repressors and promoters) are transmitted to rice florigen genes through Ehd1 in flowering rice (Fig. 10.1).

The shared regulation by flowering time genes downstream of Hd3a and RFT1 is an underlying mechanism under both SD and LD conditions (Fig. 10.1). Hd3a interacts with GF14C, and then the Hd3a-GF14C complex interacts with OsFD (Taoka et al. 2011; Tsuji et al. 2013). The resultant protein complex induces the expression of floral meristem identity genes (e.g. OsMADS15) to initiate the floral transition in the shoot apex.

An additional description of the genes described above is provided in Table 10.1.

10.5 Circadian Clock Genes Control the Expression of Flowering Time Genes

The expression of many rice flowering time genes depends on the day length (Izawa 2007; Itoh et al. 2010). For example, Hd3a and Ehd1 are expressed in the morning under SD conditions, whereas Ghd7 is expressed in the morning under LD conditions. The expression of OsGI shows daily circadian oscillations with a peak at the end of the light period, and the expression of OsGI is regulated by the circadian clock and activates the expression of Hd1 (Table 10.1; Hayama et al. 2002, 2003) (Fig. 10.1). OsGI also activates the expression of Ehd1 either directly or via OsMADS51 (Kim et al. 2007; Itoh et al. 2010). The expression levels of certain flowering time genes are regulated by the circadian gating of light responses through phytochromes (red-light receptors) and cryptochromes (blue-light receptors) (Itoh et al. 2010). The expression of Ehd1 is induced by blue light in an OsGI-dependent manner regardless of day length; however, the expression is repressed by Ghd7 under LD conditions. The expression of Ghd7 is induced by phytochrome signalling, and the sensitivity to red light is gated at the beginning of the light period under LD conditions.

Recently, Matsuzaki et al. (2015) developed a statistical model of the expression of multiple genes with phase setting by sunlight and the circadian clock under field conditions. The integration of the expression patterns of individual flowering time genes can accurately estimate the internal biological time determined by both the circadian clock and the actual physical time of day.

To date, the control of flowering by the circadian clock in rice remains largely unknown compared to that in Arabidopsis likely because experiments in the laboratory are difficult to perform due to its large plant size. However, field experiments, such as those performed by Matsuzaki et al. (2015), will provide a better understanding of the role of the circadian clock in rice flowering.

10.6 Genetic Architecture of the Natural Variations in Flowering Time

To clarify the natural allelic variations in flowering time, we carried out QTL analyses in 12 F2 populations derived from crosses of ‘Koshihikari’ (ssp. japonica), which is an elite Japanese variety that is commonly used as a parental line, with varieties originating in various regions in Asia (Ebana et al. 2011; Shibaya et al. 2011). A limited number of loci with large effects that corresponded to Hd1, Hd2, Hd6, RFT1, Ghd7, DTH8, and Hd16 accounted for some varietal differences, but additional QTLs are likely to be involved in the flowering variation in these populations.

To detect QTLs with small effects, we analysed advanced backcross progeny derived from each cross combination by Ebana et al. (2011) and Shibaya et al. (2011) and detected a total of 255 QTLs widely distributed across the genome (Hori et al. 2015). We detected 128 QTLs with a relatively large effect, which corresponded to the genomic positions of previously detected flowering time genes, such as Hd1, Hd2, Hd6, RFT1, Ghd7, DTH8, and Hd16. The sequence analyses revealed that the chromosomal positions of the large-effect QTLs mainly corresponded to those of different alleles of the flowering time genes in 12 rice varieties. The other 127 QTLs were detected in chromosomal regions other than those of the flowering time genes and had relatively small effects. These results indicate that much of the variation in the flowering time can be explained by combinations of alleles in large- and small-effect QTLs.

Genome-wide association studies have also supported the hypothesis that allelic variations at multiple QTLs play an important role in the differences in the flowering time among rice varieties (Zhao et al. 2011; Huang et al. 2012; Yano et al. 2016). Zhao et al. (2011) detected ten genomic regions that were significantly associated with the flowering time variation, although only Hd1 was detected as a major QTL. These genomic regions explained less than 50% of the flowering time variation. Huang et al. (2012) found 14 significant genomic regions: 5 regions surrounding Hd1, Ghd7, RCN1, OsGI, and Hd3a and 9 newly discovered regions. The detected regions explained 36% of the flowering time variation. More recently, a genome-wide association study revealed that two novel QTLs on chromosomes 1 and 11 contributed to the flowering time variation in japonica rice varieties (Yano et al. 2016). The above-mentioned studies suggest that QTLs that have not yet been discovered are associated with the natural variation in the flowering time in rice varieties.

10.7 Regional Adaptation Based on Allelic Differences in Flowering Time Genes

Early flowering conferred by deficient alleles in the flowering time genes is important for expanding the range of rice cultivation to high latitudes (LD conditions) (Izawa 2007; Shrestha et al. 2014), where early heading and maturity are required for seed production. A sequence analysis of the known flowering time genes, including Hd1, Ghd7, DTH8, Hd16, OsPRR37, DTH2, and Ehd4, indicated that allelic differences contribute to regional adaptation (Takahashi et al. 2009; Naranjo et al. 2014; Gómez-Ariza et al. 2015; Zheng et al. 2015; Goretti et al. 2017). The functional alleles of Hd1 are associated with late flowering, and its non-functional alleles are associated with early flowering under natural day-length conditions; the geographical distribution of the Hd1 alleles suggests that favourable alleles have been selected by breeders to enhance rice productivity and adaptability in each region (Takahashi et al. 2009; Fujino et al. 2010; Ebana et al. 2011; Takahashi and Shimamoto 2011; Naranjo et al. 2014; Gómez-Ariza et al. 2015; Goretti et al. 2017). The deficient or weak alleles of Ghd7, DTH8, DTH2, Hd16, and OsPRR37 are distributed in cultivation areas at high latitudes (Xue et al. 2008; Wei et al. 2010; Fujino et al. 2013; Hori et al. 2013; Koo et al. 2013; Wu et al. 2013; Kwon et al. 2014; Goretti et al. 2017), suggesting that these alleles must be involved in the expansion of rice cultivation areas.

10.8 Conclusions and Perspectives

During the last two decades, tremendous progress in genome sequencing has improved our understanding of the genetic and molecular mechanisms that control the flowering time in rice. For example, Ehd1 is an important integrator in genetic control pathway, the putative homolog has found in sorghum (SD plant); on the other hand, it has not been found in Arabidopsis and wheat (LD plants) (Brambilla et al. 2017). Such a finding in rice provides valuable suggestion about the genetic control of flowering time in sorghum; additionally, it plays an important role in understanding of the diversity and evolution of flowering time control in plants.

This progress was due not only to genomic approaches, such as QTL analyses and map-based cloning, but also to the large number of rice accessions (including wild relatives) and genetic mapping populations derived from artificial crosses. Most genes with major effects on flowering time have already been identified during the last 20 years. Although some genes with minor effects have also been analysed, additional new QTLs with minor effects need to be further examined. In fact, genome-wide association studies could be an effective method for detecting new chromosomal regions (QTLs) responsible for the flowering time (Yano et al. 2016). The verification of the allelic effects of these QTLs with minor effects must be performed using experimental populations derived from single crosses. This approach will lead a more in-depth understanding of the genetic control of flowering time in rice. Information derived from these analyses has also been applied to the modulation of the flowering time of cultivars for regional adaptation and cropping system (Takeuchi et al. 2006; Takeuchi 2011; Hori et al. 2016).

Recently, genome-wide prediction models of the flowering time in rice have been tested and demonstrated high prediction accuracy by adding environmental variables (Nakagawa et al. 2005; Onogi et al. 2016; Spindel et al. 2016). Furthermore, it appears that these models can predict the flowering time using genome-wide genotypes and trait values in various types of populations, such as experimental biparental populations or an array of varieties (Spindel et al. 2016). Further progress using these approaches will enhance the fine-tuning of flowering time in rice breeding programmes.