Analysis of QTLs is important for facilitating breeding by isolating valuable alleles of the genes controlling agronomic important traits (Yano 2001). However, traditional QTL mapping is labor-intensive and time-consuming, requiring the generation of a large number of advanced generation progeny, polymorphic molecular marker screen, and marker genotyping. The QTL-seq method was first reported in rice in 2013, which employed the bulk segregant analysis and whole-genome resequencing for rapid and cost-effective identification of QTLs (Takagi, et al. 2013). By using the QTL-seq approach, advanced generation progeny or repeating backcrossing is not required for mapping population development. F2 populations containing hundreds of recombinant individuals were proofed for efficiently isolating QTLs, which greatly expedite the mapping process (Illa-Berenguer, et al. 2015; Lu, et al. 2014; Takagi, et al. 2013). Moreover, the number of polymorphic DNA markers is not a limiting factor due to DNA maker development, and marker genotyping is not required for QTL-seq (Illa-Berenguer, et al. 2015; Takagi, et al. 2013). Taking this advantage, QTL-seq is applicable for rice cultivars with a low level of DNA polymorphisms that are difficult for molecular marker development in conventional QTL mapping. In our study, HSS is a mutant derived from a tissue culture line of Nipponbare indicating that genetically closely related parents are suitable for the QTL-seq approach.
The QTL-seq method has been successfully applied to F2, F3, F4, and RIL (recombinant inbred lines) populations (Das, et al. 2015; Illa-Berenguer, et al. 2015; Lu, et al. 2014; Pandey, et al. 2017; Singh, et al. 2016; Takagi, et al. 2013; Wei, et al. 2016). RIL has a high degree of homozygosity and each individual can be treated as proxy clones, and thus is suitable for detecting QTLs for minor effects (Takagi, et al. 2013). While the F2 population is not suitable for detecting minor effect QTLs but quicker for population generation (Takagi, et al. 2013). Takagi et al. (2013) evaluated the power of F2 and RILs of F7 generation to detect a QTL and revealed that F7 RILs have higher codominance and complete dominance power than in F2 populations (Takagi, et al. 2013). In this study, we used F3 plants for QTL mapping. More background variations are fixed in F3 compared to F2 progeny and the selection of individuals with extreme phenotypic values was more solid as it was continuously confirmed from the two generations. The size of the mapping population is also critical for the successful implementation of QTL-seq. Usually, a population with hundreds of individuals is sufficient for conducting the QTL-seq approach, and only 20–50 plants with extreme opposite trait values for each bulk are needed (Das, et al. 2015; Lu, et al. 2014; Singh, et al. 2016; Takagi, et al. 2013; Wei, et al. 2016). A population harbors more recombinant F2 plants is desirable with the aim to clone the underlying gene (Illa-Berenguer, et al. 2015). And a larger population would result in a shorter interval of the genomic region of the gene (Illa-Berenguer, et al. 2015). On the contrary, the relatively smaller size of the F2 population is suitable for identifying closely linked markers without further gene identification (Illa-Berenguer, et al. 2015). In our study, we selected 75 early flowering and 78 late-flowering extremes from the F2 population with 532 plants. To double confirm the phenotype of selected individuals, approximately 1600 F3 plants were grown in the next year and, 37 earliest and 35 latest plants were selected for QTL-seq. These resulted in the identification of major QTL controlling flowering time located in an approximate of 8.3 Mb intervals. The efficiency to identify QTLs by QTL-seq approach can be influenced by the read depth at a specific locus and coverage rate of the genome as well. These are critical for detecting reliable variations and analyzing the contribution of each parent to the bulked DNA (Illa-Berenguer, et al. 2015). Rice has a compact genome of roughly 400Mb that allows read depth more than or equal to 20-fold of coverage to sufficiently detect variants by QTL-seq approach (Takagi, et al. 2013). We applied 30-fold coverage for whole-genome sequencing in this study that is highly effective to detect variants associated with quantitative traits of interest.
Genetic analysis indicated that Nipponbare might have an allele of Hd1 with a strong response to photoperiod sensitivity (Yano, et al. 2000). In this study, HSS has a weaker sensitivity to photoperiod. Relative mRNA expression analysis indicated that the function of the Hd1 allele in HSS was not completely lost but expressed at a lower level, especially during night time (Fig. 5 and Fig. 6). Therefore, HSS can be categorized as a knock-down-functional mutant at the Hd1 locus. Hd1 protein possesses a zinc finger domain and a CCT domain. The zinc finger domain is typically involved in nucleic acid binding (Jantz, et al. 2004), and the CCT domain functions as a nuclear localization signal (Robert, et al. 1998). Comparison of the sequences of HSS and Nipponbare revealed a single-base substitution and a 36bp insertion located at the C terminal region of the zinc finger domain located in the first exon. It is noteworthy that the sequence of the 36bp insertion induces the basic amino acid motif RRHQR to the zinc finger domain terminus of the Hd1 allele in HSS. This basic amino acid motif is conserved among rice Hd1, Arabidopsis CO, and Brassica napus BnCOA1 protein (Hd1 orthologs) downstream from the zinc finger (Robert, et al. 1998), and considered to be important for protein-DNA interaction (Omichinski, et al. 1993). The 36bp insertion also present in other rice cultivars of Zhonghua11(Luan, et al. 2009), Kasalath, Ginbouzu (Mo, et al., 2021; Yano, et al. 2000), Bengal, Cypress (Subudhi, et al. 2018), Kitaake (Gao, et al. 2013), and weedy rice of PSRR-1 (Subudhi, et al. 2018). It is more like a 36bp deletion that happened in Nipponbare during the evolutionary process. This allelic variation in Hd1 gene has not obtained enough attention and characterizing its function by genetic analysis is not easy, because other sequence variations also present in Hd1 except for the 36bp insertion in some cultivars, such as in Kasalath, Bengal and Cypress (Subudhi, et al. 2018; Yano, et al. 2000), that influence functions of Hd1. Moreover, allelic differences in other flowering time regulators obstruct studies on clarifying functions of the insertion. For example, apart from the 36bp insertion in Hd1 gene, Kitaake has an immature stop in Ghd7 compared to Nipponbare (Gao, et al. 2013). According to our result, the 36bp insertion in HSS could at least partially explain its insensitivity or Nipponbare’s sensitivity to day-length.
Hd1 is a major photoperiodic floral regulator in rice that promotes flowering in SD and represses flowering in LD by regulating the expression of Hd3a (Du, et al. 2017; Kojima, et al. 2002). Diurnal expression pattern analyses revealed a lower transcripts level of Hd1 in HSS under both SD and LD conditions compared to Nipponbare, especially during the dark period (Fig. 5 and Fig. 6). In HSS, the corresponding mRNA accumulation of Hd3a was lower in SD and higher in LD, resulting in phenotypic divergences between HSS and Nipponbare in response to different photoperiods. Recent studies demonstrated that Hd1 can physically interact with DTH8 forming the DTH8-Hd1 complex (Du, et al. 2017; Zhu, et al. 2017). DTH8 is an NF-YB transcription factor in rice which is homologous with the Arabidopsis HEME ACTIVATOR PROTEIN (YEAST) HOMOLOG 3 subunit of the CCAAT box-binding transcription factor (Du, et al. 2017; Matsubara, et al. 2014). Liu et al. (2020) observed dominance complementation of Hd1 and DTH8 resulted in extremely late flowering under LD condition. Du et al. (2017) elucidated that Hd1 represses Hd3a expression with a prerequisite of the presence of functional DTH8 under LD conditions, without DTH8, Hd1 plays an activator role in flowering (Du, et al. 2017). Similar results were observed in this current study and a previous report (Du, et al. 2017) that plants harboring weakened-functional or non-functional Hd1 allele, but with a functional DTH8 background, the Hd3a mRNA level was dramatically elevated under LD condition and brought forward flowering time. Moreover, under LD conditions, DTH8 represses Ehd1 and Hd3a and therefore suppresses flowering (Du, et al. 2017; Wei, et al. 2010; Yan, et al. 2011; Zhu, et al. 2017). While, DTH8 promotes flowering under SD conditions (Du, et al. 2017; Yan, et al. 2011). The dual role of DTH8 in flowering time regulation is functionally similar to Hd1, indicating that Hd1 and DTH8 may regulate the same pathway in photoperiodic flowering by forming the DTH8-Hd1 complex (Sun, et al., 2022). Wei et al. (2010) examined the expression of DTH8 at ZT0 in the NIL(hd1) line under LD conditions and found no changes of DTH8 expression level compared to WT (Wei, et al. 2010). However, in our study, a distinct difference was observed of DTH8 mRNA level between HSS (has weak functional allele of Hd1) and Nipponbare in both SD and LD via diurnal expression analyses from ZT0 to ZT20 (Fig. 5 and Fig. 6), and the changings of DTH8 expression were in a way following Hd1 in LD (Fig. 5). These results demonstrated that the Hd1 expression level could affect the transcripts of DTH8 in a positive manner (Fig. 7). Hd1 and DTH8 may regulate downstream genes, such as Hd3a, together in the form of the DTH8-Hd1 complex (Fig. 7). Additionally, it is possible that the abundance of the DTH8-Hd1 complex is affected by day length through regulating Hd1 protein abundance. Whether the basic amino acid motif (RRHQR) in HSS makes the interaction between Hd1 and DTH8 more closely and thus enables Hd1 to regulate DTH8 is worth further study. Moreover, we found that the pattern changes of the transcripts RFT1 in SD and LD, and Ehd3 in SD were also similar to the changes of Hd1 or DTH8. It is not clear whether RFT1 and Ehd3 can be regulated by Hd1 or DTH8 directly.