Quantitative trait locus analysis for flowering-related traits using two F 2 populations derived from crosses between Japanese common wheat cultivars and synthetic hexaploids

Flowering time is an important trait for Japanese wheat breeding. Aegilops tauschii, the D-genome donor of hexaploid wheat, is a useful resource to enlarge the D-genome diversity of common wheat. Previously, we identified floweringrelated QTLs in F2 populations of synthetic hexaploid wheat lines between the tetraploid wheat cultivar Langdon and Ae. tauschii accessions. Here, to evaluate the usefulness of the early-flowering alleles from Ae. tauschii for Japanese wheat breeding, QTL analyses were conducted in two F2 populations derived from crosses between Japanese wheat cultivars and early-flowering lines of synthetic hexaploid wheat. Only two chromosomal regions controlling flowering-related traits were identified, on chromosomes 2DS and 5AL in the mapping populations, and no previously identified QTLs were found in the synthetic hexaploid lines. The strong effect of the 2DS QTL, putatively corresponding to Ppd-D1, was considered to hide any significant expression of other QTLs with small effects on flowering-related traits. When F2 individuals carrying Ae. tauschii-homozygous alleles around the 2DS QTL region were selected, the Ae. tauschii-derived alleles of the previously identified flowering QTLs partly showed an early-flowering phenotype compared with the Japanese wheat-derived alleles. Thus, some early-flowering alleles from Ae. tauschii may be useful for production of early-flowering Japanese wheat cultivars.


INTRODUCTION
Flowering time (FLT) determination is one of the most important events for higher plant adaptation to regional growth habitats.Common wheat cultivars show wide variation in heading time (HT) and FLT, and control of HT and FLT has been a target for wheat breeding (Snape et al., 2001).Early-maturing wheat cultivars are required, particularly in Japan, because the rainy season overlaps with the season of wheat maturation.High moisture at maturation time (MAT) results in degradation of wheat grain quality through preharvest sprouting, Fusarium damage, and so on.
Wheat HT and FLT are controlled by three major genetic components, vernalization requirement, photope-riodic sensitivity and narrow-sense earliness (Murai et al., 2005).A few major genes control the vernalization requirement and photoperiodic sensitivity, whereas some minor genes have been identified as quantitative trait loci (QTLs) controlling narrow-sense earliness (Cockram et al., 2007).The vernalization requirement and photoperiodic sensitivity are mainly determined by the Vrn-1 loci on the long arms of homoeologous group 5 chromosomes and the Ppd-1 homoeologous loci on the short arms of group 2 chromosomes, respectively (Murai et al., 2005).Most Japanese wheat cultivars, except those in the Hokkaido area, carry the Ppd-D1a allele on 2DS, which contributes to their early-flowering phenotype (Seki et al., 2011;Nguyen et al., 2013a).The Ppd-D1a mutant allele is photoperiod-insensitive because of a 2-kb deletion upstream of the coding region in a pseudo-response regulator gene (Beales et al., 2007;Nishida et al., 2013).Most spring-type landraces of southwest Japan carry a domi-nant Vrn-D1 allele on 5DL (Iwaki et al., 2001), and the spring habit caused by this allele shows slightly shorter narrow-sense earliness (Kato et al., 2001).Structural mutations at the Vrn-1 loci, such as insertion/deletion events in the promoter region and large deletions in the first intron, generate dominant alleles in spring-type cultivars of einkorn, durum and common wheat (Yan et al., 2004;Fu et al., 2005).Moreover, Vrn-D4 is known as another vernalization gene located in the centromeric region of chromosome 5D in common wheat, and wheat varieties with the dominant allele of Vrn-D4 show the spring growth habit (Yoshida et al., 2010).
To enrich D-genome variation in common wheat, synthetic hexaploid wheat lines produced through hybridization of tetraploid wheat and Aegilops tauschii are useful resources (Mujeeb-Kazi et al., 1996;Trethowan and Mujeeb-Kazi, 2008;Jones et al., 2013).Several agronomically important traits such as disease resistance have been transferred from Ae. tauschii to common wheat through synthetic hexaploids (Kerber, 1987;Ma et al., 1995;Mujeeb-Kazi et al., 1996).Our empirical studies using a series of wheat synthetics have also shown enriched diversity of the D genome for some morphological and physiological characters (Takumi et al., 2009;Kajimura et al., 2011;Iehisa and Takumi, 2012;Okamoto et al., 2012).The wide variation in HT observed among Ae.tauschii accessions is retained in the synthetic hexaploids (Kajimura et al., 2011).Early heading and flowering in synthetic hexaploids are presumably associated with shortened narrow-sense earliness due to effects of the added D-genome (Fujiwara et al., 2010).Recently, we selected three early-flowering lines from the series of wheat synthetics and determined QTLs for floweringrelated traits in four mapping populations of the synthetic wheat lines with distinct D-genome origins (Nguyen et al., 2013b).These QTLs significantly contribute to the variation in flowering-related traits among the synthetic wheat lines, implying that D-genome QTLs might be at least partly effective in the hexaploid wheat background.However, their utility for determination of flowering-related traits remains unclear in progeny crossed with Japanese common wheat cultivars.The Ae. tauschii populations used as the direct D-genome donors for common wheat are considered to be distributed in a narrow range relative to the entire species range (Feldman, 2001), suggesting that the huge genetic diversity of Ae. tauschii is not necessarily represented in common wheat and that the early-flowering lines of wheat synthetics should include early-flowering D-genome alleles that have not been introduced into the D-genome of Japanese common wheat cultivars (Matsuoka et al., 2013;Wang et al., 2013).Here, we conducted QTL analyses for floweringrelated traits using two F 2 populations derived from crosses between Japanese common wheat cultivars and two selected lines of the early-flowering wheat synthetics to examine whether the D-genome QTLs from Ae. tauschii are responsible for flowering-related trait variations in the cross progeny.Based on the results, we discuss the usefulness of the synthetic wheat lines for wheat breeding.

MATERIALS AND METHODS
Plant materials Two cultivars of common wheat (Triticum aestivum L.) from Japan, Kitanokaori and Norin 61, and two synthetic hexaploid wheat lines were used in this study.Kitanokaori is a winter-type elite cultivar of bread wheat grown on Hokkaido, and Norin 61 is a standard cultivar grown in the central to southwestern parts of Honshu.Synthetic wheat lines showing an earlyflowering phenotype were produced from two cross combinations between a tetraploid wheat variety, Triticum turgidum ssp.durum (Desf.)Husn.cv.Langdon (Ldn), and two Aegilops tauschii Coss.accessions, KU-2097 and PI476874 (Kajimura et al., 2011;Nguyen et al., 2013b).These two synthetic wheat lines, Ldn/KU-2097 and Ldn/ PI476874, were selected from the 82 wheat synthetic lines derived from crosses between Ldn and various Ae.tauschii accessions (Fujiwara et al., 2010;Kajimura et al., 2011) as showing early flowering.Ldn contains a dominant allele of Vrn-A1 for the spring habit and no Ppd-1 dominant allele for photoperiodic insensitivity (Fu et al., 2005;Fujiwara et al., 2010).Two F 2 populations were derived from crosses between Kitanokaori and Ldn/KU-2097 and between Norin 61 and Ldn/PI476874.F 2 seeds in the Kitanokaori//Ldn/KU-2097 population (N = 132) and the Norin 61//Ldn/PI476874 population (N = 108) were sown in November and respectively grown in the 2012-2013 and 2010-2011 seasons.From each of the 10 F 2 plants of Kitanokaori//Ldn/KU-2097, 16 selfed seeds (F 3 generation) were sown in November 2013 and grown in the 2013-2014 season.All F 2 and F 3 individuals as well as four plants of each parent were grown individually in pots arranged randomly in a glasshouse of Kobe University as previously reported (Kajimura et al., 2011).

Phenotype measurement and statistical analyses
Four flowering-related traits were measured in the field.HT and FLT were recorded as days after sowing (Kajimura et al., 2011).MAT was measured as the number of days that had passed before the peduncle turned yellow (Kajimura et al., 2011).The grain-filling period (GFP) was the number of days from flowering to maturation.HT, FLT and MAT were measured for the three earliest tillers of each plant, and mean values were calculated using the data for each F 2 and F 3 plant.The data were statistically analyzed using JMP software ver.5.1.2(SAS Institute, Cary, NC, USA).Pearson's correlation coefficients were estimated among the traits measured in each mapping population.

Detection of polymorphisms and genotyping with markers
To amplify PCR fragments of simple sequence repeat (SSR) markers, total DNA was extracted from the parents and F 2 individuals using standard procedures.SSR genotyping was performed according to our previous report (Nguyen et al., 2013b).Information on SSR markers and their respective annealing temperatures was obtained from the NBRP KOMUGI website (http://www.shigen.nig.ac.jp/wheat/komugi/strains/aboutNbrpMarker.jsp) and the GrainGenes website (http://wheat.pw.usda.gov/GG2/maps.shtml).Polymorphisms at the Ppd-D1 and Vrn-A1 loci were detected using allele-specific primers (Fu et al., 2005;Beales et al., 2007;Nguyen et al., 2013aNguyen et al., , 2013b)).The PCR products were separated by electrophoresis through a 1.2% agarose gel and stained with ethidium bromide.
For map construction of chromosome 2D in the Norin 61//Ldn/PI476874 population, we used genotyping data from our previous report (Iehisa et al., 2014).In these genotyping data were included two EST-derived sequence-tagged site markers, TE6 and WE6, that were developed with the Iw2-linked markers (Liu et al., 2007;Nishijima et al., 2014).Some single nucleotide polymorphism (SNP) markers, which were converted to cleaved amplified polymorphic sequence (CAPS) or high resolution melting (HRM) markers, were also included to enrich the number of markers.The SNP information was originally obtained from our D-genome SNP database based on RNA-seq data for Ae.tauschii (Iehisa et al., 2012(Iehisa et al., , 2014)).PCR and analysis of the CAPS and HRM markers were performed according to our previous studies (Iehisa et al., 2012;Matsuda et al., 2012).

Map construction and QTL analysis
Polymorphic markers of the parents were genotyped and used for map construction.Genetic mapping was performed using MAPMAKER/EXP software ver.3.0b (Lander et al., 1987).The threshold value for log-likelihood (LOD) scores was set at 2.5, and the genetic distances were calculated using the Kosambi mapping function (Kosambi, 1944).Chromosomal assignment of SSR markers was generally based on reported reference maps (Somers et al., 2004;Torada et al., 2006;Kobayashi et al., 2010).
QTL analyses were carried out by composite interval mapping using Windows QTL Cartographer software ver.2.5 (Wang et al., 2011) with the forward and backward method.A LOD score threshold for each trait was determined by computing a 1,000-permutation test.The percentage of phenotypic variation explained by a QTL for a trait and any additive effects were also estimated using this software.

Flowering-related traits in two F 2 populations and their parental accessions
The mean values of the parental lines involved in each mapping population differed for flowering-related traits (Table 1).HT, FLT and MAT were significantly shorter in Ldn/KU-2097 than in Kitanokaori.Norin 61 showed significantly earlier attributes for HT, FLT and MAT than Ldn/PI476874, whereas GFP values were significantly shorter in Ldn/PI476874 than in Norin 61.Little variation in the four traits was observed among individual plants of each parental line, but all four traits varied widely in the F 2 populations.Much earlier-and later-heading F 2 individuals were present than in the parental lines, especially in the Kitanokaori//Ldn/KU-2097 population, which suggested transgressive segregation.
Significant (P < 0.001) positive correlations were observed among HT, FLT and MAT in the two F 2 populations (Table 2).However, GFP was negatively correlated with HT, FLT and MAT in the two F 2 populations.The negative correlation between GFP and the other three traits indicated that the earlier-flowering F 2 individuals required a longer period for grain maturation, and that GFP of the late-flowering F 2 individuals tended to be shorter.
Construction of a linkage map and QTL analysis for flowering-related traits In the Kitanokaori//Ldn/ KU-2097 population, 914 primer sets were tested and 194 (21.2%) were found to be polymorphic.Of these, 155 markers including Iw2, B1 and Vrn-A1 formed 22 linkage groups.The total map length was 1929.0 cM with an average spacing of 12.4 cM between markers.In the Norin 61//Ldn/PI476874 population, 270 (28.2%) of the 958 tested primer sets were polymorphic between the parental lines.In total, 241 loci including Ppd-D1 were available for construction of a genetic map with 26 linkage groups.The total map length was 3117.0 cM with an average distance of 12.9 cM between markers.
In total, seven QTLs for flowering-related traits were detected on chromosomes 2D and 5A on the Kitanokaori// Ldn/KU-2097 map (Table 3).Of these, four QTLs for HT, FLT, MAT and GFP,with LOD scores of 12.20,14.20,9.56 and 4.24, respectively, were found in the same chromosomal region of 2DS (Fig. 1A).These 2DS QTLs explained 16.69 to 46.56% of the phenotypic variation for flowering-related traits.The remaining three QTLs for HT, FLT and MAT, with LOD scores of 9.00, 9.00 and 6.55, respectively, were assigned to the Vrn-A1 region of 5AL.The 5A QTLs contributed 19.25, 20.44 and 18.65% of the HT, FLT and MAT variation in the Kitanokaori// Ldn/KU-2097 population.
For all 2DS QTLs identified for HT, FLT and MAT, alleles from the Japanese wheat cultivars showed negative values when assessed for additive effects (Table 3), indicating that the alleles for these QTLs that are derived from the Japanese wheat cultivars contribute to early phenotypes for HT, FLT and MAT.For the 5A QTLs for HT, FLT and MAT, alleles from the Japanese wheat cultivar showed positive values for additive effects, indicating that the alleles for these QTLs that are from the synthetic wheat line contribute to early phenotypes.The GFP QTL on chromosome 2DS showed a positive value for additive effect.Thus, the GFP allele for prolonged filling period was derived from the Japanese wheat cultivars for the 2DS QTL.
Effects of known QTLs on flowering-related traits Our previous QTL analyses using the F 2 mapping populations of synthetic wheat lines identified QTLs for flowering-related traits on chromosomes 7D in the Ldn/ PI476874//Ldn/KU-2069 population and 1D, 5D and 6D in the Ldn/KU-2097//Ldn/IG126387 population (Nguyen et al., 2013b).However, we failed to identify the 1D, 5D, 6D or 7D QTLs in the Kitanokaori//Ldn/KU-2097 or Norin 61//Ldn/PI476874 populations.Therefore, data for each flowering-related trait were grouped based on the genotype around the reported QTL regions of each F 2 individual to study the effects of the QTLs identified previously.No significant differences were observed among genotypes around the 1D, 5D and 6D QTLs for HT in the Kitanokaori//Ldn/KU-2097 population (Fig. 2A).A similar result was obtained for the 7D QTL in the Norin 61// Ldn/PI476874 population (data not shown).
To mask the large effects of 2DS QTLs identified here on flowering-related traits, F 2 individuals carrying the Ldn/KU-2097-homozygous (photoperiod-sensitive) and Ldn/PI476874-homozygous (photoperiod-sensitive) alleles around the 2DS QTL region were selected and the genotypic effects of the 1D, 5D, 6D and 7D QTLs for HT were compared based on the genotype around the reported QTL regions of each selected F 2 individual.Around the 6D QTL, the selected F 2 individuals with heterozygous and Ldn/KU-2097-homozygous alleles showed significantly (P < 0.05) earlier HT than those with Kitanokaori-homozygous alleles (Fig. 2B).The F 2 individuals carrying Ldn/KU-2097-homozygous alleles around the 1D and 5D QTLs for HT appeared to display earlier HT attributes than those with the Kitanokaori-homozygous allele, although these differences were not significant (Fig. 2B).In the Norin 61//Ldn/PI476874 population, no significant differences in flowering-related traits were observed among genotypes around the 7D QTL.
We then chose 10 F 2 individuals with Ldn/KU-2097homozygous alleles around the 2DS QTL region from the Kitanokaori//Ldn/KU-2097 population, and the effects of the 1D, 5D and 6D QTLs for HT on flowering-related traits were compared in their selfed progeny (F 3 generation).The mean values of each F 2:3 plant varied for the four flowering-related traits (Table 4).F 2:3 plants with Ldn/KU-2097-homozygous alleles around the 5D QTL for HT tended to display earlier attributes for HT, FLT and MAT than other F 2:3 progeny or Kitanokaori.

DISCUSSION
Allohexaploidization between the tetraploid wheat (AABB) and Ae.tauschii (DD) genomes affects gene expression profiles in synthetic hexaploid wheat (Pumphrey et al., 2009;Chagué et al., 2010).This alteration of gene expression is considered to be due to epistatic interaction between the AB and D genomes and genetic and epigenetic modification of the chromatin structure (Khasdan et al., 2010;Mestiri et al., 2010;Feldman et al., 2012).For control of flowering-related traits, the Ppd-1 and Vrn-1 genotypes of the A and B genomes largely affect the expression of flowering-related traits in synthetic hexaploid wheat (Gororo et al., 2001).On the other hand, when Ldn was used as the single AB donor parent for production of synthetic wheat, D-genome variation for some morphological and physiological traits including flowering-related traits in Ae. tauschii was at least partly expressed in the hexaploid background (Kajimura et al., 2011;Iehisa and Takumi, 2012;Okamoto et al., 2012).In the present study, QTL analyses revealed only two chromosomal regions controlling flowering-related traits in each mapping population, and failed to find any QTL previously identified using mapping populations of the synthetic hexaploid wheat lines.The chromosomal location of the QTLs identified on 2DS corresponds to that of Ppd-D1 in both the Kitanokaori//Ldn/KU-2097 and Norin 61//Ldn/PI476874 populations (Fig. 1).At the centromeric region of barley chromosome 2H, EPS2, a barley homolog of Antirrhinum CENTRORADIALIS and Arabidopsis TERMINAL FLOWER1, is located as a flowering time regulator (Comadran et al., 2012).The 2DS QTL position was much more closely linked to the Ppd-D1 locus than was the EPS2 centromeric region.The Ppd-D1 dominant allele is frequently found in Japanese wheat cultivars, and contributes to their early flowering phenotype (Seki et al., 2011;Nguyen et al., 2013a).In the Kitanokaori//Ldn/KU-2097 and Norin 61//Ldn/ PI476874 populations, the 2DS QTLs showed high LOD scores and explained large amounts of the phenotypic variation for flowering-related traits (Table 3).The strong effects of the 2DS QTL could hide significant expression of other QTLs with weak effects on floweringrelated traits.In fact, when F 2 individuals with the Ldn/ KU-2097-homozygous allele around the Ppd-D1 region were selected, the 6D QTL for HT on the D genome, identified previously in the synthetic wheat population (Nguyen et al., 2013b), significantly affected HT in the F 2 progeny of Kitanokaori//Ldn/KU-2097 (Fig. 2).In addition, the Ldn/KU-2097-homozygous allele around the 5D QTL tended to cause earlier HT, FLT and MAT than the Kitanokaori-homozygous allele under ppd-D1 recessive conditions (Table 4).On the other hand, we failed to detect statistically significant effects of the 1D and 5D QTLs on HT in the Kitanokaori//Ldn/KU-2097 population and of the 7D QTLs on the flowering-related traits in the Norin 61//Ldn/PI476874 population even when F 2 individuals with the Ldn/KU-2097-homozygous allele and the Ldn/PI476874-homozygous allele around the Ppd-D1 region were selected.Because the number of available F 2 and F 2:3 plants was limited due to the restricted size of the F 2 populations, precise effects of the three D-genome QTLs on HT should be evaluated in future studies using near-isogenic lines with the genetic background of Japanese wheat cultivars.
The negative correlations between MAT and GFP were significant in both the Kitanokaori//Ldn/KU-2097 and Norin 61//Ldn/PI476874 populations (Table 2).A similar negative correlation between MAT and GFP was previously reported in synthetic hexaploid wheat lines (Kajimura et al., 2011), whereas positive correlations between MAT and GFP were detected in four F 2 mapping populations of synthetic hexaploids and two F 2 mapping populations of common wheat (Nguyen et al., 2013a(Nguyen et al., , 2013b)).In the Kitanokaori//Ldn/KU-2097 and Norin 61// Ldn/PI476874 populations, the strong effect of the Ppd-D1 region on GFP might have been responsible for the negative correlations.As previously reported, GFP can be influenced by temperature in wheat (Wiegand and Cuellar, 1981;Knott and Gebeyehou, 1987).The grain-filling process seemed to progress more quickly under higher temperatures in the late-flowering F 2 individuals with the ppd-D1 allele than in the early-flowering individuals with the Ppd-D1 allele in both populations in the present study.
The 5AL QTLs detected in the Kitanokaori//Ldn/KU-2097 population appeared to correspond to Vrn-A1 (Fig. 1A).Ldn contains a spring allele with a large deletion in the first intron of Vrn-A1 (Fu et al., 2005), whereas Kitanokaori, a winter-type cultivar, putatively has a recessive allele with no deletions in the promoter or first intron regions of Vrn-A1.Thus, the Vrn-A1 dominant allele from Ldn could contribute to earliness for HT and FLT in the mapping population.This suggestion supports previous data showing slightly shorter narrowsense earliness for the spring allele of Vrn-D1 (Kato et al., 2001).In addition, a phytochrome gene, PHYC, acts as one of the major flowering accelerators under long-day conditions in tetraploid wheat, and the PHYC loci are linked to the Vrn-1 loci in wheat and barley (Chen et al., 2014).An allelic difference of PHYC might contribute to the HT and FLT variations in the mapping population, although little information about natural variation in PHYC has been reported in common wheat.The relationship between the 5A QTL and PHYC in the Kitanokaori// Ldn/KU-2097 population should be clarified in further studies.On the other hand, although the vernalization requirement of Norin 61 should be affected by the spring allele of Vrn-D1 (Yan et al., 2004), no Vrn-1 locus was found for flowering-related traits in the Norin 61//Ldn/ PI476874 population.These observations suggest that detection of the Vrn-1 loci as QTLs for flowering-related traits depends on the environment during the growing season, or that effects of the spring allele of Vrn-1 on earliness are small after over-winter growth of wheat plants.Dominant alleles of the Ppd-1 loci more markedly affect earliness of HT, FLT and MAT than do the Vrn-1 loci in the wheat mapping populations.The narrow-sense earliness genes are thought to be involved in the fine-tuning of wheat flowering (Griffiths et al., 2009;Zikhali et al., 2014), and allelic divergence in a narrow-sense earliness gene results in a one-to five-day difference in flowering time in wheat (Valarik et al., 2006;Zikhali et al., 2014).The KU-2097-derived alleles of the 5D and 6D QTLs for HT showed a significantly earlyflowering phenotype under ppd-D1 recessive conditions (Fig. 2, Table 4), suggesting that the earliness alleles may be useful for fine-tuning of wheat flowering.For efficient marker-assisted selection, molecular markers linked to the two KU-2097-derived alleles of the 5D and 6D QTLs for HT should be developed.Although epistatic interaction between the AB and D genomes makes it difficult to detect the effects of useful alleles from Ae. tauschii on flowering-related traits in hexaploid wheat, at least some of the early-flowering alleles from Ae. tauschii should be made available for production of early-flowering Japanese wheat cultivars.

Fig. 1 .
Fig. 1.Linkage maps for chromosomes 2D and 5A and QTL likelihood curves of LOD scores for four flowering-related traits.Genetic distances (in centimorgans) are given to the right of each chromosome.Black arrowheads indicate the putative positions of centromeres.(A) Kitanokaori//Ldn/KU-2097 mapping population.(B) Norin 61//Ldn/PI476874 mapping population.

Fig. 2 .
Fig. 2. Genotypic effects at each previously detected QTL on the observed variation in flowering-related traits in the Kitanokaori//Ldn/KU-2097 mapping population.Markers used to deduce the genotype at a QTL are listed above each graph, and the number of the F 2 individuals is indicated within each bar.Means (days) ± standard error with the same letter are not significantly different (P > 0.05, Tukey-Kramer HSD test).(A) F 2 individuals selected independently of the genotype around the Ppd-D1 region.(B) F 2 individuals with the Ldn/KU-2097homozygous allele around the Ppd-D1 region.

Table 1 .
Parental and F 2 population means for four flowering-related traits measured in two F 2 mapping populations Data are represented as mean (days) ± standard deviation.Student's t-test was used to determine the statistical significance (*P < 0.05, ***P < 0.001) of parental differences.

Table 2 .
Correlation coefficient (r) matrices for four traits measured in two F 2 populations

Table 3 .
Summary of QTLs for flowering-related traits identified in the two F 2 mapping populations