Dissection of two quantitative trait loci with pleiotropic effects on plant height and spike length linked in coupling phase on the short arm of chromosome 2D of common wheat (Triticum aestivum L.)

Key message Two QTL with pleiotropic effects on plant height and spike length linked in coupling phase on chromosome 2DS were dissected, and diagnostic marker for each QTL was developed. Abstract Plant height (PHT) is a crucial trait related to plant architecture and yield potential, and dissection of its underlying genetic basis would help to improve the efficiency of designed breeding in wheat. Here, two quantitative trait loci (QTL) linked in coupling phase on the short arm of chromosome 2D with pleiotropic effects on PHT and spike length, QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2, were separated and characterized. QPht/Sl.cau-2D.1 is a novel QTL located between SNP makers BS00022234_51 and BobWhite_rep_c63957_1472. QPht/Sl.cau-2D.2 is mapped between two SSR markers, SSR-2062 and Xgwm484, which are located on the same genomic interval as Rht8. Moreover, the diagnostic marker tightly linked with each QTL was developed for the haplotype analysis using diverse panels of wheat accessions. The frequency of the height-reduced allele of QPht/Sl.cau-2D.1 is much lower than that of QPht/Sl.cau-2D.2, suggesting that this novel QTL may be an attractive target for genetic improvement. Consistent with a previous study of Rht8, a significant difference in cell length was observed between the NILs of QPht/Sl.cau-2D.2. By contrast, there was no difference in cell length between NILs of QPht/Sl.cau-2D.1, indicating that the underlying molecular mechanism for these two QTL may be different. Collectively, these data provide a new example of QTL dissection, and the developed diagnostic markers will be useful in marker-assisted pyramiding of QPht/Sl.cau-2D.1 and/or QPht/Sl.cau-2D.2 with the other genes in wheat breeding. Electronic supplementary material The online version of this article (10.1007/s00122-019-03318-z) contains supplementary material, which is available to authorized users.


Introduction
Wheat (Triticum aestivum L.) is an important food crop worldwide, providing calories and proteins consumed by humankind (Fischer et al. 2014;Shiferaw et al. 2013). Plant height (PHT) is a crucial trait related to plant architecture and yield potential (Cadalen et al. 1998;Peng et al. 1999; Communicated by Yunbi Xu.

Electronic supplementary material
The online version of this article (https ://doi.org/10.1007/s0012 2-019-03318 -z) contains supplementary material, which is available to authorized users. Sakamoto and Matsuoka 2004). Consequently, the use of dwarfing genes to reduce PHT and improve yield has been one of the main strategies in breeding modern high-yielding hexaploid bread wheat varieties. For example, during the green revolution, the introduction of semi-dwarf varieties into wheat (T. aestivum L.) appropriately reduced PHT and contributed significantly to a worldwide increase in potential grain yield (Peng et al. 1999). PHT is known to be typically under polygenic control (Bellucci et al. 2015;Tang et al. 2007). Thus, identification of QTL/gene controlling PHT would help to improve the efficiency of designed breeding in wheat.
To date, 24 genes influencing PHT have been identified and designated reduced height genes in wheat (McIntosh et al. 2017). Of these 24 major genes, Rht-B1b and Rht-D1b, which are located on chromosomes 4B and 4D, respectively (Pearce et al. 2011), were cloned and most widely used in wheat breeding (Borner et al. 2002;Cadalen et al. 1998). They encoded DELLA proteins, which are transcriptional regulators that reduce response to gibberellin. Thus, Rht-B1b and Rht-D1b belonged to the group of dwarfing genes that is insensitive to gibberellic acid (Pearce et al. 2011). The characterization of these two genes has enhanced our knowledge about PHT determination in wheat, and functional markers developed for them have been used in wheat breeding (Akman and Bruckner 2012;Borrell et al. 1991;Tang et al. 2009).
The other extensively used gene in wheat breeding is Rht8, which is derived from the Japanese variety Akakomugi and was introduced into Southern European wheat breeding in the 1930s by the Italian breeder Strampelli, together with the photoperiod-insensitive, early flowering Ppd-D1a allele (Lorenzetti 2000). Rht8 was mapped on the short arm of chromosome 2D, and a closely linked SSR marker named Xgwm261 was detected ). The 192-bp allele of Xgwm261 corresponds to a height-reducing phenotype of Rht8, attributing to a 7-8 cm reduction in PHT without pleiotropic effects on other agronomic traits ). Since then, the 'diagnostic' 192-bp allele was used to survey the presence of Rht8 in wheat cultivars (Ahmad and Sorrells 2002;Asplund et al. 2012;Bai et al. 2004;Chebotar et al. 2001;Liu et al. 2005;Worland et al. 2001a, b;Zhang et al. 2006). However, the 192-bp allele of Xgwm261 was not always linked to Rht8 (Ellis et al. 2007). A recent study showed that Rht8 is located in a genetic interval of 1.29 cM (DG279-DG371), which is 1.95 cM away from Xgwm261 (Gasperini et al. 2012). Thus, the development of a robust marker tightly linked to Rht8 is the priority for marker-assisted selection.
Over the past two decades, the successful application of quantitative-genetic methodology has facilitated identification of numerous QTL for PHT in wheat (Borner et al. 2002;Cadalen et al. 1998;Wurschum et al. 2015Wurschum et al. , 2017Yu et al. 2014;Zanke et al. 2014). In our recent study, two QTL (QPht.cau-2D.1 and QPht.cau-2D.2) for PHT linked in coupling phase were mapped on the short arm of chromosome 2D (Zhai et al. 2016). Since both QTL are located on chromosome 2DS, this investigation was undertaken to determine the genetic relationship between QPht.cau-2D.1 and QPht.cau-2D.2 using segregating populations and NILs derived from a residual heterozygous line. Moreover, the diagnostic marker for each QTL was developed for markerassisted selection in wheat breeding programmes. In addition, haplotype distribution of QPht.cau-2D.1 and QPht. cau-2D.2 in diverse panels of wheat accessions was also investigated.

Plant materials
Following our previous study of QTL mapping using the recombinant inbred lines (RILs) of Yumai 8679 (Y8679)/ Jing 411 (J411) (Zhai et al. 2016), SSR (simple sequence repeat) markers were further developed for linkage map construction and QTL analysis. In generation F 9 of the RIL population, a RHL (RIL171) that carried the heterozygous segment at the genetic region from SSR markers SSR-2212 to SSR-2429 was self-pollinated to produce the F 10 generation for further study (Fig. 1). The homozygotes (F 10 ) without recombinant were selected as NILs (NIL Y8679 and NIL J411) to validate the presence of the QTL. Two plants that carried heterozygous segments covering the intervals from SSR-2212 to Xwmc503 and from Xcfd53 to SSR-2429 were identified to produce segregate populations (F 11 , population I and II) for genetic analysis (Fig. 2b). Then, the non-recombinant homozygotes of each population were selected and self-pollinated to produce NIL-I (F 12 , NIL-I Y8679 and NIL-I J411 ) and NIL-II (F 12 , NIL-II Y8679 and NIL-II J411 ).

Field experiments
Populations I and II were tested in Hebei (37°56′N, 114°42′E). NIL Y8679 and NIL J411 were grown in two locations, Shanxi (36°08′N, 111°34′E) and Hebei, and NIL-I and NIL-II were grown in three locations, Beijing, Shanxi and Hebei (Supplementary Table S5). In each environment, two segregating populations were planted with a sowing rate of 20 seeds per row (1.5 m long and 0.3 m apart), while the NILs were planted 25 seeds per line. The irrigation and other management of all sites were in accordance with local standard practices.

Phenotypic evaluation and statistical analysis
Plant height (PHT), spike length (SL) and spikelet number (SN) of the individuals in populations I and II were measured before harvest. For the NILs, five plants from each line were used for phenotypic evaluation. PHT, SL, SN and length of 5 internodes from the main tillers were measured. For NILs, the mean values over three replications at each site were used for data analysis. Spikelet compactness (SC) was calculated by dividing the SN by the SL.
Statistical analysis was performed with IBM SPSS Statistics 20 (SPSS, Chicago, USA). Significance analysis was calculated using Student's t test.

SSR marker development
Microsatellite markers were designed to construct the genetic map of chromosome 2DS. The reference sequence 1 3 of wheat variety Chinese Spring (http://www.wheat genom e.org/) and Aegilops. tauschii genomic DNA contig sequence (http://aegil ops.wheat .ucdav is.edu/ATGSP /) were extracted to design new SSR markers by using BLASTN method with the sequences of markers DG279 and DG371 (Gasperini et al. 2012) and the flanking sequence of the polymorphic SNPs in our previous results (Zhai et al. 2016). The PCR system included 5 μl 2 × Taq PCR StarMix, 2 μl primer (mixture of left and right primer), 2 μl DNA template (50-100 ng/μl) and 1 μl H 2 O. The PCR protocol was performed using PCR amplification as follows: 94 °C for 5 min; 36 cycles of 94 °C denaturation for 30 s, 55-57 °C primer annealing for 30 s, and 72 °C extending for 30 s; and finally, 72 °C for 10 min. For polymorphism detection, PCR products were separated using the method of 8% non-denaturing polyacrylamide gel electrophoresis (PAGE) (Marklund et al. 1995). The primers of 17 codominant SSR markers are listed in Supplementary Table S14.

Linkage map construction and QTL analysis
JoinMap 4.1 was used for constructing genetic linkage map of chromosome 2DS with LOD value above 5. A regression mapping algorithm (Stam 1993) and Kosambi's mapping function (Kosambi 1943) were applied to determine the marker order and convert recombination into distances.
QTL mapping was determined with Windows QTL Cartographer 2.5 (Wang et al. 2006). Briefly, composite interval mapping (CIM) was performed using model 6 with forward and backward regression, five markers, and a 1-cM scanning window as cofactors. Empirical threshold LOD scores estimated with 1000 permutations at P ≤ 0.05 detected QTL with overlapping confidence intervals (± 2 LOD away from the peaks of likelihood ratios) were considered equivalent and named as suggested by McIntosh et al. (2017). 1 3

Histological analysis
The medial sections of peduncle at flowering stage were collected from NIL-I Y8679 , NIL-I J411, NIL-II Y8679 and NIL-II J411 plants, then fixed in FAA solution (50% (v/v) ethanol, 5% (v/v) glacial acetic acid and 4% (v/v) formaldehyde) and subjected to vacuum pumping for 30 min. Next, the internodes were dehydrated in a series of ethanol solutions (75% (v/v) ethanol and anhydrous ethanol) and destained in a series of xylene solutions (3:1 ethanol: xylene, 1:1 ethanol: xylene, 1:3 ethanol: xylene, and pure xylene). The internodes were soaked in each ethanol and xylene solution for 1 h and then embedded in paraffin . Tissue sections were cut into 4-μm-thick slices, fixed on a glass slide, and stained with 1% sarranine and 0.5% fast green (G1031, http://www.servi cebio .cn/).

Sequence analysis of candidate genes
Genomic DNA for Y8679 and J411 was used to build pairedend sequencing libraries with insert sizes of approximately 500 bp, according to vendor-provided instructions (Illumina). An average 5 × coverage of the assembled genome, with 150-bp paired-end reads for each accession, was generated with the Illumina HiSeq X Ten platform. All the sequence reads for each parent were mapped to the newly updated genome of Chinese Spring (RefSeq v1.0) in the Burrows-Wheeler Aligner program (BWA, ver. 0.7.15) with default parameters (Li and Durbin 2009). SNPs and InDels are identified by the HaplotypeCaller module. The sequence of each candidate gene including intron, exon, 3′-UTR region and 2-kb sequence upstream of translation start codon was used as query for analysis.

Saturation of genetic Linkage map and QTL mapping of QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2
To saturate the genetic linkage map for plant height (PHT) and spike length (SL) on the short arm of chromosome 2D, 632 SSR markers were developed, among which 17 markers exhibited polymorphism between parents Y8679 and J411 (Supplementary Table S14). The collinearity of these markers with Ae. tauschii markers was analysed, and the physical positions on the reference sequence of Chinese Spring are shown in Fig. 1. These 17 polymorphic SSR markers were used to genotype the 191 RIL lines for QTL analysis. The resulting genetic linkage map of chromosome 2D consisted of 75 markers spanning 70.11 cM in length (Supplementary Table S13). Consistent with our previous finding, two stable QTL in coupling phase controlling PHT (QPht.cau-2D.1 and QPht.cau-2D.2) were detected on chromosome 2DS by reanalysing the phenotypic data from 191 RIL lines in 9 individual environments and 1 combined analysis (BLUP) (Zhai et al. 2016). QPht.cau-2D.1 was located between SNP markers BS00022234_51 and BobWhite_rep_c63957_1472, which explained 4.39-11.94% of the phenotypic variation. QPht.cau-2D.2 was mapped to the genomic interval between SSR markers SSR-2062 and Xgwm484, which explained 4.12-12.96% of the phenotypic variations under different environments ( Fig. 2a, Table 1). In the same positions of QTL for PHT, two major QTL for SL (QSl.cau-2D.1 and QSl.cau-2D.2) were also detected under all environments, which explained 30.94-40.63 and 31.31-41.95% of the phenotypic variations, respectively. All increasing alleles for PHT and SL came from J411 (Table 1). For simplification, hereafter, the two genomic regions covering PHT and SL were designated as QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2, respectively.

Development of near isogenic lines harbouring QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2
To further evaluate the genetic effects of QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2 on PHT and SL, one line (RIL171) from the F 9 generation of the RIL population was selected for investigation because of its residual heterozygosity at the target QTL genomic regions. First, one pair of NILs (NIL Y8679 and NIL J411 ) from the self-pollinated progeny of RIL171 (F 10 ) harbouring QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2 were genotyped using the wheat 660 K SNP array (https ://wheat .pw.usda.gov/ggpag es/topic s/Wheat 660_SNP_array _devel oped_by_CAAS. pdf). Of 630517 SNP markers with detected signal, only 434 (0.07%) exhibited polymorphism between NIL Y8679 and NIL J411 , indicating that the genetic backgrounds of these two NILs had very high similarity (Supplementary  Table S6). Notably, based on the information of SNP markers with genomic position, 154 of 434 polymorphic SNP markers (35.48%) were located on genomic  Table S7). Second, the progenies of the above NILs (F 11 ) were used for phenotypic analysis. As shown in Table 2, significant differences were observed for SL, length of peduncle (PED) and length of 1st internode under peduncle (1IL) between NIL Y8679 and NIL J411 (P < 0.001) under different environments. By contrast, the differences in length of the second internode under peduncle (2IL), length of the third internode under peduncle (3IL) and length of the fourth internode under peduncle (4IL) were only detected under a specific environment. In addition, no difference was observed for spikelet number (SN). Consequently, PHT of NIL J411 was 12.62 and 15.08% (P < 0.001) higher than that of NIL Y8679 under two different environments. Correspondingly, spikelet compactness (SC) of NIL Y8679 was significantly (P < 0.001) higher than that of NIL J411 ( Table 2).  (Fig. 2b). The single marker analysis for PHT, SL and SN was performed using the derived segregating populations I and II (F 11 ). As expected, polymorphic SSR markers in each population were significantly (P < 0.001) associated with PHT and SL (Table 3). To further test the pleiotropic effects of QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2, two sets of NILs (F 12 , NIL-I and NIL-II) (Fig. 3) were developed from the two F 11 populations. For PHT, SL and PED, NIL-I J411 and NIL-II J411 (genotype same with J411) had significantly higher values than those of   NIL-I Y8679 and NIL-II Y8679 (genotype same with Y8679) under all environments, with average differences of 9.97% and 12.62, 16.25 and 19.56%, and 13.43% and 18.11%, respectively (Fig. 2c, Supplementary Table S9). In addition, the 1IL and 4IL of NIL-II J411 were also much higher (P < 0.001) than those of NIL-II Y8679 under all environments (Supplementary Table S9).

Effects of QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2 on culm cell length
In wheat, internode elongation is caused by cell division and/or cell elongation (Chowdhry and Allan 1966;Gasperini et al. 2012). To investigate the underlying physiological bases of two QTL, longitudinal cell length in the median section of peduncle at the flowering stage in NILs of each QTL was measured. For QPht/Sl.cau-2D.1, NIL-I Y8679 showed no difference in cells length compared to NIL-I J411 (Fig. 6a, b and e). By contrast, there was significant difference in cell length between NIL-II Y8679 and NIL-II J411 of QPht/Sl.cau-2D.2. The cell length of NIL-II J411 with increasing allele was significantly longer than that of NIL-II Y8679 with decreasing allele for PHT (Fig. 6c, d, and f, P < 0.001).
Collectively, our data exhibited that the alleles of STARP-2001-Y and SSR-2433-Y were significantly associated with the height-reduced alleles of QPht/Sl.cau-2D.1 and QPht/ Sl.cau-2D.2, respectively. To further explore implications of our findings on wheat breeding, the haplotype of 724 common wheat varieties/ lines from China was first analysed using diagnostic markers  Table S11). The frequency of SSR-2433-Y allele in different zones was in the order of II (86.67%) > X (78.05%) > IV (51.22%) > III (48.15%) > I (46.51%) > VIII (33.33%) (Fig. 5b, Supplementary Table S11). When comparing the alleles of STARP-2001-Y and SSR-2433-Y in the same zone, the frequency of SSR-2433-Y was much higher than that of STARP-2001-Y in zones I, II, III, IV and X, especially in the major wheat production zone X (Fig. 5a, b, Supplementary Table S11). Furthermore, 442 common wheat varieties/lines from other countries were tested by the two diagnostic markers. The STARP-2001-Y allele was present in 48 of 442 accessions (10.86%), spreading in five continents, including Asia, Europe, North America, South America and Oceania. The SSR-2433-Y allele was detected in 63 of 442 accessions (14.25%), which are derived from Asia, Europe, Africa and North America (Fig. 5c, d, Supplementary Table S12).
To trace the origin of dwarf genes in the QTL intervals of south-western winter wheat region; V, southern winter wheat region; VI, north-eastern spring wheat region; VII, northern spring wheat region; VIII, north-western spring wheat region; IX, Qinghai-Tibet spring-winter wheat region; X, Xinjiang winter-spring wheat region Table S2). In addition, a diverse panel of 86 accessions of Aegilops tauschii, the D-genome progenitor of T. aestivum, was also analysed. The results exhibited that the STARP-2001-Y allele was not detected in 86 accessions of Aegilops tauschii, but the SSR-2433-Y allele presented in 39 of 86 accessions (45.35%) (Supplementary Table S4). NS, no significance P = 0.05; ***, t test P < 0.001. Bars represent the standard deviation

Sequence variations in candidate genes for QPht/ Sl.cau-2D.1 and Rht8
To identify the candidate genes for QPht/Sl.cau-2D.1 and Rht8, we analysed the gene models in the mapping interval of the Chinese Spring RefSeq v.1.0 sequence. In rice, WRKY transcription factor Dlf1 and gibberellin-inactivating 2-betadioxygenase gene played important roles in the regulation of PHT (Cai et al. 2014;Lo et al. 2008). Interestingly, of 38 predicted genes in the 2.5-Mb interval of QPht/Sl.cau-2D.1 (BS00022234_51-BobWhite_rep_c63957_1472), TraesC-S2D01G051000 and TraesCS2D01G051500 encoded two WRKY transcription factors, and TraesCS2D01G049700 encoded gibberellin 2-beta-dioxygenase, which may be the candidate genes (Supplementary Table S10). DNA sequence analysis showed that there were 5 SNPs and 3 InDels between TraesCS2D01G051500 of parental lines Y8679 and J411, including 3 SNPs in exon 1, 1 InDel in 3′-UTR region, 1 InDel in intron 2 and 2 SNPs and 1 InDel in the upstream region of translation start codon (Fig. 8, Supplementary  Table S15). However, no sequence difference between Y8679 and J411 was detected for TraesCS2D01G049700 and TraesCS2D01G051000.
The well-known reduced height genes Rht-B1 and Rht-D1 in wheat and several plant height related genes in rice (DLT, SMOS1/SHB and SLR1) encoded transcriptional factors belong to the GRAS family (De Vleesschauwer et al. 2016;Qiao et al. 2017;Tong et al. 2009). In rice, OsCCC1 was involved in cell elongation by regulating ion (Cl − , K + , and Na + ) homoeostasis to maintain cellular osmotic potential, which affected PHT in turn (Chen et al. 2016). Based on the new reference sequence of Chinese Spring (RefSeq v.1.0), the mapped region of Rht8 between SSR-2062 and Xgwm484 contained 419 predicted genes, among which TraesCS2D01G055700 and TraesCS2D01G058700 encoded two GRAS transcription factors, and TraesC-S2D01G059300 encoded a cation-chloride cotransporter, which may be the candidate genes (Supplementary  Table S10). DNA sequence analysis revealed that there were 38 SNPs and 8 InDels between TraesCS2D01G055700 of Y8679 and J411, 3 SNPs and 3 InDels in the exon, and 35 SNPs and 5 InDels in the upstream region of translation start codon. For TraesCS2D01G058700, only 2 SNP variants were found, which were located in intron 4 and the upstream region of translation start codon, respectively (Fig. 8, Supplementary Table S15). By contrast, the sequence of TraesC-S2D01G095300 in Y8679 was identical to that in J411.

Two QTL for PHT/SL linked in coupling phase on chromosomes 2DS
Neighbouring QTL that are linked in coupling phase are commonly observed in primary QTL analysis (Fan et al. 2017;Wu et al. 2015). To date, many studies have tried to dissect QTL in coupling phase using NILs or residual heterozygous lines and found that coupling QTL were partially attributed to tightly linked independent QTL (Chemayek  Shen et al. 2011). For example, Wu et al. dissected two tightly linked QTL for PHT (qPH3) and yield per plant (qYD3) by using two rice near isogenic populations (Wu et al. 2015). In wheat, tight repulsion linkage between Sr36 and Sr39 was revealed by genetic, cytogenetic and molecular analyses (Chemayek et al. 2017). Recently,using 191 RILs derived from Y8679 and J411,2) controlling PHT and SL were mapped on the short arm of chromosome 2D. However, it is difficult to conclude whether QPht/Sl.cau-2D.1 is a shadow or genuine QTL because it is only 7 cM away from QPht/Sl.cau-2D.2 (Zhai et al. 2016). Here, we dissected these two QTL (QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2) using segregating populations and NILs derived from a residual heterozygous line (RIL171). They were linked in coupling phase on chromosome 2DS, with increased alleles from the same parent (J411). QPht/Sl.cau-2D.1 was located between SNP markers BS00022234_51 and Bob-White_rep_c63957_1472, and QPht/Sl.cau-2D.2 was located between SSR markers SSR-2062 and Xgwm484. The results of NILs indicated that the genetic effect of these two QTL is similar, which could reduce plant height by 4. 79-14.32 and 9.49-14.43 cm under different environments (Table 2). Our present study provides a new example of QTL dissection in wheat, but the underlying molecular basis of two QTL (QPht/Sl.cau-2D.1 and QPht/Sl.cau-2D.2) is still an area for further elucidation.

Robust diagnostic SSR marker for QPht/Sl.cau-2D.2
Numerous studies exhibited that there were major QTL/ genes on chromosome 2DS, such as the well-known and widely used dwarf gene Rht8 (Chebotar et al. 2013;Griffiths et al. 2012;Zanke et al. 2014). Recently, using a fineresolution mapping approach, the Rht8 genetic interval has been reduced from 20.5 to 1.29 cM, flanking by DG279 and DG371 (Gasperini et al. 2012). Comparative analysis revealed that QPht/Sl.cau-2D.2 of the present study is 1 3 located on the same genetic interval of Rht8 (Gasperini et al. 2012). Moreover, consistent with the effect of Rht8 on culm cell elongation, significant difference in cell length was also observed between NILs of QPht/Sl.cau-2D.2 (Fig. 6c, d). This, we speculated that QPht/Sl.cau-2D.2 may be the wellcharacterized Rht8 gene.
Due to the widespread use of Rht8 in wheat breeding, assessing the distribution of height-reducing alleles at the Rht8 is of interest for breeding purposes. Following the identification of Xgwm261, a closely linked marker to Rht8 with the 192-bp allele corresponding to a height-reducing phenotype, the gwm261 192bp allele has been taken as diagnostic marker for Rht8 (Ahmad and Sorrells 2002;Bai et al. 2004;Chebotar et al. 2001;Liu et al. 2005;Zhang et al. 2006). However, several instances were found in which the 192-bp allele was not associated with a height-reducing phenotype (Ellis et al. 2007). For example, a source variety of the Green Revolution semi-dwarfing genes, Norin10, carrying a 192-bp allele of Xgwm261, was independent of the height reduction effect of Rht8 (Ellis et al. 2007). Consistent with this, our data revealed that Xgwm261 was located in the genetic interval of QPht/Sl.cau-2D.1, instead of QPht/Sl.cau-2D.2. Thus, we developed a codominant SSR marker (SSR-2433) that was tightly linked to QPht/Sl.cau-2D.2/Rht8. Moreover, the SSR-2433 locus only has two allelic variants in the diversity panel of wheat accessions, which may be an ideal diagnostic marker for the QPht/Sl.cau-2D.2/Rht8 gene.

A novel QTL for PHT/SL tightly linked to dwarf gene Rht8
Previous studies exhibited that numerous QTL/genes controlling PHT were detected in wheat (Borner et al. 2002;Gao et al. 2015a, b;Peng et al. 2003;Quarrie et al. 2005;Singh et al. 2016;Tian et al. 2017). Specifically, 24 dwarfing genes (Rht1-Rht24) have been catalogued in wheat (McIntosh et al. 2017). However, quite a few genes for reduced stature have been used in wheat breeding, because most showed strongly negative effects on grain yield (Chapman et al. 2007;Zhang et al. 2013a). Thus, it is necessary to explore and utilize new QTL/genes controlling PHT. For example, the new reduced PHT gene Rht24 was important and extensively used in wheat breeding programmes (Wurschum et al. 2017). In the present and our previous studies, we identified a novel QTL for PHT neighbouring to Rht8, QPht/Sl.cau-2D.1, by the following evidences: (1) QPht/ Sl.cau-2D.1 was mapped to a position distal to Rht8 on the short arm of chromosome 2D (Fig. 2a, Table 1); (2) QPht/ Sl.cau-2D.1 had no significant effect on culm cell elongation, which is obviously different from that of Rht8 (Fig. 6).
Common wheat has an allohexaploid genome structure, which was hybrid by T. turgidum (AABB) and Aegilops tauschii (DD) 0.43 million years ago (International Wheat Genome Sequencing 2014; Petersen et al. 2006) To investigate the origin of the novel QPht/Sl.cau-2D.1 during wheat evolution, we analysed the allelic variation in the diagnostic marker STARP-2001 in a diverse panel of 86 accessions of Aegilops tauschii., but the STARP-2001-Y allele was not detected, which is significantly associated with the heightreduced allele of QPht/Sl.cau-2D.1. Thus, it seems that the novel height-reduced allele of QPht/Sl.cau-2D.1 was generated after the formation of allohexaploid wheat. Notably, haplotype analysis revealed that the height-reduced allele of QPht/Sl.cau-2D.1 has not been widely used in wheat breeding compared to Rht8. Thus, QPht/Sl.cau-2D.1 may be a favourable QTL in the genetic improvement in PHT in wheat, but the effect of different alleles of QPht/Sl.cau-2D.1 on grain yield and their potential utilization needs further investigation.

Candidate genes of Rht8 and QPht/Sl.cau-2D.1
To date, several genes controlled PHT have been cloned in wheat, such as Rht-B1, Rht-D1, TaSTE and GA2oxA9 (Rht18) (Ford et al. 2018;Gasperini et al. 2012;Peng et al. 1999;Zhang et al. 2013b). Specifically, Rht-B1 and Rht-D1 encoded transcriptional factors belong to the GRAS family (Peng et al. 1999;Sun 2010). Different alleles of Rht-B1 and Rht-D1 have been also identified, which produce dwarfs with a broad range of plant height (Pearce et al. 2011). Semi-dwarfing alleles Rht-B1d and Rht-B1e introduce premature stop codons within the amino-terminal coding region, whereas the severe dwarfism of Rht-B1c gene is caused by an intragenic insertion of 90 bp within the highly conserved amino-terminal DELLA domain (Pearce et al. 2011). In the present study, we found that two of the three candidate genes for Rht8 (TraesCS2D01G055700 and TraesCS2D01G058700) encoded GRAS transcription factors. Notably, a 35-bp deletion in the coding region of TraesCS2D01G055700 was found in parental line J411 as compared to Y8679, which resulted in frame shift for translation. Moreover, another two SNP variations between Y8679 and J411 leaded to amino acid substitutions. In addition, 35 SNPs and 5 InDels in the upstream region of translation start codon were detected (Fig. 8, Supplementary Table S15). Collectively, we proposed that the role of TraesCS2D01G055700 in plant height merit for further investigation.
In rice, Dlf1, a WRKY transcription factor, acts as a transactivator to downregulate Ehd2/RID1/OsId1 in the signal transduction pathway of flowering and plays an important role in the regulation of PHT (Cai et al. 2014). Of the three candidate genes for QPht/Sl.cau-2D.1, TraesCS2D01G051000 and TraesCS2D01G051500 encoded two WRKY transcription factors, but only TraesCS2D01G051500 has sequence variations between Y8679 and J411, including one SNP (A/C) in exon 1 that leaded to the substitution of 1 amino acid (I/L) and 2 SNPs and 1 InDel in the upstream region of translation start codon (Fig. 8, Supplementary Table S15). Therefore, detailed studies would be necessary to analyse the effect of these variations in the function of TraesCS2D01G051500 and plant height.
Epigenetics regulate gene expression without effects on gene sequences, which was involved in diverse aspects of plant development (Bender 2002;Heer et al. 2018;Paszkowski and Mittelsten Scheid 1998). The repression of FLC expression in Arabidopsis by vernalization is one of the well-studied examples of the epigenetic regulation of a gene in plants (Bastow et al. 2004;Questa et al. 2016). Recently, Zhang et al.(2012) identify a DNA hypomethylation epi-allele in rice and demonstrate that repression of FIE1 via DNA methylation and H3K9me2 is essential for plant height. Thereby, although the sequences of TraesC-S2D01G049700, TraesCS2D01G051000 and TraesC-S2D01G095300 in Y8679 were identical to those in J411, these genes may also be the candidate genes and further experiment is needed to validate which one is the real gene for Rht8 and QPht/Sl.cau-2D.1.
In conclusion, two QTL with pleiotropic effects on plant height and spike length linked in coupling phase on chromosome 2DS in common wheat were separated, and the diagnostic marker tightly linked with each QTL was developed, which corresponded to a novel QTL QPht/ Sl.cau-2D.1 and well-known Rht8 gene, respectively. Of six candidate genes for two QTL, three have sequence variations between parental lines Y8679 and J411. Remarkably, the frequency of the height-reduced allele of QPht/ Sl.cau-2D.1 in diverse panels of wheat accessions is much lower than that of QPht/Sl.cau-2D.2. Taken together, these data provide a new example of QTL dissection, and QPht/Sl.cau-2D.1 may be an attractive target for genetic improvement in wheat breeding.
Author contribution statement ZN and AZ conceived the project; LC, ZC and RB carried out experiments; XC and HZ participated in field trials; LC, ZC and XC performed marker development of the QTL region of interest; LC and ZH developed the segregation populations and near isogenic lines; LC, ZC and RB performed phenotyping of the segregation populations and the near isogenic lines; QS, HP, YY, ZH, MX and WG assisted in revising the manuscript; LC analysed experimental results and wrote the manuscript.