Cross population verified frost QTL and related potential candidate genes
Thirty individual frost QTL were detected in the six DH populations. Out of these, there were 18 major QTL that were each responsible for a phenotypic variation greater than 9.5% and were distributed across 13 chromosomes (2A, 2B, 2D, 3A, 4A, 4B, 4D, 5A, 5D, 6D, 7A, 7B, and 7D). Most frost QTL were closely linked to the QTL for anthesis and physiological maturity Zadok stages and to anthesis-related genes. The QTL results for individual DH populations are included in Table 1.
A large proportion (83%) of the detected frost QTL were showed up consistently in two or more DH populations, except for the QTL on SpM_1A, SpM_1D, SpM_3D, Sp7A_6D, and G7A_3B (Supplementary Fig. 3). A frost QTL was detected on the homologous region on the short arm of chromosome 2A in the SpB and SpM populations, while the other two frost QTL distant from each other in SpM were closely linked to the anthesis and physiological maturity loci on the long arm of chromosome 2A (Table 1; Supplementary Fig. 4). The dominant photoperiod (Ppd ) allelePpd-A1 is located on the short arm of 2A. In the SpB and SpM populations, the Ppd1_2A (2A: 36.9 Mb) allele is located between 2AM5988 (2A: 31.7Mb) and 2AM67517 (2AL 40.8 Mb) and tightly linked to the frost QTL on 2A. The frost QTL on the 2A long arm in the SpM population is a new locus segregating for frost.
A highly significant frost QTL (the highest LOD score 9.2) was detected on homologous regions of the 2B chromosome in four populations of Sp7A, SpB, G7A and St3B, and the additive effects were all contributed by Bethlehem and its substitution lines, while another QTL (LOD score 3.9) in SpM was also closely associated with the homologous region, with the phenotype contributed by Mace (Table 1; Fig. 3). A Ppd1_2B gene (gene bank number DQ885765) was used to identify the physical map location. The potential location of Ppd1_2B was on the short arm of 2B at 56.2 Mb, which was above the marker 2BM46469 (2B: 58.8 Mb) in Sp7A, SpB and G7A, and between markers of 2BM29812 (2B: 53.4Mb) and 2BM73250 (2B: 77.2 Mb) in St3B. Strikingly, the location ofPpd1_2B was in the frost and anthesis QTL region in those four populations, and the QTL results matched the potential segregation ofPpd-B1 gene copy number (Fig. 2d), which points toward a contribution ofPpd-B1 copy number to the phenotype. This QTL corresponds to the newly identified Ppd1_2B location in this study.
On linkage groups 2D3 and 2D2, frost QTL were detected in both SpM and Sp7A, contributed by Mace and Spitfire, respectively. These QTL were associated with anthesis time, as anthesis QTL were closely linked to the homologous region of 2D (Supplementary Fig. 5). Gene Ppd1_2Dwas located at 2D: 33.9 Mb from the upstream on the short arm of 2D. SNP M22365 (2D: 18.2 Mb) on Sp7A_2D1 and SNP M15795 (2D: 17.3 Mb) on SpM_2D1 were close to Ppd1_2D on the physical map. The frost and anthesis QTL on SpM_2D3 and Sp7A_2D2 tend not to be associated with the Ppd1_2D locus, since the Ppd_D1a locus does not expected to segregate in those populations. This frost QTL may imply a new phenology gene on 2D, associated with anthesis time.
On chromosome 3A, the frost QTL in SpB population was located on the short arm, whereas in St3B it was located on the long arm. The QTL on the short arm overlapped with anthesis QTL (Supplementary Fig. 6).TaGI gene sequences (AF543844) were isolated (Sawa et al., 2007) and were located on 3A (3A: 84.1 Mb), 3B (3B: 117.9 Mb) and 3D (3D: 71.9 Mb). One frost QTL in SpB_3A was located between SNP markers 3AM63098 (3A: 56.4 Mb) and 3AM34922 (3A: 107.7 Mb). TaGI_3A is most likely the candidate gene contributing to this frost QTL and the associated anthesis QTL. The 3A long arm QTL in the St3B population was located between 3AM9626 (3A: 711.0 Mb) and 3AM39004 (3A: 729.7 Mb), which are not associated with the TaGI gene. The Genebank number of KF769443 was used to identify the location ofEps_3Am on the physical map.Eps_3Am (3A: 740.1 Mb) was located between the markers 3AM10770 (3A: 739.3 Mb) and 3A73079 (3A: 741.2 Mb), about 10 cM away from the frost QTL.
Significant frost QTL were consistently detected on the long arm of chromosome 4A in the SpM, G7A and St3B populations (Supplementary Fig. 7). Interestingly, the phenotypes were all attributable to the male parents of Mace, B7A and B3B. The frost QTL in SpM and G7A populations were located on the anthesis QTL region, and the late anthesis QTL were contributed by the female parents, Spitfire and Gregory. In other words, the early-flowering phenotype contributed by male parents led to frost susceptibility. The homologous gene of the DELLA protein (rht1_D1a ; AJ242531) on 4A is TraesCS4A02G271000 (4A: 582.4 Mb), which is located between markers M76744 (4A: 575.0 Mb) and M43375 (4A: 597.6 Mb), and above the common marker 4AM77169. Interestingly, the newly identified rht1_4A locus in the current study was located 20 cM away from the minor frost QTL regions on the short arm of 4A in SpM and G7A populations. The homologous gene sequence ofWSOC1_4D is TraesCS4A02G320300, on chromosome 4A: 608.8 Mb, near SNP marker 4AM77381(4A: 612.1 Mb), which is 30-40 cM away from the frost QTL on the long arm regions of 4A in the G7A and St3B populations. The potential WSOC1_4A locus is further away from the frost QTL in the SpM population.
Significant frost QTL were detected on the Rht1 region on chromosome 4B in G7A and St3B populations segregating for Rht1(Fig. 4a), which were contributed by the female parents, Gregory and Suntop, which harbour the Rht-B1b allele. No flowering QTL were detected in those regions. A significant frost QTL with a LOD score of 16.0, attributable to Mace, was repeatedly detected on the distal region of chromosome 4B short arm in the SpM population. The QTL region was about 80 cM away from the recessive rht1 gene loci. The contributing gene for this QTL is unknown. A minor frost QTL was detected in the BW and G7A populations on the homologous region of the long arm of chromosome 4B. In the BW population, the frost QTL were closely linked to the anthesis QTL. Interestingly, frost QTL in the BW population was in the 4BM4887 (4B: 646.6 Mb) region, which is only 6.4 Mb away from the WSOC1_4B and the anthesis QTL, close to 4BM8859 (4B: 660.7 Mb). These results suggest that WSOC1_4B might be a new gene influencing flowering time and contributing to segregation for frost.
Likewise, significant frost QTL were detected on the Rht2 gene region on 4D in the G7A and St3B populations segregating for theRht2 gene, contributed by B7A and B3B, which harbour theRht2 allele (Fig. 4b).
Frost QTL were consistently detected on the homologous region of chromosome 5A in the SpM, Sp7A and SpB populations, with the highest LOD score of 7.7 found in Sp7A (Fig. 5). The frost QTL were in theVrnA1a region, attributed to B7A and Bethlehem in the Sp7A and SpB populations, respectively, whereas it was contributed by Spitfire in the SpM population. The frost QTL in the Sp7A population overlapped with the anthesis QTL, contributed by Spitfire. Both B7A and Bethlehem host aVrnA1a mutant and flower early. The results further indicate that frost damage was caused by the early flowering genotypes of B7A and Bethlehem. A minor frost QTL in the St3B population was on the distal region of the short arm 5A, away from the anthesis QTL, and not associated with the flowering genes or their related phenotypes.
In the case of chromosome 5D, frost QTL were detected in the SpM, BW and St3B populations (Fig. 6). The largest frost QTL (the highest LOD score 14.2) on 5D was in the VrnD1a region in the SpM population while in the BW population it was close to VrnD1a . The frost phenotype in the SpM and BW populations was contributed by Spitfire and Bethlehem, which harbour VrnD1a , whereas for the St3B population in the Williams trial, the contributed phenotype was attributed to Suntop. Several anthesis and maturity QTL were detected in the VrnD1aregion, indicating that frost damage is closely associated with plant growth stages. Another vernalisation related gene, Vrn-D4 , which is a homologue of Vrn1_5A (VrnA1a ), originated from a large segment of chromosome 5A inserted into the short arm of 5D (Kippes et al., 2014; Kippes et al., 2015). According to the physical map ofVrn-D4 (5D: 193 Mb), the gene location should be close to marker 5DM45442 (5D: 278 Mb) on the St3B _5D map, while it is next to 5DM62708 (5D: 154 Mb) on the BW map. The frost QTL may stem from theVrn-D4 locus.
Significant frost QTL were detected on chromosome 7A in the SpM, SpB and BW populations (Fig. 7), with the highest LOD score of 10. The phenotype in the SpB and BW populations was contributed by Bethlehem. Frost QTL in the SpM population were located in three positions, mainly attributed to Mace. According to the physical map of TaVRT-2_7A (7A: 128.8 Mb), the closest SNP marker is 7AM1849 (7A:128.5 Mb), close to 7AM75587, which was tightly linked to the frost QTL in the SpB and BW populations, while in the SpM population the location was closely linked to the anthesis QTL on 7A. One frost QTL was detected in the SpM and SpB populations on the homologous distal region on chromosome 7A. This corresponds to a new segregating locus for frost impact on 7A, which was not associated with the flowering genes or their related phenotypes.
For chromosome 7B, frost QTL were detected in the SpM and BW populations (Fig. 8a). However, the frost QTL regions were distinct from each other. The frost QTL in the SpM population, contributed by Mace, was on the terminal region of the short arm of chromosome 7B, away from the anthesis QTL, whereas in the BW population, the frost QTL were closely linked to the QTL for anthesis and maturity, and were contributed by Westonia. The anthesis QTL in the SpM and BW populations were located on the homologous region on 7B. Another anthesis QTL detected exclusively in the SpM population was located on the distal region of the long arm of 7B. In the SpM population, the physical map of VRN-B3 (7B: 9.7 Mb) was close to marker 7BM76084 (7B: 6.7 Mb), which may contribute the frost phenotype in SpM, while the location was 30 cM away from the frost QTL in the BW population. Based on the physical map, TaVRT-2_7B(7B: 90.1 Mb) is located between 7BM53206 (7B: 64.7 Mb) and 7BM10089 (7B:115.2 Mb) on the BW map, closely linked to the QTL for anthesis and frost, suggesting the involvement of this locus in frost damage in the BW population. In the SpM population, TaVRT-2_7B is not close to the frost QTL.
A QTL with a LOD score of 4.6 also appeared on the homologous regions on chromosome 7D in the Sp7A and SpB populations (Fig. 8b). The frost phenotype was contributed by B7A and Bethlehem and was highly associated with anthesis QTL. TaVRT-2_7D (7D: 128.9 Mb) was close to the upper SNP marker 7DM76171 (7D: 112.0 Mb) in Sp7A population and distant to the lower SNP marker 7DM42766 (7D: 182.6 Mb) in both the Sp7A and SpB populations. This gene tends to contribute the significant frost QTL on 7D.