QTL Detection for Wheat Kernel Size and Quality and Their Response to Low Nitrogen Stress

QTL detection for wheat kernel size and quality and the response of these traits to low nitrogen stress

Fa Cui · Xiaoli Fan · Mei Chen · Na Zhang · Chunhua Zhao · Wei Zhang · Jie Han · Jun Ji · Xueqiang Zhao · Lijuan Yang · Zongwu Zhao · Yiping Tong · Tao Wang · Junming Li

Supplementary information

1 Results

1.1 Phenotypic variation and correlations between traits

In nearly all cases, all 11 kernel-related traits (KRTs) widely varied, showed transgressive segregation and were normally distributed, with absolute values of skewness and kurtosis of less than 1 for the 188 KJ-recombinant inbred lines (RILs). The average coefficients of variation (CVs) among the ten environments ranged from 1.23 % (test weight (TW), 0.94–1.58 %) to 14.90 % (Zeleny sedimentation value (ZEL), 10.90–19.92 %) (Supplementary Table S3). These results indicated that all the 11 traits were typical quantitative traits suitable for analysis of quantitative trait loci (QTLs).

Kernel length (KL) showed a positive correlation with kernel width (KW) under high-nitrogen (HN) and low-nitrogen (LN) conditions, although this correlation was statistically non-significant under HN. All seven kernel quality-related traits, except for TW, were significantly positively correlated with each other. Although statistically non-significant under HN, TW showed negative correlations with water absorption (ABS), ZEL, and kernel hardness (KH) under HN and LN, and TW was negatively correlated with grain protein content (GPC) and wet gluten content (WGC) under HN, but these correlations were positive under LN. Although statistically non-significant under both LN and HN, positive correlations existed between KL and GPC and between KW and ZEL. KL showed a positive correlation with DT, ABS, ZEL, and KH under both LN (significant) and HN (non-significant), and KW showed positive correlations with GPC and TW under both LN (non-significant) and HN (significant), as did thousand-kernel weight (TKW) with GPC, WGC, and ZEL. Although non-significant in some cases, KL and kernel diameter ratio (KDR) were negatively correlated with TW and dough tractility (DT), respectively. The remaining factors showed inconsistent correlations with each other under LN and HN, including either a negative correlation under LN but a positive correlation under HN or a positive correlation under LN but a negative correlation under HN. Theses correlations were statistically non-significant in the majority of cases.

1.2 Genotypic analysis of the ten genes related to kernel size and quality traits and the novel genetic linkage map

Concerning the three high-molecular-weight (HMW) glutenin loci, the diagnostic markers of Glu-B1 showed polymorphisms between KN9204 and J411. ZSBy9aF1/R3, the dominant marker of By9, was detected by the presence of a diagnostic fragment (662 bp) in KN9204, but no corresponding amplification product was observed in J411. For Glu-A1, KN9204 had the Ax-null allele, whereas J411 had Ax-2* (Cui et al. 2014a). In addition, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) revealed that KN9204 carried the null, 7+9, and 2+12 alleles, whereas J411 carried the 2*, 6+8, and 2+12 alleles at the Glu-A1, Glu-B1 and Glu-D1 loci, respectively (data not shown). These findings were consistent with the results identified by assessment of the functional markers of the three HMW glutenin loci. Concerning the three low-molecular-weight (LMW) glutenin subunits, the diagnostic markers of Glu-A3 showed polymorphisms between KN9204 and J411. Of them, LA3F/SA2R and LA3F/SA4R, the dominant markers of Glu-A3b and Glu-A3d, respectively, were detected by the amplification of 894 bp and 967 bp diagnostic fragments, respectively, in KN9204, but no corresponding amplification products were obtained in J411. LA1F/SA3R, the dominant marker of Glu-A3a and Glu-A3c, was detected by amplification of the diagnostic fragment (573 bp) in J411, but no corresponding amplification product was observed in KN9204. These findings indicated that KN9204 had the Glu-A3b or Glu-A3d alleles and that J411 had the Glu-A3a or Glu-A3c alleles. In addition, J411 had the Glu-B3h allele, whereas KN9204 had a 1BL.1RS translocation that resulted in the loss of Glu-B3 (Cui et al. 2014a). The two parents had the same Glu-D3 allele.

1.3 QTLs for the 11 kernel-related traits

Nine putative additive QTLs for GPC were identified in an individual environment QTL mapping analysis and were distributed on chromosomes 1A, 1B (2 QTLs), 2A, 2B, 3B, 4A (2 QTLs), and 4B. These QTLs individually explained an average of 4.27–8.08 % of the phenotypic variance with average LOD values ranging from 2.07 to 3.66. All of these QTLs were verified in at least two of the ten environments. QGpc-1A and QGpc-1B.2 were stable QTLs that were identified reproducibly in six and five environments, respectively; QGpc-2A, QGpc-2B, and QGpc-4B were all verified in four of the ten environments. Five and four favorable alleles that increased GPC were contributed by J411 and KN9204, respectively (Table 3; Supplementary Table S4; Fig. 2).

For WGC, nine QTLs were significant together in ten individual environments and were distributed on chromosomes 1A, 1B (2 QTLs), 2A, 2B, 3B, 4B, 5D, and 7D. These QTLs individually explained an average of 4.27–9.83 % of the phenotypic variance, with average LOD values ranging from 2.17 to 4.91. All of these QTLs were significant in at least two of the ten environments. QWgc-1A, QWgc-1B.1, and QWgc-5D were stable QTLs that were all identified reproducibly in five environments, and both QWgc-1B.2 and QWgc-2A were significant in four of the ten environments. KN9204 and J411 contributed favorable alleles that increased WGC at five and four QTLs, respectively (Table 3; Supplementary Table S4; Fig. 2).

For DT, nine putative additive QTLs were detected together in ten individual environments and were distributed on chromosomes 1A (2 QTLs), 2A, 2D, 3B, 4A, 4B, 5D, and 7D. These QTLs individually accounted for an average of 3.90–11.44 % of the phenotypic variance, with average LOD values ranging from 2.66 to 6.03. All of these QTLs were detected reproducibly in at least two of the ten environments. QDt-1A.2, QDt-2A, QDt-4A, QDt-4B, and QDt-5D were stable QTLs that were verified in six, eight, five, nine, and eight environments, respectively, and QDt-7D was detected reproducibly in four of the ten environments; in addition, QDt-4B was a major stable QTL for DT that individually explained 11.44 % (6.59–15.78 %) of the phenotypic variance with a LOD value of 6.03 (2.68–9.63). KN9204 and J411 contributed to increased DT at five and four chromosomal regions, respectively (Table 3; Supplementary Table S4; Fig. 2).

Eight QTLs for TW were mapped to chromosomes 2B, 2D, 3B, 4A, 4B (2 QTLs), 5D, and 7A, and they individually explained an average of 4.57–11.91 % of the phenotypic variance, with average LOD values of 2.28–5.61. All of these QTLs were verified in two of the ten environments, with the exception of QTw-4B, which was significant in six environments and individually explained 11.91 % (9.36–14.97 %) of the phenotypic variance as the only major stable QTL for TW. Two and six QTL alleles that increased TW were contributed by KN9204 and J411, respectively (Table 3; Supplementary Table S4; Fig. 2).

Eight QTLs for ABS were mapped to chromosomes 1B, 2A, 2B, 2D (2 QTLs), 3B, 5D, and 6B and they individually accounted for an average of 2.85–32.75 % of the phenotypic variance with average LOD values of 2.55–18.23. They all could be verified in at least two different environments. QAbs-1B.1, QAbs-3B, and QAbs-5D were stable QTLs that could be verified in nine, six, and ten environments, respectively, and QAbs-2B could be verified in four environments; in addition, QAbs-5D individually explained 32.75 % (11.78–49.06 %) of the phenotypic variance, with a LOD value of 18.23 (4.58–26.41), and it was the only stable QTL for ABS. Four and four QTL alleles that increased ABS were donated by KN9204 and J411, respectively (Table 3; Supplementary Table S4; Fig. 2).

In the ten environments, six QTLs for ZEL were identified and were distributed on chromosomes 1B, 2A, 3B (2 QTLs), 4B, and 5D. These QTLs individually explained an average of 5.04–16.54 % of the phenotypic variance, with average LOD values of 2.88–7.71. Both QZel-2A and QZel-5D were stable QTLs that were significant in eight of the ten environments. In addition, QZel-5D individually explained 16.54 % (10.96–30.54 %) of the phenotypic variance, with a LOD value of 7.71 (5.23–14.4), and it was the only major stable QTL for ZEL. The remaining four QTLs were verified in at least two of the ten environments. All six QTL alleles that increased ZEL were contributed by J411 (Table 3; Supplementary Table S4; Fig. 2).

A total of six QTLs for KH were identified in the ten environments and were mapped to chromosomes 1B, 2A, 2D (2 QTLs), 5D, and 6B. These QTLs individually accounted for an average of 2.92–35.57 % of the phenotypic variance, with average LOD values of 2.33–18.14. QKh-5D, the only major stable QTL for KH, was mapped to chromosome 5DS in the vicinity of the known location of Ha, and it was confirmed in all ten environments. QKh-2D was significant in four different environments and the remaining QTLs were significant in only two different environments. Equal number of favorable alleles that increased KH were contributed by KN9204 and J411 (Table 3; Supplementary Table S4; Fig. 2).

In total, 11 putative additive QTLs for KL were identified and were distributed on chromosomes 1A, 1B, 1D, 2A (2 QTLs), 2B, 3B, 4A (3 QTLs), and 7A. These QTLs individually accounted for an average of 4.91–17.64 % of the phenotypic variance, with average LOD values of 2.38–6.33. QKl-1B and QKl-2A.1 were stable QTLs that could be detected reproducibly in seven and six environments, respectively. Moreover, QKl-1B individually exhibited 13.47 % (5.48–18.28 %) of the phenotypic variance, with a LOD value of 6.33 (2.22–9.48), as the only major stable QTL. Both QKl-1A and QKl-2A.2 were major QTLs that individually caused 13.47 and 10.20 % of the phenotypic variance, respectively; however, they were only significant in one and two environments, respectively. The remaining QTLs had a moderate additive effect and could be identified in only two or three different environments. Five and six QTL alleles that increased KL were contributed by KN9204 and J411, respectively (Table 3; Supplementary Table S4; Fig. 2).

A total of 13 QTLs for KW were mapped to chromosomes 1A, 1B, 2A, 2B, 2D, 4A (2 QTLs), 4B (2 QTLs), 5D, 6A, 6B, and 7A. These QTLs individually explained an average of 4.32–21.70 % of the phenotypic variation, with average LOD values of 2.65–6.61. QKw-1A, QKw-1B, QKw-2A, and QKw-4A.2 were major QTLs that could be identified in only one environment. QKw-4B.1, QKw-4B.2, and QKw-6B were major QTLs that could be identified reproducibly in two, four, and four different environments, respectively. QKw-2D, the only stable QTL for KW, was verified in nine of the ten environments and individually explained 9.05 % (5.11–15.36 %) of the phenotypic variation, with LOD values ranging from 2.12 to 10.54. The remaining QTLs showed moderate additive effects and were unstable across the environments. Seven and six QTL alleles that increased KW were contributed by KN9204 and J411, respectively (Table 3; Supplementary Table S4; Fig. 2).

Thirteen QTLs associated with KDR were detected that individually explained an average of 4.12–12.36 % of the phenotypic variation, with average LOD values of 2.20–6.37. These QTLs were located on chromosomes 1B, 2A, 2B (2 QTLs), 2D, 3D, 4A (2 QTLs), 4B (2 QTLs), 5B, 6B, and 7A. QKdr-2A, QKdr-2D, and QKdr-4B.2 were stable QTLs that were detected reproducibly in six, six, and five environments, respectively. Additionally, QKdr-4B.2 individually explained 12.36 % (4.82–18.82 %) of the phenotypic variation, with a LOD value of 6.37 (2.88–10.98), as the only major stable QTL for KDR. Although QKdr-7A individually explained an average of 10.92% of the phenotypic variance with an average LOD value of 3.05, it could be identified in only two different environments. The remaining QTLs had moderate additive effects and were unstable across environments. KN9204 and J411 contributed six and seven QTL alleles that increased KDR, respectively (Table 3; Supplementary Table S4; Fig. 2).

Up to 17 QTLs for TKW were identified in the ten individual environment QTL mapping analyses. These QTLs were located on chromosomes 1A, 1B (2 QTLs), 2A (2 QTLs), 2B, 2D, 3B, 4A, 4B (2 QTLs), 5D, 6A, 6B (2 QTLs), 7A, and 7D, and they individually explained an average of 3.19–13.85 % of the phenotypic variation, with average LOD values of 2.15–7.99. QTkw-2D, QTkw-4A, QTkw-4B.2, and QTkw-5D were stable QTLs that were verified in ten, eight, five, and five different environments, respectively. Moreover, both QTkw-2D and QTkw-4B.2 were major stable QTLs individually accounting for an average of > 10.00 % of the phenotypic variance, with average LOD values of > 3.00. QTkw-6B.2 was the third major QTL, but it was unstable across the environments. QTkw-6A could be verified in four different environments with a moderate additive effect. The remaining QTLs individually explained < 10.00 % of the phenotypic variance and were unstable across the environments. Nine and eight QTLs alleles that increased TKW were contributed by KN9204 and J411, respectively (Table 3; Supplementary Table S4; Fig. 2).

2 Discussion

2.1 Would indirect selection of superior genotypes under normal conditions maximize the genetic gain in breeding programs designed to improve tolerance to nitrogen deficient conditions?

In the present study, indirect selection for TKW under HN resulted in the selection of seven of nine (77.8 %), 12 of 19 (63.2 %), and 15 of 28 (53.6 %) common RILs at 5, 10, and 15 % selection intensities, respectively. The Spearman’s rank correlation coefficient calculated for TKW between the treatments was second (0.79) to that of KL among the 11 traits investigated (Table 2, Fig. 1). In addition, TKW showed higher heritability under both HN and LN and was less significantly or not significantly negatively affected by N stress (Supplementary Tables S2 and S3). The top five RILs had identical or similar TWK under LN and HN (50.78–53.71 and 50.10–53.64 g, respectively). This finding indicates that indirect selection under HN can maximize the genetic gain for the improvement of tolerance to N stress for TKW. Similar inferences were also made for KL, KW, and KDR because they showed strong correlations with TKW, but indirect selection under HN resulted in the selection of 50 % or fewer common RILs selected at the three selection intensities (Tables 1 and 2; Fig. 1).

Despite the moderate Spearman’s rank correlation coefficients between the N treatments for DT and WGC, direct selection under each N treatment resulted in the selection of one-half or more of the RILs as common RILs at all three selection intensities (Table 2; Fig. 1). In addition, both traits showed higher heritabilities under both HN and LN, and they were consistently negatively affected by N stress across the tested environments (Supplementary Tables S2 and S3). The top five RILs had higher DTs and higher WGCs under HN than under LN (173.38–177.79 vs. 155.27–163.73 for DT and 35.84–37.03 vs. 27.87–33.33 for WGC, respectively). This finding indicates that indirect selection under HN allows for the genetic improvement of DT and WGC under LN in most cases; however, the negative effects of N stress on both DT and WGC were difficult to counteract.

Although the Spearman’s rank correlation coefficient for ABS was relatively high (0.70) between the treatments, direct selection under each N treatment resulted in the selection of more than half of the RILs as common at only a 15 % selection intensity (15 of the 28 lines were selected as common). However, ABS was less negatively affected by N stress, and the top five RILs had a slightly higher or similar ABS under HN compared with that under LN, (61.96–63.16 under HN and 60.96–62.34 under LN, respectively) (Supplementary Tables S2 and S3). This finding indicates that indirect selection under HN allows for the genetic improvement of ABS under LN to some extent.

For GPC, TW, ZEL, and KH, indirect selection under HN resulted in the selection of fewer than 50 % of the RILs as common at the three selection intensities, indicating the lower efficiency of indirect selection under HN for the genetic improvement of these traits under LN (Tables 1 and 2; Fig. 1).

2.2 QTL co-segregation with known genes

KN9204 and J411 differed at two HMW glutenin loci (Glu-A1 and Glu-B1) and two LMW glutenin loci (Glu-A3 and Glu-B3). Only one moderately stable QTL for DT (QDt-1A.2) mapped to a location near the Glu-A1 locus, which was likely due to the positive effect of the 2*-HMW subunit encoded by the Glu-1Ab allele on J411. QGpc-1B.2 and QWgc-1B.2 mapped to chromosome 1BL in the vicinity of Glu-B1. Additionally, three QTLs for kernel size (QKl-1B, QKdr-1B, and QTkw-1B.2) also mapped to this region, and alleles from KN9204 in this region (Bx7+By9) increased GPC and WGC but reduced kernel size, which might have contributed to the increase in the grain nitrogen concentration at the expense of grain biomass accumulation (Kunert et al. 2007). Moreover, the Bx7+By9 allele from KN9204 increased the tolerances of GPC, WGC, KL, and KDR to low N stress (Table 5). Two stable QTLs, QGpc-1A and QWgc-1A, along with an environmentally sensitive QTL, QDt-1A.1, mapped to locations near the Glu-A3 locus. The alleles from KN9204 in this region (Glu-A3b or Glu-A3d) simultaneously increased GPC, WGC, and DT. In addition, J411 had the Glu-B3h alleles, whereas KN9204 had a translocation of 1BL.1RS, resulting in the loss of Glu-B3. However, no QTLs for KRTs mapped to any locations near the Glu-B3 locus, indicating that this translocation had no adverse effects on the quality traits. In fact, Johnson et al. (1999) have concluded that genetic background and environmental factors likely affect the milling and baking quality to a greater extent than the 1BL.1RS translocation.

TaCwi was mapped to 164.7 cM on chromosome 2AL in C3, which was found to harbor ten QTLs for both kernel size and kernel quality, and alleles from J411 enhanced all ten kernel related-traits, including ABS, WGC, KH, GPC, DT, ZEL, KDR, KW, KL, and TKW (Table 6; Supplementary Fig. S1). These findings indicate that the TaCwi-A1a allele may be associated with not only a greater kernel size but also enhanced kernel milling and baking quality. Moreover, TaCwi-A1a from J411 enhanced the tolerances of ABS, ZEL, and KH to low N stress and reduced those of KDR and TKW (Table 5). However, a QTL for sensitivity of wheat yield per plant to N stress (QYddv-2A.1-2) was mapped to this region, with alleles from J411 enhancing N stress tolerance (Cui et al. 2014a). These findings indicate that TaCwi-A1a enhances the tolerances of both quality and yield to N stress.

TaSus2 was mapped to 95.2 cM on chromosome 2BS in C4, which was found to harbored five QTLs for both kernel size and kernel quality. Alleles from KN9204 increased KW and TKW but decreased GPC, WGC, and KDR (Table 6; Supplementary Fig. S1). In addition, a QTL for the per-plant wheat yield (QYd-2B) was mapped to this region, and alleles from KN9204 contributed to an increased phenotypic value (Cui et al. 2014a). This finding indicates that TaSus2 has opposing effects on yield and quality traits, which might contribute to the dilution effect between grain nitrogen concentration and grain biomass accumulation (Kunert et al. 2007). Two QTLs for KLDV (QKldv-2B) and KWDV (QKwdv-2B) were identified in this chromosomal region, and alleles from KN9204 increased the phenotypic value for both KLDV and KWDV. These results indicate that TaSus2 can influence the responses of KL and KW to N stress. However, no QTL for differences between the values for per-plant wheat yield under HN and LN was identified within this chromosomal region (Cui et al. 2014a). These results indicate that TaSus2 has no effect on the response of per-plant yield to N stress (Table 5; Supplementary Table S5).