The hard grain mutant of Japanese elite wheat cultivar “Kitahonami”
To confirm the capability of whole-genome resequencing approach to detect large deletions caused by gamma irradiation in the hexaploid wheat, we focused on grain hardness because grain hardness is regulated by Pina-D1 and Pinb-D1 genes, loss of which causes the grain hardness change from soft to hard [15, 16]. Given that “Kitahonami” produces soft grains, the hard grain mutant was assumed to have mutations in the Pina-D1 and Pinb-D1 regions on the distal region of short arm of chromosome 5D. Based on the grain colour and the Single Kernel Characterization System (SKCS) value, we obtained the grain hardness mutant “30579” from the gamma-irradiated mutant population of “Kitahonami” (Fig. 1). Whereas the SKCS value of the wild-type is 24, the grain hardness of “30579” showed 102 of SKCS value (Table 1). Typically <40 and >70 of SKCS value is regarded as soft and hard grain, respectively. Therefore, the grains of mutant “30579” is hard. Grain protein content was also slightly elevated in the mutant (Table 1). Agronomic characteristics of the mutant “30579” were also compared with those of the wild-type. Grain weight (kg/10a) and 1000-grain weight of the mutant “30579” were low to those of the wild-type, indicating that the yield of the mutant was inferior to that of the wild-type.
Table 1 Agronomic and quality characteristics of “Kitahonami” and the mutant “30579”
|
Kitahonami WT
|
30579
|
Agronomic characteristics
|
|
|
Maturity stage (date)1
|
28th July
|
27th July
|
Grain weight (kg/10a)
|
687
|
578
|
Ratio (%)
|
100
|
84
|
1000-grain weight (g)
|
41.2
|
30.4
|
|
|
|
Quality characteristics
|
|
|
Grain Hardness
|
24
|
102
|
Grain protein content (%)
|
9.4
|
10.2
|
1Maturity stage was examined in 2019-2020 season.
Comparative genomics of “Kitahonami” and its hard grain mutant “30579” revealed large deletions induced by gamma irradiation
To identify the causal genome region of hard grain of the “Kitahonami” mutant “30579,” genome resequencing of the wild-type and the mutant “30579” were performed. In total, 2.3 to 2.9 billion reads were obtained. After removing short reads with bad quality and PCR duplicates, we obtained 1.8 to 2.3 billion reads aligned to the reference genome of CS (Table 2). Average depth-of-coverage was 15.51 for the wild-type and 18.85 for the mutant. The aligned short reads covered over 94% of the reference genome sequences. The number of single-nucleotide polymorphisms (SNPs) and short indels between “Kitahonami” and CS were 30,597,871 and 1,328,442, while the number of SNPs and short indels between the mutant “30579” and CS were 32,254,934 and 1,713,770.
Table 2 Summary of alignments and coverages of genome sequencing of wild-type and mutants of “Kitahonami”
Name
|
Total filtered reads (%)1
|
Aligned reads (%)2
|
Aligned reads after removal of PCR duplicate (%)3
|
Reference bases covered (%)
|
Average
depth-of-coverage
|
Kitahonami
|
2,736,865,184 (94.05%)
|
2,732,548,516 (99.84)
|
2,334,492,027 (85.30%)
|
95.99
|
15.51
|
Kitahonami (5x)4
|
933,526,244 (99.51%)
|
932,129,013 (99.85%)
|
876,039,611 (93.84%)
|
94.38
|
6.01
|
30579
|
2,042,471,104 (87.84%)
|
2,036,948,509 (99.73%)
|
1,830,711,894 (89.63%)
|
94.57
|
18.85
|
28511
|
2,107,121,470 (78.79%)
|
2,104,170,687 (99.86%)
|
1,908,476,075 (90.57%)
|
95.39
|
19.43
|
28511 (15x)4
|
1,626,684,636 (78.79%)
|
1,624,412,275 (99.86%)
|
1,498,723,901 (92.13%)
|
95.20
|
15.35
|
28511 (10x)4
|
1,084,461,526 (78.79%)
|
1,082,945,323 (99.86%)
|
1,025,292,441 (94.68%)
|
94.86
|
10.50
|
28511 (5x)4
|
542,210,918 (78.79%)
|
541,454,612 (99.86%)
|
526,465,742 (97.23%)
|
93.63
|
5.39
|
28511 (2.5x)4
|
271,156,212 (78.79%)
|
270,777,750 (99.86%)
|
266,912,156 (98.57%)
|
87.30
|
2.73
|
1 An average base quality score per 4bp >=20
The filtered reads rate = Total filtered reads / Total reads × 100
2 The aligned reads rate = Aligned reads / Total filtered reads × 100
3 The filtered reads rate = Aligned reads after removing PCR duplicates / Total filtered reads × 100
4 The sets of read data were used for the simulations. The number in the paratheses indicates approximate depth-of-coverage.
SNP density distribution over chromosome for the mutant “30579” was almost identical to the wild-type (Fig. 2). The SNP density for the wild-type and the mutant “30579” was unevenly distributed over the chromosomes. In chromosomes 1A, 2A, 1B, 2B, 3B, 5B, and 6B, “Kitahonami” was genetically divergent from CS, while the chromosomes 3A, 4A, 6A, 4B, and 7B were genetically close in the two cultivars, particularly in their proximal region. D genome chromosomes showed less divergence than the other genomes between “Kitahonami” and CS. Less SNP density in the 420-450 Mbp position of chromosome 2A and the short arm of chromosome 5D was uniquely detected in the mutant “30579.”
Large deletions are often observed in gamma-irradiated mutants [29, 30]. If a large deletion causes Pina-D1 and Pinb-D1 to be lost, a decrease of depth-of-coverage in the short arm of chromosome 5D should be uniquely observed in the mutant “30579.” To identify genomic regions where depth-of-coverage uniquely decreased in the mutant, we analyzed moving average of depth-of-coverage per three Mbp over the chromosomes (Fig. 3). Depth-of-coverage in the mutant was close to zero in 420-450 Mbp position of chromosome 2A, around 90 Mbp position of chromosome 4B, and 0-130 Mbp position of chromosome 5D, indicating that these chromosomal regions were uniquely deleted from “30579” mutant. The genome regions of the mutant “30579” showing less SNP density than the wild-type (Fig. 2) were consistent with these deleted regions. The distribution of difference of depth-of-coverage (∆depth) per three Mbp between the wild-type and the mutant “30579” was also visualized over the chromosomes (Fig. 3). Peaks beyond the 99% confidence interval were observed at the uniquely deleted regions on chromosomes 2A, 4B, and 5B of the mutant “30579.” Several sharp peaks of ∆depth over 99% confidence interval corresponded to regions with irregularly deep depth-of-coverage, where mapped reads were derived from repetitive sequences. Since the chromosome 5D region with the large deletion corresponded to the genome region containing Pina-D1 and Pinb-D1 [12], the hard grain mutant “30579” lost Pina-D1 and Pinb-D1.
Genotyping of the mapping population with indel markers confirmed causal regions of hard grains in mutant “30579”
We developed a F2 (n = 72) population of a cross between the wild-type and the mutant “30579” to evaluate segregation of grain hardness. The SKCS analysis was conducted to examine grain hardness of the seeds harvested from the F2 population. Of the tested lines, 22, 42, and eight lines showed soft, intermediate, and hard grain phenotypes, respectively (Fig. 4a). To validate which deletions were linked to the hard grain phenotypes, indel markers were designed for each deletion (Fig. 4b). Genotyping of 22 soft grain lines and eight hard grain lines was performed using the indel markers. The indel marker of chromosome 5D was linked to the hard grain phenotypes, whereas the indel markers of chromosomes 2A and 4B were not (Fig. 4c). This result indicates that the large deletion on chromosome 5D caused “Kitahonami” to change from soft to hard.
Characteristics of the PHS-tolerant mutant of “Kitahonami”
A PHS-tolerant mutant, called “28511,” was selected from “Kitahonami” mutant population. The PHS tolerance test showed that the mutant “28511” was more tolerant to PHS than the wild-type (Fig. 5, Table 3). PHS tolerance of the mutant “28511” was comparable to that of “Kitakei 1831,” used as a control variety for PHS tolerance. In the seed dormancy test, the mutant “28511” showed a lower germination rate than the wild-type under both 10 °C and 15 °C conditions in three seasons (Table 3). The mutant “28511” had higher germination rate than “Kitakei 1831” in 2016-2017 season whereas had similar germination rate to “Kitakei 1831” in 2017-2018 season. Agronomic and quality characteristics of the mutant “28511” were also evaluated (Table 4). Maturity stage, flour yield, and flour color of the mutant “28511” were almost the same as those of the wild-type. However, grain weight (kg/10a) and 1000-grain weight of the mutant “28511” were inferior to those of the wild-type. Moreover, the mutant “28511” showed higher flour protein content than those of the wild-type.
Table 3 The tests of PHS tolerance and seed dormancy of wild-type “Kitahonami” and the mutant 28511
Season
|
Name
|
PHS tolerance test
PHS level
(0 — 5)
|
Seed dormancy test
Germination rate (%)
|
15 °C
|
10 °C
|
2016/2017
|
Kitahonami WT
|
1.1
|
64.7
|
96
|
|
28511
|
0.3
|
27.5
|
83.7
|
|
Kitakei 1838
|
0
|
8
|
54
|
2017/2018
|
Kitahonami WT
|
0.4
|
83.7
|
82.4
|
|
28511
|
0.1
|
20
|
45.8
|
|
Kitakei 1838
|
0.1
|
18
|
48
|
2019/2020
|
Kitahonami WT
|
—
|
72.0
|
82.0
|
|
28511
|
—
|
13.7
|
26.0
|
Table 4 Agronomic and quality characteristics of wild-type “Kitahonami” and the mutant “28511”
|
Kitahonami WT
|
28511
|
Agronomic characteristics
|
|
|
Maturity stage (date)1
|
25th July
|
26th July
|
Grain weight (kg/10a)
|
747
|
646
|
Ratio (%)
|
100
|
87
|
1000-grain weight (g)
|
41.5
|
36.9
|
|
|
|
Quality characteristics
|
|
|
Grain protein content (%)
|
11.2
|
13.0
|
Flour ash content (%)
|
1.01
|
1.07
|
Flour yield (%)
|
71.0
|
70.1
|
Flour color L*
|
87.34
|
86.55
|
a*
|
-0.94
|
-0.56
|
b*
|
15.43
|
14.38
|
1Maturity stage was examined in 2016-2017 season.
Whole genome resequencing of the PHS-tolerant mutant “28511” identified a large deletion on the long arm of chromosome 3B
To identify the causal region of PHS tolerance of the mutant “28511,” genome sequencings of the mutant were performed in the same method as the hard grain mutant “30579.” Of the qualified 2.1 billion reads, 99.9% were successfully aligned to the reference genome of CS (Table 2). Average depth-of-coverage for the mutant was 19.43 for the mutant. The number of SNPs and short indels between the mutant “28511” and CS were 35,368,664 and 1,865,274, respectively. Uneven distribution of SNP density over chromosome for the mutant “28511” was also observed as shown in that of the wild-type and the mutant “30579” (Additional file 2: Fig. S1).
To detect deletions in the mutant “28511,” moving averages of depth-of-coverage per 3 Mbp and ∆depth per 3 Mbp were calculated over the chromosomes. At around 700 Mbp position of chromosome 3B, depth-of-coverage uniquely decreased in the mutant “28511” (Fig. 6). ∆depth also showed above 99% confidence interval. The reduced area extended to 67.8 Mbp, indicating that the mutant “28511” had a large deletion at the long arm of chromosome 3B. Such a remarkable reduction of depth-of-coverage was not observed in the other chromosomes.
Association between the large deletion at chromosome 3B and PHS tolerance
To validate the large deletion at chromosome 3B, we constructed a co-dominant marker to detect the deletion (Fig. 7a, Additional file 1: Table S1). Primer pairs (pre- and post-deletion primers) were designed outside the deletion boundary. Another primer (in-deletion primer) was designed inside the deletion boundary. In wild-type, a 712 bp PCR fragment amplified by in-deletion primer and post-deletion primer was observed, whereas in the mutant, a 414 bp PCR fragment amplified by pre-deletion primer and post-deletion primer was observed (Fig. 7b). When the CS nulli-tetrasomic line of nulli-3B tetra-3D, the chromosome 3B of which is replaced with chromosome 3D, was used, no PCR fragment was detected, indicating that this marker was specific to the deletion detected on chromosome 3B.
To confirm whether the large deletion was the causal genetic factor of PHS tolerance, we tested germination rate under the three conditions, three, seven, and nine days at 15 °C after seed sowing, for F2 segregation population from a cross between the wild-type and the mutant “28511.” Also, we examined the genotype for the F2 population by using the co-dominant marker, whether the large deletion was associated with low germination rate. Under every condition, the wild-type genotype had significantly high germination rate, followed by the heterozygous genotype, and the mutant genotype had the lowest germination rate (Fig. 7c). This result indicates that a large deletion was associated with the low germination rate, which could facilitate PHS tolerance of the mutant “28511.”
Since ABA sensitivity rather than the ABA level in seeds regulates seed dormancy, wheat plants with low ABA sensitivity decrease seed dormancy, resulting in PHS [31]. To test the ABA sensitivity of the wild-type and the mutant “28511,” suppression rate of germinated root elongation under exogenous ABA treatment was examined. No significant difference in the suppression between “Kitahonami” and the mutant “28511” was detected (Fig. 7d), implying that they have similar ABA sensitivity.
Vp-B1, associated with PHS tolerance [25, 26], was located in the long arm of chromosome 3B where the large deletion was detected in the mutant “28511.” A gene encoding GRAS family transcription factor was also found as a candidate gene of PHS tolerance in this region. Gibberellic acid (GA) promotes seed germination by inducing biosynthesis of α-amylase and protease in aleurone layers [32]. GRAS family transcription factor SCARECROW-LIKE 3 (SCL3) is a positive regulator of GA signaling in A. thaliana [33]. We designed two markers for each gene to test whether these genes were lacking in the mutant (Fig. 8a). The markers for Vp-B1 (Vp-1B_2 and Vp-1B_3) showed no amplification in the mutant “28511” and the nulli-3B tetra-3D line. The markers for GRAS family transcription factor (GRAS-TF_2 and GRAS-TF_3) exhibited no amplification in the either the mutant “28511” or the nulli-3B tetra-3D line (Fig. 8b). These results confirmed that Vp-B1 and GRAS family transcription factor were deleted in the mutant “28511.”
Verification of detection ability in the gamma-irradiated wheat mutant with whole genome sequencing
To reduce resequencing cost, it is essential to know how much depth-of-coverage is necessary to detect large deletions in a gamma-irradiated mutant genome. By subsampling short reads and adjusting average depth-of-coverage per genome, we evaluated the deletion detection power of the method based on depth-of-coverage, visualized over chromosomes. Four depth-of-coverage conditions, 2.5x, 5x, 10x, and 15x were tested using short reads from the PHS mutant “28511” and the wild-type (Fig. 9). The distributions of moving average of depth-of-coverage and ∆depth over the chromosomes were almost identical among all the conditions. The 67.8 Mbp deletion on chromosome 3B was significantly detected under all the depth-of-coverage conditions, implying that over 2.5x depth-of-coverage was enough to detect the large deletion in the gamma-irradiated wheat mutant.
In addition, to clarify how long deletions could be detected under the 5x depth-of-coverage conditions, simulation of deletion length was conducted by changing deletion length to 40 Mbp, 20 Mbp, 10 Mbp, 5 Mbp, 3 Mbp, and 1 Mbp (Fig. 10). When the deletion length was 1 Mbp or more, ∆depth was beyond 99% confidence interval. The length of the detectable deletion was found to depend on the window size of moving average. If window size was larger than the length of targeted deletion, the detectability of deletion decreased. For example, when the target deletion length was 3 Mbp, the estimated ∆depth for the 3 Mbp or 1 Mbp window size was over 99% confidence interval, but the estimated ∆depth for more than 3 Mbp window size was not. Another peak of ∆depth around 200 Mbp position on the chromosome. This detected peak was caused by repeats derived from transposable elements. By confirming both distributions of depth-of-coverage and ∆depth, such irregular peaks can be distinguished from deletions.
Characterization of mutations detected in the gamma-irradiated wheat mutants
Nucleotide substitutions and small indels with less than 25 bp, which was the maximum length detected by indel calling of bcftools, were also detected in the gamma-irradiated wheat mutants (Table 5, Additional file 1: Table S2). Between the wild-type and the grain hardness mutant “30579,” 2,412 SNPs and 329 small indels were detected. Between the wild-type and the PHS-tolerant mutant “28511,” 2,715 SNPs and 266 small indels were detected. Gamma irradiation is assumed to cause these SNPs and indels. The ratio of transition and transversion (Ts/Tv ratio) between the wild-type and the mutants is lower than that between the wild-type and CS and is regarded as a natural variation. Over 98% of SNPs were detected in the intergenic regions. SNPs between the wild-type and the mutants were more randomly distributed over chromosomes compared with SNP density between wild-type “Kitahonami” and CS (Fig. 2, Fig. 11, Additional file 2: Fig. S1). The limited natural variations on chromosomes 3A, 4A, 5A, 6A, 4B, 7B, and all the D genome chromosomes were observed between these two cultivars, while the putative mutations induced by gamma irradiation covered these chromosomes.
Table 5 Comparisons between putative gamma-irradiated and natural SNPs
|
Wild-type vs. 30579
(Hard grain mutant)
|
Wild-type vs. 28511
(PHS tolerance mutant)
|
Wild-type vs. CS
|
Total
|
2,412
|
2,715
|
16,850,714
|
Transition
|
1,570
(65.091)
|
1,710
(62.983)
|
12,023,124
(71.351)
|
Transversion
|
842
(34.909)
|
1,005
(37.017)
|
4,827,590
(28.649)
|
Ts/Tv ratio
|
1.865
|
1.701
|
2.491
|
Exon
|
11
(0.455)
|
12
(0.442)
|
110,591
(0.651)
|
Synonymous
|
3
(0.124)
|
2
(0.074)
|
47,365
(0.279)
|
Nonsynonymous
|
8
(0.331)
|
10
(0.368)
|
63,226
(0.372)
|
Intron
|
15
(0.620)
|
20
(0.736)
|
252,919
(1.489)
|
UTRs
|
4
(0.165)
|
2
(0.074)
|
51,665
(0.304)
|
Intergenic
|
2,389
(98.760)
|
2,683
(98.749)
|
16,565,968
(97.555)
|
High impact variants1
|
|
|
|
Total
|
0
(0.000)
|
1
(0.037)
|
1,696
(0.010)
|
Nonsense mutations
|
0
(0.000)
|
1
(0.037)
|
856
(0.005)
|
Start codon lost
|
0
(0.000)
|
0
(0.000)
|
93
(0.001)
|
Stop codon lost
|
0
(0.000)
|
0
(0.000)
|
282
(0.002)
|
Splice sites
|
0
(0.000)
|
0
(0.000)
|
465
(0.003)
|
Percentage is shown in parentheses.
1Variant types are defined in SnpEff (Cingolani et al. 2012).