Genetic biofortification is an effective method for enhancing crop microelement contents. A large number of genetic loci controlling microelement accumulation in wheat, rice and other crops have been identified in previous research [22–25]. However, there are relatively few studies on the genetic mechanisms of Ca ion accumulation in wheat grain; consequently, limited information is available on wheat grain genetic control and molecular physiological mechanisms. GWAS is a powerful tool for dissecting the genetics of complex traits and identifying the chromosomal regions harboring genes suitable for use in breeding programs. In this study, a GWAS was used to dissect the genetic basis of Ca accumulation in wheat grain using a natural population.
When using a GWAS, the probability of detecting the causal variant and associated loci for a target trait depends on the marker density, population size and statistical methods [26]. Owing to the rapid development and application of molecular marker assays, Wheat 35K, 90K, 660K and 820K SNP genotyping arrays have been designed and utilized for GWAS and linkage analyses in common wheat [27–29]. Comparative analyses revealed that the Wheat 660K SNP array is reliable and cost-effective, making it the ideal choice for genotyping a population of individuals [30]. In the present study, a credible number of markers (244,508 SNPs) was identified using the Wheat 660K SNP array, and the population met the requirements of a GWAS for Ca accumulation in wheat grains. Population size is another factor limiting the detection efficiency of a GWAS. The effect of increasing the population size on QTL detection efficiency is greater than that of the marker density [31], and increasing the population size may lead to the identification of more smaller-effect QTLs [32]. Previously, population sizes ranging from 100 to 500 have been used for wheat association analyses [33–35]. In this study, although the natural population of 207 diverse accessions was not sufficiently large, the dramatic phenotypic variations in grain Ca concentration was very large, ranging from 121.71 mg/kg to 685.17 mg/kg, and it showed a normal distribution, which is conducive to GWAS. Dramatic phenotype variation may be associated with high genetic diversity [36]. In this study, a population structure analysis based on the genotype data of the natural population interpreted the diversity from another perspective. Our population could be divided into eight subpopulations (Figure 1), which indicated high genetic diversity and suitability for GWAS. The lack of proper modifications to population structure may lead to spurious associations [37, 38]. To eliminate such associations, the population structure, as shown in the Q matrix, was considered as a fixed-effect factor in the GWAS. Additionally, complex traits are sensitive to different statistical models, which vary in their abilities to control type I or type II errors. Thus, a suitable statistical model can effectively increase the confidence of the associated sites. For example, Chen et al. (2017) analyzed 13 agronomic traits of wheat and found that the MLM model effectively controls false positive and false negative errors, making it the best model for a GWAS [39]. In a maize GWAS for arsenic accumulation, the GLM model (only controls the influence on the population structure) was better than the MLM model [40], although it could not completely control the population structure influence. In the present study, the QQ plots indicated that the influence of type I and type II errors was controlled in the GWAS for wheat grain Ca accumulation by MLM [simultaneously controls the influence of fixed-effect factors (Q matrix) and random-effect factors (K matrix)] better than the GLM and FarmCUP statistical models. These results were in accordance with previous GWAS studies for Ca accumulation in wheat grain, in which the MLM model was used for the GWAS [20].
The manifestations of complex quantitative traits (such as microelement accumulation in wheat grain) are often controlled by multiple genetic loci [21, 41, 42]. Several genetic loci affecting Ca accumulation in wheat grain were identified previously by GWAS and linkage analyses [13, 20, 21], which allowed for a comparison among loci identified in known QTLs and those identified in the present study. Here, 18 non-redundant loci for Ca accumulation were distributed on chromosomes 1D, 2A, 2B, 3A, 3B, 3D, 4A, 4B, 5A, 5B, 6A, 7A and 7B, which suggested the involvement of loci on these chromosomes in the natural population regulating Ca concentration variation. QTLs for Ca accumulation are scattered on chromosomes 4A, 4B, 5B, 6A and 7B [13] in a durum wheat × wild emmer RIL population, which overlapped with loci identified in the current study, indicating that the linkage mapping results were complementary to those of the GWAS for wheat grain Ca accumulation. Bhatta et al. (2018) identified 15 significant marker-trait associations for wheat grain Ca accumulation distributed in 14 different genomic regions on chromosomes 1B, 2B, 2D, 3A, 3B, 3D, 6A, 6B and 7A that explained 2.7–21.5% phenotypic variation using the Hard Winter Wheat Association Mapping Panel (including 299 varieties) and the GWAS method [21]. We only identified eight loci on chromosomes 2B, 3A (2 loci), 3B (2 loci), 3D, 6A and 7A, and not all of the QTLs detected previously were identified in this study. This may be because of (a) the different origins of the populations, or (b) the use of different genotypic identification platforms. This can make it difficult to align the complete genomes of the population’s individuals. It is worth noting that, like other complex quantitative traits, the accumulation of Ca in wheat grain is controlled by multiple genetic sites and is susceptible to environmental influences [43]. Therefore, the ideal target genetic loci should be stably identified under multi-environmental conditions. In this study, we found that all six loci were detected in at least two environments with relatively higher PVE values (9.66–26.93%), which suggests that these were stable QTLs significantly associated with wheat grain Ca accumulation that were critical for target trait phenotypic variation. The peak SNP AX-110634514 has been mapped closely to Kukri_c40035_258 on chromosome 2A [20], and AX-109541359 is co-located in the vicinity of regions harboring S3A_593702925 and CFA2262-wPt_3816 on chromosome 3A [21, 44]. On chromosome 3B, the SNPs AX-110013515 and AX-110922471 were mapped to genomic regions near gwm389-wPt-8093(C) and QGCaUE-3B, respectively. The former QTL has been identified in a natural wheat population [44], and the latter QTL has been detected in a double-haploid wheat population [42]. The peak SNP AX-94729264 was simultaneously identified in SQ, YY, and BLUP, which indicated it co-localizes with S3D_45073985 [21]. These findings validate the results of the GWAS and increase the confidence in some loci identified in the present study. The hotspot at the end of chromosome 4A linked with SNP AX-108912427 (identified in all the environments with the highest PVE values, ranging from 12.89–26.93%) was simultaneously mapped in the vicinities of three QTLs, gwm165a-wmc420, wmc106-gwm165a and gwm610 [13, 44], which implies that this hotspot is a key factor that harbors a major gene for regulating Ca accumulation in wheat grain.
In this study, six peak SNPs, together with corresponding non-redundant QTLs that are associated Ca accumulation, were identified by the GWAS in at least two environments. Combined with the gene’s physical positions and functional annotation information, six genes were identified as the most credible candidate genes for Ca accumulation. Two SNP markers, AX-108912427 on chromosome 4A and AX-110922471 on chromosome 3B, were associated with genes encoding V-type proton ATPase subunit e (TraesCS4A01G428900) and V-type proton ATPase subunit d (TraesCS3B01G241000), respectively. Both subunits are important components of the V-type proton ATPase that is the typical generator of a proton gradient involved in Ca ion sequestration in plant cells, and it may influence Ca ion homeostasis in wheat grains [45, 46]. The hot spot QTL linked with 39 significant SNPs at the end of the 4A chromosome. The peak SNP, AX-108912427, has the highest average PVE at 21% for Ca concentration and was identified in all the environments, which implies that V-type proton ATPase is a key factor affecting Ca accumulation in wheat grain. Another marker on chromosome 3B, AX-110013515, corresponds to a transmembrane protein (TraesCS3B01G018800). This is a novel resident protein on the endoplasmic reticulum (ER) and has an ER luminal domain containing rich acidic residues [47]. This amino-terminal luminal domain has a high capacity, and moderate affinity, for binding Ca ions, and it plays an essential role in ER Ca ion handling [47], which may be closely associated with Ca accumulation in wheat grain. The SNP marker AX-109541359 on chromosome 3A was associated with the gene TraesCS3A01G424700, which encodes cytochrome P450. The sequential actions of cytochrome P450 may result in the synthesis of a pleiotropic hormone that regulates several genes that have actions associated with Ca homeostasis as well as cellular growth [48]. The remaining SNPs, AX-110634514 on chromosome 2A and AX-94729264 on chromosome 3D, were associated with candidate genes TraesCS2A01G098100 and TraesCS3D01G079600, respectively. The former encodes calmodulin, the predominant Ca receptor with a flexible conformation that regulates both fluxes of transient Ca-ion influx through Ca-ion channels and Ca-ion efflux by Ca transporters [49]; calmodulin also has the capacity to regulate Ca-ion pumps in multiple subcellular locations [50, 51]. The latter encodes a ubiquitin family protein involved in the process of ubiquitin conjugation. Mutating ubiquitin proteins may alter cell coupling and the resulting Ca elevation [52], which implies that ubiquitin family proteins may participate in altering Ca homeostasis in wheat. These results reveal the complex nature of Ca accumulation in wheat grain and imply that various mechanisms are involved in controlling Ca accumulation.
Organismal Ca requirements must be met through dietary uptake. The Ca intake in the adult population of Asia has been reported to be less than 500 mg/day, and in Africa and South America the Ca intake of the adult population is between 400 and 700 mg/day [15]. These values are far below the recommended standards of the Food and Agriculture Organization, which are 1,300 mg/day for children over 9 years of age and 800–1,300 mg/day for adults. Biofortification is an effective strategy to increase the microelement content of wheat and improve the human intake of Ca. However, early breeding pro-grams mainly focused on yield and ignored the microelement levels. Because of the so-called " dilution effect " in which high yields are negatively correlated with micro-element levels [3], it is difficult to select wheat varieties with high Ca contents at high yield levels using traditional breeding methods. Identifying superior allele loci and developing corresponding molecular markers have been beneficial to pyramid breeding, and this strategy could significantly enhance the microelement, including Ca, levels in wheat grain [53–55]. In this study, 6 of 18 loci were identified as harboring superior alleles and exhibited significantly higher Ca accumulations in two or more environments. By comparing the abilities of lines with both superior and inferior alleles for Ca accumulation in wheat grain at the six loci, we found that phenotypic differences reached significant levels between individuals with either superior or inferior alleles (Figure 4, Table 4). Additionally, a linear regression showed that with an increase in the number of favorable alleles, the Ca concentrations of accessions gradually increased, revealing a significant additive effect, which provides guidance for pyramid breeding (Figure 5). Markers identified by the GWAS that were significantly linked to these loci and associated with wheat grain Ca concentration may be converted into competitive allele-specific PCR markers for molecular marker-assisted selection-based breeding programs [56]. Using marker-assisted selection, the superior alleles for Ca accumulation could be integrated for multi-loci pyramid breeding, which will provide guidance for biofortification breeding. In the future, our studies will focus on validating the effects of these QTLs and developing molecular markers for wheat Ca biofortification pyramid breeding.