The Hot QTL Locations for Potassium, Calcium, and Magnesium Nutrition and Agronomic Traits at Seedling and Maturity Stages of Wheat under Different Potassium Treatments

Potassium (K) is one of the most important mineral nutrients for wheat. In this study, the effects of low K (LK) treatments and the quantitative trait loci (QTLs) for K, calcium (Ca), and magnesium (Mg) use efficiency traits, both at the seedling and maturity stages of wheat, were investigated. The set of “Tainong 18 × Linmai 6” recombinant inbred lines (RILs) were used to identify the QTLs under different K treatments using hydroponic culture and field trials. The majority of K concentrations and content-related traits at seedling and maturity stages decreased with reduced K supply, but the K use efficiency-related traits increased. In contrast, with reduced K supply, the contents of Ca and Mg increased, while the Ca and Mg use efficiency decreased. A total of 217 QTLs for seedling traits and 89 QTLs for adult traits were detected. Four relatively high-frequency QTLs (RHF-QTLs) and 18 QTL clusters (colocation of QTLs for more than two traits) were detected. Eight clusters were detected for K-, Ca-, and Mg-related traits simultaneously. This means that these traits might be controlled by the same QTL. In addition, we highlight that 4B might be an important chromosome regulating the nutrition of K, Ca, and Mg in wheat. The 4B chromosome and four hot QTL clusters, which located 45 QTLs, might be important potential targets for further investigation.


Introduction
Wheat (Triticum aestivum L.) contributes one-third of the world's edible dry matter. It is one of the most important grain crops in the world. Potassium (K) is one of the essential macronutrients for crops, which is not only important for crop growth, development, and fecundity, but also significant for crop yield and quality [1]. It can increase the salt, drought, and disease tolerance of plants [2][3][4][5][6][7][8]. The average reserves of K in soil are usually large. However, most of the K in soil is not plant-available, and K deficiency is one of the most common limiting factors for crop production [9].
Potassium deficiency can significantly affect the use of K or other elements and ultimately affect crop yield [3,10]. There are complex interactions among K, calcium (Ca), and magnesium (Mg). Potassium can reduce the uptake of Mg in numerous plant species, such as soybeans, wheat, and rice [11][12][13]. However, the mechanisms for this K-inhibited Mg uptake have not been researched clearly. One possible explanation is the competition for apoplast binding between K + and Mg 2+ [14], while another possible explanation is the competition for the unidentified transporters between them. Tomoaki et al. [15] suggested that OsHKT2;4 (a K + -permeable transporter/channel)-mediated currents could also exhibit permeability to both Mg 2+ and Ca 2+ , which would be smaller with the competitive lines were 3 cm × 3 cm. The nutrient solutions were continuously aerated and renewed every 4 days. The 0.1 mmol·L −1 HCl or NaOH solution was used to regulate the pH of nutrient solution between 6.0 and 6.2 every day. The plants grew for 28 days in nutrient solution and were harvested. The details referenced the method of Guo et al. [32].

Field Trial
The field trials were carried out at the agronomy experimental station of Shandong Agricultural University from 2012-2013. The soil type was loamy cinnamon soil (pH 7.8). The average contents of N (available N), P (Olsen P), and K (available K) in the 0 to 20 cm soil profile sampled were 58.2, 21.3, and 86.4 mg·kg −1 without fertilizing. Two K concentration treatments, moderate K (114 kg/hm 2 ) and low K (0 kg/hm 2 ), were designed. In addition, 50% of the total N (97.5 kg/hm 2 ), all the P 2 O 5 (102 kg/hm 2 ), and the corresponding K 2 O in two treatments were applied as base fertilizer before sowing, and the other 50% of N (97.5 kg/hm 2 ) was applied at the stem elongation stage. Each treatment was replicated twice. Twenty seeds of each line were sown on October 10, and ten seedlings were retained after germination, with a 10 cm spacing between plants and 25 cm between rows. All of the materials were harvested on June 10. Twenty plants of each line in the same K treatment were put together as one sampling during harvest and then threshing for further testing.

Trait Measurements
All of the investigated traits and their abbreviations are listed in Table 1. For hydroponic trials, the three replicates for each line (nine plants) of each treatment were pooled together as one mixed sample and separated into roots and shoots. After being dried at 105 • C for two hours and dried at 60 • C for 72 h in an oven, the dry weight and the concentrations of K, Ca, and Mg in roots and shoots were measured. The concentrations of K, Ca, and Mg in roots and shoots were determined using atomic absorption spectroscopy (AA7000) after microwave digestion using HNO 3 . K-use efficiency (RKUE) was the ratio between dry weight and the concentration of K in the corresponding part of plant. For example, root K-use efficiency (RKUE) was the ratio between root dry weight per plant and the concentration of K in root (RDW/RKCE). The measurement methods of other nutrient use efficiencies were similar to the root K-use efficiency.
For field trials, the plant height (PH), spike number per plant (SN), and grain number per spike (GN) were determined from five random plants for each replicate of each line. All of the plants for each line of one K treatment were pooled together and then measured for dry weight and K, Ca, and Mg concentrations of the straw and grain, separately. The concentrations of K, Ca, and Mg in grains and straws were determined using atomic absorption spectroscopy (AA7000) after microwave digestion using HNO 3 . The calculation methods of nutrient use efficiencies were similar to the root K-use efficiency in the hydroponic culture trial. The K harvest index (KHI) is the ratio between grain K content per square meter and aboveground K content per square meter (GKC/AKC). The measurement methods of Ca and Mg harvest indexes were similar to the KHI.

Data Analysis
The SPSS 18.0 software (SPSS Inc., Chicago, IL, USA) was employed to conduct the analyses of variance (ANOVA), the least significant difference (LSD) test, and Spearman's correlation coefficients (r) between different traits. In a no-repeat trial design, using a two-factor model was adequate for ANOVA. All factors including RILs (n − 1) degrees of freedom, treatments (t − 1), and random error ((n − 1)(t − 1)) were considered sources of random effects. Multiple comparison tests for the traits between "treatments" were calculated by taking all of the RILs as replicates and using the mean value of the same K condition for each trait. The variance of K conditions was excluded when the broad-sense heritability (h B 2 ) was estimated according to the formula: h B 2 = σg 2 /(σg 2 + σe 2 ), where σg 2 was the genotypic variance and σe 2 was the total error variance. The high-density genetic map for 184 RILs of "TN18 × LM6" [28] was employed in the QTL analysis. The map comprised of 10739 loci (5399 unique loci) assigned to 21 chromosomes, with a total map length of 3394.47 cM and a density of 0.63 cM/marker. The Windows QTL Cartographer 2.5 software (Http://statgen.ncsu.edu/qtlcart/wqtlcart.htm) [33] was used to perform the QTL mapping in this study. The presence of the significant QTL was declared via the threshold that was defined by 1000 permutations at p ≤ 0.05 [34]. The identification of QTL cluster and its confidence interval referenced the results of the meta-analysis (Biomercator 2.0 software, AIC = 4 (model 4) in the step Meta-analysis 2/2 (http://www.genoplante.com)) [29].

Naming Method of QTLs
QTLs were named according to the method of "Q + trait name + chromosome name + experimental treatment." Among them, traits are represented by their English abbreviation, and "−" was added between traits and chromosomes. QTLs for the same trait on the same chromosome were distinguished using an Arabic numeral (1, 2, 3, . . . ). In addition, E1 and E2 stand for the hydroponic trial of February 2013 and March 2013, respectively. T1 and T2 stand for CK and LK treatments, respectively.

Phenotypic Variation and Correlations Between Traits
The significant differences among most of the investigated traits of the RIL population were found in both the hydroponic and field trials (Tables 2 and 3, Tables S2 and S3). The coefficients of variation (CVs; CV = SD/Average × 100%) exhibited wide ranges among the 184 RILs. They ranged from 9.51% to 50.43% in the hydroponic culture trial and from 6.63% to 51.27% in the field trials. The CVs for 69.50% of all the traits were greater than 20% in the two trials (Table S2). The 31 and 38 investigated traits under hydroponic culture and field trials in each treatment (respectively) showed a continuous distribution. The h B 2 for all investigated traits ranged from 50.16% (TKUE-total K-use efficiency) to 80.00% (RMgCE-concentration of root Mg) in the hydroponic trial and from 42.41% (StKCE-concentration of straw K) to 89.67% (TGW-thousand grains weight) in the field trials (Table S2). The ANOVA results showed that the variance for either genotype or treatment effects on most investigated traits were significant at a p ≤ 0.05, excluding genotypic effects on SKCE (concentration of shoot K), TKCE (concentration of total K), SKUE (shoot K-use efficiency), and TKUE, as well as the treatment effects on RMgCE (concentration of root Mg) in the hydroponic culture trial. In the field trial, the genotypic effects on HI (harvest index), StKCE, AKCE (concentration of aboveground K), StMgCE (concentration of straw Mg), CaHI (Ca harvest index), MgHI (Mg harvest index), GKUE (grain K-use efficiency), GCaUE (grain Ca-use efficiency), and ACaUE (aboveground Ca-use efficiency), as well as the treatment effects on HI, GCaCE (concentration of grain Ca), StCaC (straw Ca content per plant), ACaC (aboveground Ca content per plant), MgHI, and GCaUE, were not significant at p ≤ 0.05 (Table S3). The LSD (least significant difference) test of the RIL population showed that the average values of the investigated traits were significantly different among the treatments of the hydroponic trial and the field trial in most cases (Table S2). These results indicated that the treatments and genetic background assisted in explaining the overall phenotypic variation. Most correlation coefficients (r) among traits were significant at the p ≤ 0.01 level (Table S4) in the hydroponic culture trial, except for the r between RCaC (root Ca content per plant) and RKC (root K content per plant), SKC (shoot K content per plant), TKC (Total K content per plant); TCaUE (total Ca-use efficiency) and RCaC, SCaC (shoot Ca content per plant), TCaC (total Ca content per plant). In the field trials, the correlation coefficients (r) between yield traits show that PH (plant height), SN (spike number per square meter), GWP (grain weight per plant), StWP (straw weight per plant), and AWP (total aboveground weight per plant) were significantly and positively correlated with each other. There were significant and negative correlations between TGW (thousand grains weight) and SN (spike number), HI (harvest index), and StWP (straw weight per plant) (Table S4).

QTL Clusters
A total of 18 QTL clusters (C1-C18) (a cluster was defined as the co-location of QTLs for more than two traits) were mapped to nine chromosomes, involving 96 out of the 306 QTLs (31.37%) (Figure 1, Table 5). All these QTL clusters could be classified into three types: detected only for seedling traits (Type I, including C1, C2, C4, C6-C14, and C16), only for adult traits (Type II, including C15 and C18), and simultaneously for seedling and adult traits (Type III, including C3, C5, and C17). Of these QTL clusters, C7, C12, C16, and C18 (Table 5, Figure 1) were the most important for seedling traits and/or adult traits. Table 5. Clusters comprising QTLs for more than two traits at seedling and mature stage.

K Effects on Biomass, K-, Ca-, and Mg-Related Traits of the RIL Population
Potassium is one of the essential nutrient elements for wheat. A deficiency in K can slow plant growth and decrease biomass production [35,36]. Compared with the CK treatment, the seedling traits of SDW and TDW significantly decreased in the LK treatment. In a similar manner, the SH, SN, GN, TGW, GWP, StWP, and AWP of maturity traits all decreased with reduced K concentration (Table S2).
It has been widely reported that there is a competitive relationship among K, Ca, and Mg as it relates to absorption [13,20,21,37]. The uptake of K may be affected by Mg, while Mg is affected by K [15]. Similarly, uptake of Ca could also be depressed by increasing the concentration of Mg [38]. In this study, the K content-related traits decreased, and the K use efficiency-related traits increased with the decreasing K supply compared to the normal treatment at the seedling and mature stage. In contrast, content-related traits of Ca and Mg increased, while the use efficiency-related traits decreased in the LK treatment at the seedling and maturity stage. These results also indicated that there was an antagonistic effect in absorption among K, Ca, and Mg.

K Effects on QTLs for K, Ca, and Mg Efficiency-Related Traits
For wheat, nutrient treatment can significantly affect the expression of nutrient-related QTLs. Some studies of QTLs or QTL clusters had been conducted under different nitrogen concentrations [27,[39][40][41][42]; under conditions of different K concentrations [28]; and under various concentrations of N, P, and K treatments [32]. In different nutrient environments, the number and the location of most QTLs detected for certain traits were quite different in these previous experiments. Similar to these studies, 302 (98.69%) QTLs and 10 QTL clusters (including 55 QTLs) of this study were detected only once in a single (moderate K or low K) K treatment (Table S5, Table 5). These sites might be important for adaptation to different K environments.
This indicated that K treatment greatly affected the expression of QTLs related to K, Ca, and Mg, and there might be quite different mechanisms for K, Ca, and Mg nutrition under different K treatments. The QTL clusters also provided some evidence for this indication.

Hot QTL Clusters for K, Ca, and Mg Efficiency-Related Traits
Many studies have reported that there were some genes which can affect the absorption of K, Ca, or Mg simultaneously. For example, the CBL-CIPK (CBL-CIPK: CBL-interacting protein kinase, Ca 2+ sensors) complex participates in the regulation of plant K + uptake under K + -deficiency stress [21]. The OsHKT2;4 (a K + -permeable transporter/channel)-mediated currents could also exhibit permeability to both Mg 2+ and Ca 2+ [15]. Finally, some genes or transporters/channels, such as the PaAlr1 gene in ascospore [17] and OsHKT2;4 in rice [15] showed sensitivity to Mg 2+ and Ca 2+ simultaneously [16]. The QTL clusters in this investigation might also provide us with some evidence about that. Eight QTL clusters included QTLs for K, Ca, and Mg simultaneously under the same K treatment, and these clusters included 61 QTLs. C7, C12, and C16 involved QTLs for K, Ca, and Mg simultaneously (C7, C16 in CK; C12 in LK); C5, C9, and C10 involved QTLs for K and Ca simultaneously (C10 in LK; C5, C9 in CK); and C15 and C18 (in LK) involved QTLs for K and Mg simultaneously. These QTL clusters highlight the important hot sites on the chromosome where some genes might be located that can affect the absorption of K, Ca, or Mg simultaneously for wheat and required further investigation.
Recently, Yuan et al. [29] reported the QTL mapping for P efficiency and morphological traits in wheat used the same RIL population derived from a cross of "Tainong 18 × Linmai 6." They found that four (C3, C4, C5, C6) out of 10 clusters were mapped to the 4B−1 chromosome and that the C3 and C5 clusters contained one and four RHF-QTLs, respectively. It is worth noting that many similar locations of QTL clusters were found between Yuan et al. [29] and this study ( Table 6). These results showed that 4B is a very important chromosome for mineral nutrition related to P, K, Ca, and Mg in wheat. In addition, the chromosome 4B also contained two (C3 and C5) out of three QTL clusters that contained QTLs, both for seedling and adult traits. The chromosome 4B and the QTL clusters are obvious and important targets that require further investigation. Table 6. QTL clusters detected in the same or adjacent marker regions in this paper and in Yuan et al. [29].