Comparative Genomic Analysis of Quantitative Trait Loci Associated With Micronutrient Contents, Grain Quality, and Agronomic Traits in Wheat (Triticum aestivum L.)

Comparative genomics and meta-quantitative trait loci (MQTLs) analysis are important tools for the identification of reliable and stable QTLs and functional genes controlling quantitative traits. We conducted a meta-analysis to identify the most stable QTLs for grain yield (GY), grain quality traits, and micronutrient contents in wheat. A total of 735 QTLs retrieved from 27 independent mapping populations reported in the last 13 years were used for the meta-analysis. The results showed that 449 QTLs were successfully projected onto the genetic consensus map which condensed to 100 MQTLs distributed on wheat chromosomes. This consolidation of MQTLs resulted in a three-fold reduction in the confidence interval (CI) compared with the CI for the initial QTLs. Projection of QTLs revealed that the majority of QTLs and MQTLs were in the non-telomeric regions of chromosomes. The majority of micronutrient MQTLs were located on the A and D genomes. The QTLs of thousand kernel weight (TKW) were frequently associated with QTLs for GY and grain protein content (GPC) with co-localization occurring at 55 and 63%, respectively. The co- localization of QTLs for GY and grain Fe was found to be 52% and for QTLs of grain Fe and Zn, it was found to be 66%. The genomic collinearity within Poaceae allowed us to identify 16 orthologous MQTLs (OrMQTLs) in wheat, rice, and maize. Annotation of promising candidate genes (CGs) located in the genomic intervals of the stable MQTLs indicated that several CGs (e.g., TraesCS2A02G141400, TraesCS3B02G040900, TraesCS4D02G323700, TraesCS3B02G077100, and TraesCS4D02G290900) had effects on micronutrients contents, yield, and yield-related traits. The mapping refinements leading to the identification of these CGs provide an opportunity to understand the genetic mechanisms driving quantitative variation for these traits and apply this information for crop improvement programs.


INTRODUCTION
Most agricultural systems focus on increasing crop productivity and grain yield (GY) and fewer efforts have been devoted to the grain yield-quality tradeoff. However, a shift from prioritizing yield to more emphasis on quality, such as nutrient content is gaining ground in breeding programs (Khush et al., 2012). Extending the existing concepts for a simultaneous selection of GY, quality traits, and micronutrient contents seems necessary to facilitate the development of varieties with an effective combination of yield potential and end-use quality (Michel et al., 2019). A rapid increase in micronutrient deficiency in food grains has resulted in micronutrient malnutrition among consumers. Fe and Zn deficiencies are serious and prevalent sources of malnutrition in developing countries with high consumption of cereals, such as wheat (Black et al., 2013;Shahzad et al., 2014;Kumar et al., 2019). Diets based on staple food crops with Zn and Fe deficiency have been widely recognized as a major global health problem that affects almost three billion people (Murray and Lopez, 2013). Breeding crops through biofortification is a practical approach to cope with Fe and Zn deficiencies by increasing the grain Fe content (GFeC) and grain Zn content (GZnC) within the edible parts of staple food crops, especially cereals (Stein, 2010;Liu et al., 2019;Shariatipour and Heidari, 2020).
Wheat biofortification through breeding methods is a promising strategy to ameliorate Fe and Zn deficiencies in developing countries . One of the most important challenges in breeding for micronutrients are negative genetic trade-offs between yield and micronutrient traits (Flatt and Heyland, 2011;Fabian and Flatt, 2012). At the genetic level, such trade-offs are thought to be caused by alleles with antagonistic pleiotropic effects or by linkage disequilibrium between loci (Fabian and Flatt, 2012). While GFeC and GZnC biofortification is an important objective in wheat breeding programs, other important traits, such as GY and grain protein content (GPC) typically cannot be compromised. The wheat quality is measured by its rheological traits, such as GPC (Goel et al., 2019) since wheat is a major source of protein accounting for 19% of human protein intake in the developing countries (Braun et al., 2010). The genetic control of quality traits and GY is complex (Zilic et al., 2011;Velu et al., 2018;Giancaspro et al., 2019;Liu et al., 2019). A high priority of breeders is to develop highquality genotypes that balance acceptable yield potential while maintaining quality characteristics, both of which are highly dependent on the co-variance between GY and the major quality (i.e., GZnC, GFeC, and GPC) criteria (Michel et al., 2019). However, a negative correlation between the GY and quality traits is challenging (Simmonds, 1995;Velu et al., 2016;Michel et al., 2019). In addition to this, our research indicates that protein concentration in grain decreases under elevated air CO 2 Abbreviations: AIC, akaike information criterion; AICc, corrected AIC; AWE, the average weight of evidence; BIC, Bayesian information criterion; CI, confidence interval; GCs, candidate genes; GO, gene ontology; GWAS, genome wide association studies; LOD, the logarithm of odds; MQTL, meta-quantitative trait loci; MTP, metal tolerance proteins; OrMQTL, orthologous MQTL; QTL, quantitative trait loci; SNP, single-nucleotide polymorphism.
concentrations of 550 µmol/mol (Fernando et al., 2012). By the end of the twenty-first century, it is predicted that the global temperature will rise from 1.1 to −3.1 • C and the atmospheric CO 2 concentration will reach above 550 ppm under intermediate scenarios (Pachauri et al., 2014) and that heat waves will occur with a higher frequency and longer duration (Pachauri et al., 2014). Given these future climate scenarios, it is critical to anticipate the effects of future growing environments and focus on breeding strategies that compensate for changes in grain quality. Likewise, it is important to have an inclusive breeding objective that tracks a portfolio of indirect traits responsible for grain quality and productivity, such as the assessment of dry matter accumulation, photosynthesis, coleoptile growth, carbon isotope discrimination, plant senescence, and rheological properties (Rebetzke et al., 2008;Liang et al., 2010;Vijayalakshmi et al., 2010;Goel et al., 2019).
Because of the large genome size and limited genome sequence information in wheat, the typical mapping intervals are quite large in most studies especially for the complex quality traits  and so further refinement is needed to narrow down the QTL intervals. Developing a statistically derived catalog of relevant loci is critical for developing marker-assisted selection (MAS) approaches in breeding programs. These markers can be applied to the quantitative trait loci (QTLs) that regulate the accumulation of high mineral nutrient concentration in grain along with QTLs for GY and grain quality traits. The QTL mapping method involves creating a QTL continuity map to identify genomic regions associated with quantitative traits (Mohan et al., 1997). Although QTL mapping is a powerful approach for detecting the genomic regions associated with complex traits, the genetic effects of QTL identified in different studies may not be present or are simply not tested in different genetic backgrounds and environments . In addition, the number of traits that can be measured in any single study is always resource limited (Acuña-Galindo et al., 2015). Overall, biparental populations are strongly influenced by different factors consisting of parents, the size and type of population, the choice of marker sets, and environmental conditions (Li et al., 2013;Izquierdo et al., 2018;Lei et al., 2018;Zhao et al., 2018).
In the last decade, an efficient approach called meta-QTL (MQTL) analysis has emerged in order to circumvent these restrictions. The MQTL method was initially developed by Goffinet and Gerber (2000) and was then improved by Veyrieras et al. (2007) is a method that gathers QTL data from independent experiments, years, location, and genetic backgrounds to detect stable QTLs (Goffinet and Gerber, 2000;Arcade et al., 2004;Hanocq et al., 2007;Sosnowski et al., 2012). The meta-QTL analysis integrates the information of QTLs from different population types and sizes identified in different environmental conditions to find stable MQTLs in a narrower genomic region with small CI (Goffinet and Gerber, 2000;Hanocq et al., 2007;Li et al., 2013).
The MQTL analysis allows for the dissection of genetic correlation among different traits (Truntzler et al., 2010;Danan et al., 2011;Xiang et al., 2012;Badji et al., 2018;Delfino et al., 2019). Hence, the MQTL analysis helps to enquire co-location of QTLs relying on dense marker maps that are responsible for different desirable traits including micronutrient contents, GY, and quality traits (Delfino et al., 2019). Currently, the MQTL analysis has become popular research in most studies (Goffinet and Gerber, 2000) related to micronutrients content (Raza et al., 2019), yield Avni et al., 2018), and quality traits (Quraishi et al., 2017) to overcome the inconsistent QTL information reported for GZnC, GFeC, GPC, and yield traits. Further, MQTL analysis helps to identify the candidate genes (CGs) and refine genomic regions for yield and quality traits (Raza et al., 2019). However, little is known about the relation of micronutrients and quality-related MQTLs with agronomic traits in wheat.
The transferability of QTLs between cereals based on the analysis of syntenic regions and genomic collinearity helps to identify stable and important QTLs for use in breeding programs. The aims of this study were to perform a QTL meta-analysis to (1) identify QTLs that are consistently associated with grain quality, yield traits, and micronutrient content (2) explore the co-localized QTLs controlling GY, GPC, GZnC, and GFeC in the wheat genome, and (3) assess the transferability of QTLs between wheat, rice, and maize based on the comparative genomics and the orthologous MQTL (OrMQTS) mining. The outcome of this study will aid plant breeders in refining micronutrients, GY, and quality traits for crop improvement through marker-assisted breeding. Refining our understanding of the genetic architecture of micronutrients, GY, and quality traits often leads to the interrelationship between the regions of the genome that may be more challenging to breed independently. The MQTL is an analytic procedure that helps refine these relationships between CGs by providing for precision and statistical power.

QTL Database Development
A database consisted of 735 quantitative trait loci (QTLs) derived from 27 independent mapping populations (assessed between 2006 and 2019) assigned to 70 traits ( Table 1) was used for the meta-QTL (MQTL) analysis. The independent populations consisted of 20 recombinant inbred lines (RILs) and 7 double haploids (DH) populations, with population sizes ranging from 92 to 485 lines (Supplementary Table 1). The reported position, the proportion of the phenotypic variance (R 2 ), and the logarithm of odds (LOD score) of the initial QTLs were used for the analysis of meta-QTLs. For QTLs with missing LOD or R 2 , the values were estimated by the following equation (Nagelkerke, 1991): where n represents the size of the population.

Constructing Consensus Genetic Map and QTL Projection
The data files of the 27 maps were integrated with the Somers (Somers et al., 2004) reference map for the construction of a consensus genetic map. Attempts to use other mapping studies consisting of single-nucleotide polymorphism (SNP) were unsuccessful due to the lack of SNP density in the regions for the meta-QTL analysis. The constructed map file for each population consisted of information on cross-type, population size, map function, map units, and the position of different markers in different linkage groups. The individual QTLs derived from independent populations were projected onto the consensus genetic map consisted of 3,394 markers with a total length of 3,412.5 cM.

Meta-QTL Analysis
Meta-QTL analysis was performed in BioMercator v4.2 (Arcade et al., 2004;Sosnowski et al., 2012). For n QTLs, the BioMercator tests the most likely assumption based on Akaike information criterion (AIC), corrected AIC (AICc), AIC 3 candidate models (AIC3), Bayesian information criterion (BIC), and an average weight of evidence (AWE) criteria in which the prevalent value among them was considered as the best fit. The consensus QTL from the optimal model was reported as MQTL. Consequently, the MQTL position and distribution on each linkage group were presented as a heatmap using pheatmap R package (Kolde, 2013). Moreover, the initial QTLs with 95% confidence interval (CI), QTL density in the identified MQTL, and the distribution of MQTLs were drawn on the linkage groups using shinyCircos web tool based on the R program (Yu et al., 2017). The variation of QTL density for different traits toward centromeric and telomeric genomic regions was estimated following the approach by Martinez et al. (2016). The QTL density was determined by counting the number of QTLs for each trait on 50 cM intervals across the wheat genome, starting from the centromere region of a linkage group at position 0. The centromere position was retrieved from the study by Wan et al. (2017).

Functional Candidate Genes in MQTLs Intervals
The MQTLs containing more than five trait-QTLs from different experiments were considered as the most stable consensus regions and were analyzed for the detection of functional candidate genes (CGs). To identify the functional CGs, the sequences of the flanking markers for each MQTL were retrieved from "Grain Genes" database (https://wheat.pw.usda. gov/browse?class=probe;query=BARC%2A;begin=351) for the simple sequence repeat (SSR) flanking markers and "Diversity Array Technology" (https://www.diversityarrays.com/) (DArT) flanking markers. For flanking markers lacking a definite position on the wheat genome, the closest markers on the genetic consensus map were selected to determine the MQTL position. Additionally, for those flanking markers lacking sequence information in databases, the forward and reverse sequences were retrieved from the "Grain Genes" database and were used for the Basic Local Alignment Search Tool (BLAST) analysis against the newest wheat reference genome (IWGSC RefSeq v2.0) for detecting the genomic position of each MQTL. The annotation and gene ontology (GO) of genes lying at the MQTL interval were retrieved from EnsemblPlants (http://plants.ensembl.org/ index.html) using the new wheat genome (IWGSC v2.0). Finally, the orthologous of genes located at each MQTL interval were

Identification of Traits Within MQTLs
To analyze traits within the MQTL regions, the MQTL results were converted into binary scores (0 or 1) on the basis of the absence/presence of an individual trait-QTL within an MQTL region. We tabulated the number of times a trait was present within an MQTL, the number of QTLs for a trait present within an MQTL (implying confirmation of the QTL), and the number of times the traits were able to be co-localized within an MQTL. A chi-squared test with one degree of freedom was performed to determine traits showing significant co-localization with grain protein content (GPC), grain zinc content (GZnC), grain Fe content (GFeC), and grain yield (GY) beyond what would be expected for a random distribution of QTL within MQTL throughout the genome. The expected number of MQTL associated with a trait and each GPC, GZnC, GFeC, and GY were separately calculated by multiplying the number of observed MQTL for a trait by the proportion of MQTL containing GPC, GZnC, GFeC, and GY QTL(s). Traits within the MQTL regions were also analyzed using IBM SPSS Statistics v.24. The simple regression analysis was performed using Minitab v. 18 to determine the effect of MQTLs on the association of GY, GPC, GFeC, and GZnC.

MQTLs and GWAS Comparison
The detected MQTLs were compared with the significant loci associated with different quantitative traits identified in wheat genome-wide association studies (GWAS). The mapped coordinates of the identified significant loci through GWAS were compared to those found with the MQTL analysis.

Orthologous MQTL
Due to the high synteny among genes in Poacaea, the most stable and promising wheat MQTLs were evaluated for the detection of the orthologous MQTLs (OrMQTL) in rice (Lei et al., 2018;Raza et al., 2019;Khahani et al., 2020), and maize (Semagn et al., 2013;Wang et al., 2013Wang et al., , 2016. A set of orthologous genes at the MQTLs regions was considered as a criterion of a syntenic region using EnsemblPlants (http://plants.ensembl.org/index.htmldatabase).

Genomic Quantitative Trait Loci Distributions
The individual traits of quantitative trait loci (QTLs) used in this study are listed in Supplementary Table 2. The QTLs for thousand kernel weight (TKW) (18.03%), grain yield (GY), (13.11%), and a number of grains per spike (NS) (13.11%) were the most frequently reported agronomic QTLs identified in the tested mapping populations. The grain Fe content (GFeC) (33.33 %) and grain Zn content (GZnC) (28.21 %) of QTLs for micronutrient traits and grain protein content (GPC) (64.00 %) and sudden death syndrome (SDS) (10.00 %) of QTLs for quality traits were frequent. Our results suggest a non-random distribution of QTLs within the wheat genome. Distribution of QTLs on the basis of physical size [χ 2 (2) = 60.17, P = 8.58E−14] showed that 254, 326, and 155 QTLs were located on the A, B, and D genomes, respectively. The QTL distribution was significantly different among the seven chromosome groups [χ 2 (6) = 47.12, P = 1.77E−8], ranging from as few as 76 QTLs on group 6 to as many as 165 QTLs on Group 2. Chromosome 3B with 75 QTLs had the highest number of QTLs, followed by chromosome 2B (62 QTLs) and 2A (61 QTLs), while chromosome 3A with 10 QTLs had the lowest QTL. The distribution of QTL over the genetic linkage map with respect to centromeric and telomeric regions was distinctly non-random (Figure 1). The non-telomeric region of each chromosome (−50 up to + 50 cM intervals) had the highest number of QTLs (Figure 1). We did not detect QTLs at 100 cM for the tested traits. The distribution of meta QTLs (MQTLs) indicated that a cluster of MQTLs was mapped to the non-telomeric regions. Besides, MQTL_5B_4 and MQTL_6B_1 were located near the centromeric region of chromosomes 5B and 6B, respectively (Figure 2). There was a significant correlation between the number of initial QTLs and MQTLs (r = 0.46, P < 0.03). The number of MQTLs per chromosome varied from two (chromosomes 3A and 5D) to eight (chromosomes 1B, 2A, and 6B).

Meta-QTL Analysis
Of the 735 initial QTLs, 449 were successfully projected onto the genetics consensus map and used in the meta-QTL analysis (Figure 3). A total of 100 MQTLs were detected and the number of individual QTL per MQTL ranged from 1 to 43 (Figures 4, 5; Supplementary Figure 1). The number of traits present per MQTL region ranged from 1 to 18. Among the identified MQTLs, MQTL_3B_1 that contained 43 QTLs had the highest number of initial QTLs followed by MQTL_7A_3 with 29 initial QTLs (Supplementary Table 3). These two MQTLs can be considered as the most stable QTLs under different experimental conditions. The detailed information of MQTLs consisted of the chromosome number, position, confidence interval (CI), flanking  Table 2. The higher marker density of the consensus map compared with the lower marker density in the independent linkage maps helped to reduce the CI of QTLs up to three-fold with an average of 4.63 cM in MQTLs compared with the mean CI of 13.73 cM for the original QTLs. Among the detected MQTLs, the CI of 11 MQTLs was reduced up to <1 cM ( Table 2).

Functional Identification of Candidate Genes
The genomic positions of the stable MQTLs and the number of functional candidate genes (CGs) in their intervals are reported in Table 3. The range for the number of the CGs annotated in the tested meta-QTLs was between 20 and 802. The MQTL_4D_1, MQTL_4B_3, MQTL_3B_1, and MQTL_5A_4 harbored the greatest number of CGs. Among the detected CGs on MQTL regions, several wellknown genes including psbL (21.896318-21.896434 Mb), psbT (21.905522-21.905638 Mb), rpl33 (21.899696-21.899896 Mb), and rps4 (24.160684-24.161289 Mb) genes were located in the MQTL_3D_4 region. In addition, the miR166 gene was detected in the MQTL_4D_1 (450.196302-450.196403 Mb) and MQTL_7A_4 (561.152174-561.152339 Mb) regions. Furthermore, several CGs with unknown annotation in wheat and were orthologous to genes in rice were identified (Supplementary Tables 4, 5). Overall, some CGs, such as TraesCS2A02G141400, TraesCS3B02G040900, TraesCS4D02G323700, TraesCS3B02G077100, and TraesCS4D02G290900 were uncovered with a possible role in micronutrient contents, yield, and yield-related traits.

Traits Analysis Within MQTLs
Our data show the unequal contribution of individual QTL across the detected MQTLs (Figure 6 and Supplementary Table 3). Individual QTL for TKW were present in 41 of the 100 MQTL regions, the most for any agronomic trait, followed by GY (38 MQTL). Among quality traits, QTLs for GPC was the most distributed QTLs which were located among 19 MQTLs. The GFeC and GZnC QTLs were presented in 19 and 12 MQTLs, respectively (Supplementary Table 3).
Analysis for the co-localization of QTLs revealed that QTLs for GY and TKW were frequently co-localized with QTLs of the target traits (GPC, GFeC, and GZnC). The QTLs for TKW showed 52% co-localizations with QTLs for GPC (Supplementary Table 6). The GY QTLs showed 52% colocalization with the QTLs for GFeC. Results also indicated 66% co-localization between grain Fe and grain Zn QTLs. Colocalization frequency for GY (R 2 = 80.81%) was strongly associated with the total number of MQTL for a trait. The association of co-localization frequency with the overall number of MQTL for a trait was relatively strong for GPC (R 2 = 65.58%), GFeC (R 2 = 58.74%), and GZnC (R 2 = 56.00%). For most traits, association with target traits (GPC, GZnC, GFeC, and GY) did not differ from the expected on the basis of chisquared analysis. However, the association of traits with GY, GPC, GFeC, and GZnC as target traits was more than expected (Supplementary Table 6).

Comparison of the Identified MQTLs and QTL Mapping in the Wheat GWAS
The comparison of the MQTL locations with genome-wide association studies (GWAS) QTL regions showed that 21 significant signals (SNPs-linked QTLs) of the available wheat GWAS map were co-located with MQTLs of seven of the 35 traits tested in our study ( Table 4). The results indicated the co-localization of significant single-nucleotide polymorphisms (SNPs) for GY (6 SNPs), a number of grains per spike (NG) (1 SNPs), plant height (PH) (4 SNPs), spike length (SL) (2 SNPs), spike number (SN) (1 SNPs), spike number per spike (SNS) (4 SNPs), and thousand kernel weight (TKW) (3 SNPs) traits in the wheat GWAS with the identified MQTLs in our study. For instance, the MQTL_4A_3, MQTL_4A_4, and MQTL_4B_3 identified for TKW in our study were positioned in the genomic regions of the major signal reported for TKW in the wheat GWSA map. Overall, the co-located MQTLs and significant GWAS signals were distributed on chromosomes 1B, 2B, 3B, 4A, 4B, 4D, 5A, 7A, and 7B (Table 4).

Orthologous MQTL Mining of Wheat in Rice and Maize
The comparative analysis for QTLs in wheat, rice, and maize resulted in the identification of orthologous MQTL (OrMQTL).

DISCUSSION
Quantitative Trait Loci and Meta-Quantitative Trait Loci Distribution Over the Wheat Genome Analysis of the genetic control of quantitative traits is a challenge in plant breeding that is due to the complexity and the   (Izquierdo et al., 2018). Meta-quantitative trait loci (MQTL) analysis made possible the consolidation of individual QTL into MQTL regions. The MQTL analysis and the comparative genomic approaches help to identify consensus QTLs and conserved genes across genomes and species. In the current MQTL analysis, we used information from over 60% of the individual QTLs to identify MQTLs that showed the statistical power of MQTL analysis for combining wheat QTLs that were in the range of 54-84% that have been reported in the previous studies of MQTL in wheat (Löffler et al., 2009;Acuña-Galindo et al., 2015;Soriano and Alvaro, 2019). In this study, we narrowed down the confidence interval (CI) of the detected MQTL regions compared to the initial QTLs used in our MQTL analysis. The efficacy of MQTL analysis in refining the CI for previously known QTLs as well as for validating their effects across different genetic backgrounds and environments is well-demonstrated (Goffinet and Gerber, 2000;Wu and Hu, 2012).
The results showed that QTLs and MQTLs had higher density in the non-telomeric and near-centromeric regions. The QTL result from the genetic segregation of sequence polymorphisms at functional elements, such as regulatory sequences upstream of genes and/or coding sequences (Flint and Mackay, 2009;Salvi and Tuberosa, 2015). Therefore, it is expected that QTL density on a genetic map is driven by gene density, polymorphism rate at functional sites in genetic regions, and the frequency of recombination. This finding was in line with the result of Martinez et al. (2016), which illustrated higher QTLs densely mapped to the near centromeres on the genetic maps.
In the quality traits category, QTLs for grain protein content (GPC), grain Fe content (GFeC), and grain Zn content (GZnC) traits were the most observed QTLs in the identified    Table 2). In agronomic traits, QTLs for thousand kernel weight (TKW), grain yield (GY), and number of grains per spike (NG) were the most frequent QTLs identified in the MQTLs regions. The outcome of a MQTL analysis in tetraploid wheat revealed that more than 10 loci were associated with TKW (Avni et al., 2018). Besides, in another MQTL analysis in wheat, individual QTLs for TKW, GY, and kernel number had the highest number of individual QTLs in MQTL regions. This result shows the importance of these QTLs for tested traits. The probable assumptions for a higher frequency of QTLs for agronomic traits are easy to measure and frequent data for these traits in different genetic mapping studies. On the other hand, the TKW, GY, NG, GPC, GFeC, and GZnC traits are multi-genic, highly heritable, and relatively insensitive to the environment (especially, TKW) which suggests a high likelihood of different populations carrying different suites of relevant alleles (Cooper et al., 1995;Bezant et al., 1997;Avni et al., 2018;Velu et al., 2018;Zhang et al., 2020).

MQTLs and Functional Candidate Genes
Gene annotation analysis for the MQTL regions helps clarify our understanding of their genetic architecture and refining the targets of breeding for these traits. The results of functional genomics for the identified MQTLs showed that several well-known genes consisted of psbL for electron transfer in photosystem II (PSII) (Ozawa et al., 1997), psbT encoding a PSII subunit for maintaining optimal PSII activity under adverse growth conditions (Monod et al., 1994), rpl33f or structural constituent of ribosome and translation which confers tolerance to cold stress (Rogalski et al., 2008;Moin et al., 2017), and the rps4 gene for the regulation of translational fidelity in wheat were located in the MQTl_3D_4 region. In addition, the miR166 gene was detected in the MQTL_4D_1 and MQTL_7A_4 regions. Comparative genomic approaches start with making some form of alignment of genome sequences and looking for orthologous sequences in the aligned genomes and checking to what extent those sequences are conserved. Based on these, genome and molecular evolution are inferred and this may, in turn, be put in the context of phenotypic evolution or population genetics. Analysis of the conservation and diversification of the miR166 family have shown that miR166 members play a wide and important regulatory role in seed development. More recently, a short-tandem target mimic (STTM) method was used to verify if miR166 regulates important agronomic traits in rice . In a study, transgenic STTM165/166 plants showed significantly reduced seed number and sterile siliques in Arabidopsis, suggesting that miR166 plays a vital role in seed development and it might be useful evidence to improve inferior grain size in wheat . Among the detected and annotated genes in the MQTLs regions in this study, the OsMED9 (TraesCS3B02G077100), which is an orthologous gene of rice, was identified in the MQTL_3B_1 region that was associated with GY and its components. A diverse array of MED genes has been identified for the regulation of GY and yield components in crop plants (Malik et al., 2016). The results indicated that the MQTL_4D_1 of our study was located in the regions of the rice orthologous BG1 (TraesCS4D02G290900), OsIDD1 (TraesCS4D02G262500), and OsPAO (TraesCS4D02G309000) genes in wheat. These genes are associated with metal ion binding, flowering time, days to heading, senescence, gravitropism, yield, and yield-related traits (Wu et al., 2008;Liu et al., 2015;Chen et al., 2016;Deng et al., 2017;Mishra et al., 2017). Overexpression of BG1 leads to larger grain size in rice  and manipulation of BG1 increases plant biomass, grain size, and GY in rice and Arabidopsis Mishra et al., 2017). OsIDD1 could rescue the never-flowering phenotype of rid1 by a transition from vegetative to reproductive growth in rice , and the PAOs protein-encoding genes (Chen et al., 2016) regulate cellular polyamine levels which are critical for embryogenesis (Bertoldi et al., 2004;De-la-Pena et al., 2008), germination (Bethke et al., 2004;Liszkay et al., 2004), root growth (Cona et al., 2005), flowering and senescence (Kakkar and Sawhney, 2002), and mineral deficiency (Moschou et al., 2008(Moschou et al., , 2009). The orthologous Ghd7 (TraesCS5A02G541200) that was among 427 genes located in the MQTL_5A_4 region in our study involves photoperiodism, flowering, days to heading, plant height (PH), and yield traits. Enhanced expression of Ghd7 under long-day conditions delays heading and increases PH and panicle size in rice. The Ghd7 gene plays crucial role in increasing the productivity and adaptability of rice globally (Xue et al., 2008). The uncovered rice orthologous D27 (TraesCS7B02G319100) and BRD2 (TraesCS7B02G484200) genes in wheat in chromosome 7B region are known to control tillering, heading date, PH, and yield traits in rice (Hong et al., 2005;Lin et al., 2009;Liu et al., 2016). The identified rice orthologous D18 (TraesCS2B02G570800), OsRPK1 (TraesCS5D02G034200), DRUS1, and DRUS2 (TraesCS4A02G133800) genes within the MQTL_2B_2, MQTL_4A_2, and MQTL_5D_2 regions of our MQTLs were associated with PH in wheat (Itoh et al., 2002;Zou et al., 2014;Pu et al., 2017). Manipulating these genes lead to varieties with dwarf and semi-dwarf phenotype and such phenotypes possess short and strong stalks, exhibit less lodging, and a greater proportion of assimilation partitioned into the grain, resulting in further yield increases (Hedden, 2003).
The rice orthologous OsMTP12 (TraesCS2A02G141400), OsMTP9 (TraesCS3B02G040900), and OsMTP1 (TraesCS4D02G323700) belonging to the metal tolerance protein (MTP) gene family were identified in the MQTL_2A_1, MQTL_3B_1, and MQTL_4D_1 interval of the wheat genome in the present study. The MTP gene family plays a critical role in metal transport, mainly in Zn, Mn, Fe, Cd, Co, and Ni, metal homeostasis, and tolerance (Gustin et al., 2011;Zhang and Liu, 2017;Ram et al., 2019). The increased expression of MTP genes during seed development and their potential role in metal homeostasis during the seed filling stage had been documented . The MTP1 and MTP12 genes share a characteristic histidine-rich loop toward the c-terminal, which is known to have a role in Zn-binding .
The MQTL_1A_4, MQTL_3B_5, MQTL_4B_1, MQTL_4B_2, MQTL_4B_3, MQTL_4B_4, MQTL_7A_1, and MQTL_7A_4 interval regions harbored rice orthologous VAL1 (TraesCS4B02G314600), PROG1 (TraesCS4B02G354000), D14 (TraesCS4B02G258200), LPL2 (TraesCS4B02G308000), OsINV3 (TraesCS7A02G009100), and OsRLCK218 (TraesCS7A02G385300) genes with a role in diverse development and growth traits including leaf development, PH, shoot branching, panicle and tiller number, grain number, grain and spikelet size, and GY (Vij et al., 2008;Zhou et al., 2016;Morey et al., 2018;Wu et al., 2018;Yao et al., 2018;Zhang et al., 2018; Deng et al., 2020). These candidate genes (CGs) at the detected MQTL regions can potentially have the same function as their orthologous varieties in rice and therefore regulate various developmental and growth-related traits in wheat. Identification of these confines stable chromosomal regions and CGs that influence economically important traits in wheat can help to expedite wheat improvement in future breeding programs. The MQTLs detected in this study will help identify CGs in these regions responsible for desirable traits and generate allelespecific markers through allele mining (Leung et al., 2015;Ogawa et al., 2018) for marker-assisted selection (MAS) application in pre-breeding population. Allele mining is a promising approach to dissect naturally occurring allelic variation at QTLs/CGs controlling desirable traits (e.g., yield, quality, and micronutrient content) which has potential applications in crop improvement programs (Kumar et al., 2010;Gokidi et al., 2017;Kumari et al., 2018). The data raised from this MQTL study help to refine genomic targets for validation and subsequent development of haplotypic markers in breeding programs. Information of the identified MQTLs can also be used for genetic transformation or allele screening in germplasm collections (ecoTilling) for finding new alleles capable of improving yield, quality, and micronutrient traits (Izquierdo et al., 2018;Belzile et al., 2020). Additionally, a promising application of the variation within these MQTL regions might be their introduction as fixed effects in genomic selection (GS) models to increase the accuracy of the prediction models in their use in wheat breeding programs (Spindel et al., 2016;Izquierdo et al., 2018).

MQTL and Traits Analysis
The QTLs projection on a consensus map allows for the inspection of co-location across traits and categories, which is especially relevant for complex traits (Delfino et al., 2019). The association between trait classes by analyzing the co-localization frequency of individual trait-QTLs demonstrated that TKW with 55% and 63% co-localization frequencies was frequently associated with GY and GPC. In addition, GY and GFeC had the highest co-localization frequency with GFeC (52%) and GZnC (66%), respectively. The results of MQTL analysis of our study confirmed correlations identified for QTLs of TKW and GY (An-Ming et al., 2011;Azadi et al., 2014;Mahdi-Nezhad et al., 2019), TKW and GPC (Wang L. I. et al., 2012;Goel et al., 2019), and GFeC and GZn (Roshanzamir et al., 2013;Pu et al., 2014;Tiwari et al., 2016;Liu et al., 2019) in other studies. Co-localization of QTLs for correlated traits has been identified by Wang et al. (2018). Co-localization of QTLs could be due to the pleiotropy or the presence of different linked genes in the same regions that can partly explain the correlation that exists between traits (Bhatta et al., 2018). The tightly linked genomic regions or pleiotropy can partly explain the correlation that exists between traits (Crespo-Herrera et al., 2016). The result of co-localization frequency of target traits (GY, GPC, GFeC, and GZnC) with the detected MQTLs in our study showed interacting MQTLs which affects the association of traits at the genomic level. Acuña-Galindo et al. (2015) demonstrated that TKW was most frequently associated with GY in MQTL regions, with 57% co-localization. The colocalization of TKW and GPC QTLs has been observed in the genetic analysis of wheat (Goel et al., 2019). The positive and highly significant associations of GPC and TKW traits have been reported in different wheat populations in other studies (Peleg et al., 2009;Badakhshan et al., 2013;Krishnappa et al., 2017). According to Liu et al. (2019), mineral nutrient concentrations and yield components have shown common QTLs which is in line with co-localized QTLs for grain Zn, Fe QTLs in the tetraploid and hexaploid wheat in other studies (Peleg et al., 2009;Xu et al., 2012;Crespo-Herrera et al., 2016;Krishnappa et al., 2017;Velu et al., 2017). In a recent QTL analysis study on rice, grain Fe and grain yield QTLs were found to be co-localized (Dixit et al., 2019). Crespo-Herrera et al. (2016) suggested the possibility of simultaneous breeding for GFeC and GZnC traits due to their co-localized QTLs. The current co-localization analysis within MQTLs regions indicates the possibility of simultaneous breeding of micronutrient content, grain quality, and GY by pyramiding the QTL regions through marker-assisted selection (MAS). Using -MAS, the specific regions can be transferred to the elite wheat genotypes to simultaneously increase the contents of various traits (Saini et al., 2020). Therefore, an attempt to pyramid QTLs responsible for GY, TKW, GPC, GZnC, and GFeC may accelerate progress in wheat variety development. The QTL pyramiding strategy has been used for simultaneous improvement of traits through MAS in wheat (Wang P. et al., 2012;Feng et al., 2018;Gautam et al., 2020;Muthu et al., 2020). A clear understanding of the co-location of QTLs and their effect on target traits, such as grain Fe and Zn concentrations and yield is very important for using the major effect of QTLs in markerassisted breeding (Swamy et al., 2018). The results of MQTL analysis in our study showed that the location of 18 MQTLs was in agreement with the position of the QTLs in the meta-QTL studies by , Acuña-Galindo et al. (2015), and Kumar et al. (2020). However, our work adds to this body of genomic mapping research by identifying new MQTL specific for GY, grain quality, and micronutrient content.  (Jamil et al., 2019;Li et al., 2019;Mathew et al., 2019;Hu et al., 2020;Li X. et al., 2020). Some of the QTL hotspots suggested common genetic markers for wheat traits for further use in marker-assisted breeding that increase genetic gain in breeding programs. The identified OrMQTLs in this study can facilitate the detection of the underlying regulatory genes with evolutionary history and conservative function. The current MQTL analysis defines a genome-wide landscape on the most stable genetic markers and CGs related to micronutrient content, yield, and yield-related traits as the most economically important traits in wheat. The straight-up comparative genomics in our study indicated high synteny between wheat, rice, and maize QTLs especially for GY suggesting possible corroborating evidence of acting the same way in different species that was in line with the results of other MQTL studies (Ahn et al., 1993;Moore et al., 1995;Gale and Devos, 1998;Feuillet and Keller, 1999;Minx et al., 2005). Despite the high interest in the identification of genes involved in GY and yield-related traits in maize and wheat as two economically important crops, the responsible genes have largely remained unknown due to their complex genomes. Given a close evolutionary relation among grass genomes (Gaut, 2002), a synteny analysis of wheat, maize, and rice through the identification of OrMQTLs enabled us to broaden our genetic information and leads to uncover the possible function of unknown CGs among these species. Here we selected the most promising wheat MQTLs to explore their conserved syntenic regions reported in similar MQTLs studies on the same traits in rice and maize to identify OrMQTLs (Table 5).
More intriguingly, the syntenic regions of the wheat MQTL_7B_3 on both chromosome 6 of rice (OrMQTL_10) and chromosome 5 of maize, harbored the rice (OsMAPK4, Os06g0699400) and maize (ZmMPK5, Zm00001d014658) and orthologous gene in wheat (TraesCS7B02G322900). The mitogen-activated protein kinase (MAPK) cascades play important roles in regulating plant growth (PH), development, and stress responses (Zhu et al., 2020). The maize ZmMPK5 is induced by various stimuli, involved in defense signaling pathways in various abiotic/biotic stress, confers tolerance to stresses Zhang et al., 2014), and subsequently leads to enhancement of plant growth and yield. Due to the key role of OsMAPK4 in plant growth, grain development Chen et al., 2021), and subsequently in GY, the results of the current syntenic analysis suggest the same function for TraesCS7B02G322900 and Zm00001d014658 genes located in OrMQTL_10 interval (Figure 7; Table 5; Supplementary Table 7).

CONCLUSION
The results of this study introduced several novel metaquantitative trait loci (MQTLs) for improving wheat in multipurpose breeding programs by identifying key genomic regions associated with agronomic performance, grain quality traits, and micronutrients content. The results of our MQTL analysis have significantly increased the power and precision of our ability to map wheat traits and will provide greater resolution for future fine mapping and marker development. Importantly, our data identify co-localization between grain yield (GY) QTLs with grain Zn content (GZnC), grain Fe content (GFeC), and grain protein content (GPC) QTLs that suggest an opportunity for simultaneous breeding for these traits. This study also provides an example for the value of comparative analysis between evolutionarily close cereal species for the identification of genomic regions and candidate genes (CGS) controlling quantitative traits. Our finding shows the utility of MQTL analysis for refining the location of genomic regions associated with a variety of traits and helps understand how their relative map positions can be exploited for crop improvement. Overall, these findings can lead to both increased selection efficiency and accuracy for breeding by providing the basis for marker development in a marker-assisted selection (MAS) program and for identifying a novel source of variation through allele mining efforts in genetic resource collections. Lastly, these refined MQTLs provide the basis for further focus on the genetic mechanisms controlling micronutrients, GY, and quality traits.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
NS conducted bioinformatics analysis and wrote the draft of the article. BH conceived and designed the project, the bioinformatics analysis, and complemented the writing of the article. AT helped in bioinformatics analysis. CR helped with reviewing and providing critical advice on the article. All authors have read and approved the manuscript.