Wheat Recombinant Inbred Line Grain Metabolomics Analysis and Metabolite-Agronomic Traits Association

Wheat Recombinant Inbred Line Grain Metabolomics Analysis and Metabolite-Agronomic Traits Association.  This study uses metabolomics and association analysis methods to better understand the genetic basis of wheat metabolism and provide a scientific basis for wheat breeding.

Guide

Wheat is one of the most important crops in the world. It provides 20% of the calories and 25% of the protein consumed by humans. Metabonomics methods have been widely used in many crops, but they are still limited to wheat. Plant metabolome is generally considered to be the bridge between genome and phenotype, because in the broadest sense, metabolome defines phenotype, and its combination with quantitative genetic analysis greatly helps researchers infer plant metabolism and phenotypic variation Genetic link between.

Combining metabolomics with genomics and transcriptomics has been shown to have a powerful role in analyzing metabolic diversity and its potential genetic variation, as well as identifying many new genes and metabolic pathways. Therefore, metabolites can be used as a predictive complex Important biomarkers for agronomic traits, allowing rapid acceleration of the breeding process while reducing its cost.

Plants are rich in specific metabolites. These metabolites play an important role in the plant life cycle and mediate their interactions in the complex environment in which they live. However, the genetic structure of the wheat metabolome has not been well studied. Therefore, this study uses a high-density genetic map to conduct a comprehensive metabonomic study on wheat grain metabolism through widely targeted LC-MS/MS, and further combines agronomic traits to analyze the genetic relationship between metabolites and agronomic traits.

The research greatly improved people’s understanding of wheat metabolomics and its relationship with agronomic traits, and provided a powerful tool for crop improvement.

Paper ID

Original name: Metabolomics analysis and metabolite-agronomic trait associations using kernels of wheat (Triticum aestivum) recombinant inbred lines Translation: Metabolomics analysis and metabolite-agronomic trait association journals of wheat recombinant inbred lines: The Plant Journal IF: 6.141 Publication time : 2020.07 Corresponding author: Chen Wei, Cui method Corresponding author units: Huazhong Agricultural University, Shandong University of East

Experimental results:

1. Metabolic profile analysis and generalized inheritance

Using 145 recombinant inbred lines of Kenong 9204 (KN9204) and Jing411 (J411) as materials, we collected mature grains. In order to identify the main genes that affect agronomic traits, parental lines have great differences in grain traits and ear traits. We used the high-throughput LC-MS/MS method (this method has broad targeting) to repeatedly detect and quantify 1,260 different metabolite characteristics of inbred mature grain extracts through 3 organisms.

Among the features, 116 were determined by directly comparing their chromatogram and fragment behavior with the structure of the real standard, and 351 were presumably annotated by using the previously described strategy. Most of the annotated compounds are flavonoids, phenolamines, polyphenols, lipids, vitamins, phytohormones and their derivatives, amino acids and their derivatives, nucleic acids and their derivatives, organic acids and sugars, which have achieved important Coverage of metabolic pathways (( Figure 1a ).

The level of metabolite accumulation varies greatly among different strains, and we can effectively analyze their genetic structure. In the RIL population, the average genetic coefficient of variation (CV) of these metabolites is 47.4% (Figure 1b). However, the difference between phenolamines and polyphenols is large, with the maximum average CV of 59.8%, which is in the range The CV from spermine is 13.6% to that of N’, N”-Di-p-coumarin spermidine is 194.5%. The generalized heritability (H2) distribution of metabolic traits shows that more than 56% of metabolites have a heritability of 0.6 or more (Figure 1b). In general, among the attached metabolites, the H2 value of secondary metabolites (average 0.63) is higher than that of primary metabolites (average 0.58), and the heritability of flavonoids is the highest (H2> 0.70). These data indicate that the diversity of metabolites is mainly affected by genetic factors.

Metabolite analysis can clarify the connection between metabolic pathways. Therefore, we use Spearman to analyze the correlation of these metabolites and construct a heat map of all detected metabolites. Figure 1c shows more positive correlations (red) rather than negative correlations (blue), and some closely related metabolite clusters, such as the colored box in the upper left corner is mainly composed of amino acids and their derivatives, nucleic acids and their derivatives The purple and blue boxes at the bottom represent the high positive correlation between lipids and plant hormones and their derivatives (Figure 1c). These closely related metabolites are likely to be of the same type.

Molecules or molecules belonging to the same biochemical pathway. Most amino acids and nucleic acids exist in a tight metabolite cluster, while flavonoids are relatively dispersed, although their relationship is closer than other substances. Lipids, polyphenols and phenolamines are found in several large clusters, indicating that these metabolites are involved in multiple metabolic pathways and may play different physiological roles.

Figure 1 Metabolic profile analysis of wheat RIL population (a) The number of metabolites detected and their classification. (b) Distribution of coefficient of variation (CV) and generalized heritability (H2) values ​​of metabolic traits in RIL population. H2 is estimated by one-one ANOVA, taking into account the differences between the three biological replicates, as the phenotypic variance derived from environmental factors. (c) The paired Pearson correlations are displayed in the heat map, and the metabolites are sorted according to the correlation-based hierarchical cluster analysis. The degree of correlation is expressed in red (positive correlation) and blue (negative correlation).

2. MQTL mapping of wheat population grains

Based on the Affymetrix Wheat 660K chip, we obtained SNPs and 746 metabolites for mQTL analysis (746 substances were selected from 1,260 metabolites based on sample repeatability), and a total of 1,005 mQTLs were obtained, of which 493 mQTLs were distributed in wheat B Genomically, compared with other classified metabolites, the number of flavonoid-related QTLs (61) is the highest, followed by amino acids, nucleic acids and their derivatives.

We analyzed the distribution of 1005 mQTLs at the entire genome level through the Chi-Squared Test, and found that there are 68 hotspots in total, mainly distributed on chromosomes 1B, 4B, and 7A, but the hotspot 1B is particularly prominent (Figure 2a), we speculate that there may be major regulatory genes in these hotspots, which affect the variation of multiple metabolites in the population. Among them, the genetic hotspots related to flavonoids and phenamines are located on chromosomes 1B and 4B (Figure 2b), and 5 lipids QTLs related to mass metabolism are located at 7A: 240.0-240.8 cM (Figure 2b). The number of mQTLs we detected on multiple chromosomes (such as chromosomes 3A and 4D) was significantly lower than expected (Figure 2a).

The number of mQTLs for each metabolite ranges from 1 to 6, and 201 metabolites contain at least two mQTLs; however, some metabolites are affected by a single major mQTLs, for example, we located an n16920 level on chromosome 2A The QTL (a polyphenol is presumably annotated as hydroxycinnamylglycerate) is located between 75.0 and 735.1 Mb (LOD = 15.3), which explains 33.4% of the phenotypic variation; the other QTL for mr1093 is located on chromosome 2B 665.2 ~ 666.4 Mb (LOD = 11.9), which explains 31.2% of the phenotypic variance. These results indicate that a single gene, rather than epistatic interaction, is directly involved in the synthesis of metabolites.

Each mQTL explained 0.8-53.1% of the observed phenotypic variation, with an average of 13.3%, and 263 loci were associated with more than 15% of the phenotypic variation. Among them, the phenotypic variation of QTLs for secondary metabolites explained (PVE) (average PVE of 14.0%) is generally greater than the explanation of the phenotypic variation of the primary metabolite QTL (average PVE of 11.9%). Different PVEs reflect the genetic structure difference between primary metabolism (central metabolism) and secondary metabolism to a certain extent.

Figure 2 Distribution statistics of mQTLs in wheat chromosomes (a) The distribution of 1005 mQTLs and hotspots. The horizontal dashed line indicates the threshold of mQTL hotspots, which is represented by the maximum number of mQTLs. This number is expected to fall into any interval by chance, and the whole genome P = 0.01. The interval size is 10 cm; (b) the distribution of mQTLs for 467 known metabolites. Each row represents the QTL mapping of a single metabolic trait. The metabolites of different chemical groups are represented by different colors. The x-axis represents the genetic position of the wheat genome. The heat map below the x-axis illustrates the QTL density of the entire genome. The window size is 10 cm.

3. Identification of candidate genes for mQTLs

The high resolution of mQTLs facilitates the allocation of metabolite candidate genes. We screened a series of candidate genes by integrating compound structures, known biosynthetic pathways, and wheat genome annotations (Table 1). In the adjacent region, TraesCS5D01G028100, which encodes the amino acid permease family protein, is considered a candidate gene because it is highly similar to the functional annotation genes Arabidopsis and rice genes AtPUT2 and OsPAR1 (at the amino acid level, 70 and 87%, respectively).

In addition, Multiple flavonoids are co-localized on the same locus on chromosome 1A (588.7-593.5Mb). Two genes in the TraesCS1A01G442200 and TraesCS1A01G442300 intervals have high homology with the rice flavonoid 30-hydroxylase encoding gene OsF3’H Sex (70 and 78% respectively at the amino acid level). We further selected two candidate genes from the candidate gene list and verified them by in vitro expression analysis, as shown below. Table 1 List of candidate genes (mQTLs)

The mQTL for the metabolite mr1092 (apigenin 7-O-rutinoside) is located at 5.6–7.2 Mb on chromosome 2B (Figure 3a). This gene region is annotated with a glycosyltransferase gene TraesCS2B01G012000. The amino acid sequence of this gene is similar to that of rice. The amino acid sequence identity of the homologous gene UGT706D1 reached 49.1%. Under the control of the 35S promoter, the coding sequence was cloned from Chinese spring wheat into a Strepll-labeled vector and expressed in tobacco (Figure 3c).

Apigenin and tritice were tested together with UDP-glucose and purified protein as co-substrates and showed that it accepts apigenin but not tritice (Figure 3d). This protein was registered as UGT88C13 by the UGT Committee. When cloning the target gene from two parents, we found that it was difficult to amplify the target gene from the J411 variety. Multiple primer pairs were used. Only KN9204 and CS got positive results. Therefore, during the evolution or domestication of J411 , It is likely that a considerable sequence change or gene loss has occurred.

The translational protein of KN9204 (named UGT88C14) was expressed, extracted and purified in tobacco (Figure 3c), showing a coding sequence and protein activity similar to Chinese spring wheat. This result proves the glycosyltransferase activity of the candidate gene, which explains the different accumulation of glycosylated apigenin in the RIL population. Similarly, using mQTL mr075 to locate another flavonoid-related gene, we found that only 3 genes were located in the interval.

One of the genes, TraesCS2B01G459900, was annotated as a glycosyltransferase, similar to rice UGT706C1 (identified as 52.1% at the amino acid level), so We cloned this gene from CS. Although activity was detected, we noticed that the two parents have the same coding sequence. Next, we used qRT-PCR to determine the relative expression level. The results showed that the relative expression level of J411 gene was 10 times that of KN9204 gene in the tissues harvested in the second week of filling, which was consistent with the accumulation of glycosylation products of J411. It is consistent with the fact that the amount is higher than KN9204.

Figure 3 Functional verification of candidate gene TraesCS2B01G012000 (a) LOD curve of QTL mapping of mr1092 accumulated on chromosome 2B. (b) TraesCS2B01G012000 gene model. (c) The protein encoded by the candidate gene is transiently expressed in Arabidopsis and purified by StrepII. The samples are in different stages of purification. The arrow indicates the purified protein. CBB, Coomassie brilliant blue staining; WB, immunoblotting. (d) Enzymatic reaction of purified protein. The structure of the substrate and product (left) and the chromatogram of the standard and biochemical reaction.

4. Correlation between agronomic traits and metabolites

The 17 agronomic traits of this RIL population were obtained from three independent harvests in the previous period. In order to analyze the relationship between the changes of metabolites and plant morphology, we first determined the CV of 17 agronomic traits, the coefficient of variation was 3.8 ~ 15.7%, and the average H2 value was 0.61, indicating that this variety has significant genetic contributions and beneficial agronomic traits. Manual selection potential. Subsequently, we constructed a metabolism-agronomic traits association network consisting of 467 annotated metabolites and 17 agronomic traits (Figure 4a).

754 significant correlations were detected, the number of positive correlations and negative correlations was roughly the same (Figure 4a), and 264 (56.5%) metabolites were correlated with at least one agronomic trait. For example, mr869 was correlated with 8 agronomic traits. Flavonoids, amino acids, nucleic acids, lipids, phenolamines, and polyphenols are significantly related to 13 agronomic traits (Figure 4a), indicating that metabolites are involved in the formation of agronomic traits, and the correlation between metabolism and agronomic traits is not as good as metabolic traits The correlation between them is close. Among agronomic traits, grain width (GW), harvest index (HI) and thousand-grain weight (KGW) are mainly positively correlated with the annotated metabolites.

Traits related to flag leaf (length of flag leaf, FLL; width of flag leaf, FLW; flag leaf area, FLA) and traits related to number of spikes (number of spikes per plant, NSPP; number of grains per spike, NGPS; number of spikelets per spike, NSPS) Most of the annotated metabolites are negatively correlated (Figure 4a). We found from related data that leaf traits (FLL, FLW, FLA) and grain traits (KGW; grain width, GW; grain length-to-width ratio, LWR) were significantly correlated with 56 and 141 metabolites, respectively (Figure 4a). It indicates that the formation of grain traits may be more complicated than the formation of leaf traits. In addition, 54 metabolites are significantly related to 3 grain traits, indicating that adjusting the content of these metabolites can be used as a strategy to improve grain yield and quality.

5. Co-localization of QTL for metabolites and QTL for agronomic traits

In order to further analyze the internal relationship between agronomic traits and metabolites, we used the data of agronomic traits to perform pQTL analysis. The results showed that a total of 97 pQTLs were obtained from the QTL analysis of 17 agronomic traits, which were mainly distributed on the 2D and 4B chromosomes of wheat. The PVE ranged from 1.9 to 37.6%, with an average of 8.3%, which was significantly lower than the average PVE of mQTLs (13.3). %). Next, we analyzed the relationship between mQTL and pQTL and found that about half of pQTL (48) overlapped with mQTL; there were a total of 369 mQTLs, representing 252 metabolic features (including 61 annotated metabolites) co-localized with pQTL . The most common sites of pQTL metabolites are mr1548 and mr2801, followed by mr107 and mr1203.

The common sites contain 6 pQTLs for 5 agronomic traits. We found several time intervals in the genome that affect the above 10 metabolites, and at the same time affect more than 2 agronomic traits; these intervals are mainly on chromosomes 1B and 4B, for example, LWR and NGPS are in the 23.7-30.9 Mb interval of chromosome 4B The above pQTL co-localizes with 42 mQTLs.

At the same time, the metabolites co-localized with agronomic traits are significantly related to agronomic traits, indicating that related metabolites affect agronomic traits and vice versa. For example, QTLs for the metabolite mr115 (ferulic acid) and 3 flavonoids (mr1114, n03958, mr1120) Co-localized with the harvest index (HI) QTL on chromosome 1B. And we also observed this result on mr1222, PH and n04711 (pyranose derivatives), and the aspect ratio of the seeds is also the same (Figure 4a). Interestingly, the two metabolites involved in auxin synthesis, mr1346 (tryptophan) and wm0034 (4-indolecarbaldehyde), are significantly related to NGPS (number of grains per panicle), and mQTL and pQTL are co-located in the similar segment of chromosome 4B (Figure 4B, C).

Based on common PCR markers and SNP markers, we further analyzed and found that the gene TraesCS4B01G155000 is located in the region. The gene functional annotation encodes auxin and dormancy-related proteins, while mr1346 (tryptophan) and wm0034 (4-indolecarbaldehyde) are located in the auxin synthesis pathway. The above results confirmed the feasibility of using metabolites to analyze complex agronomic traits.

Figure 4 Network analysis of correlation between metabolites and agronomic traits (a) Correlation analysis of 467 annotated metabolites and 17 agronomic traits. The jointly detected metabolites and agronomic traits are represented by nodes, and their correlation coefficient values ​​are represented by edges. The absolute value of Pearson’s correlation coefficient is higher than the threshold (P<0.01). Different colors represent different kinds of metabolites. The circles and green hexagons represent metabolites and agronomic traits, respectively, and the size of the shape represents the associated quantity. The correlation level is expressed in red (positive correlation) or blue (negative correlation).

The intensity of the color indicates the correlation, and the darker the color, the stronger the correlation. Yellow circles indicate metabolites that are significantly related to co-localization of near-agronomic traits. PR, ear rate; YPP, yield per plant; NSPP, number of ears per plant; AB, aboveground biomass; SDW, dry weight of straw; LWR, seed length-to-width ratio; GW, grain width; NSPS, number of spikelets per ear ; FLW, flag leaf width; FLA, flag leaf area; FLL, flag leaf length; KGW, thousand-grain weight; HI, harvest index; NGPS, grains per spike; GWPS, grain weight per spike; SL, spike length; PH, plant high. (b) Two metabolites (wm0034, 4-indole carboxaldehyde; mr1346, tryptophan) and NGPS. (c) LOD curve of QTL mapping of the number of grains per ear, wm0034 (4-indolecarboxaldehyde) and mr1346 (tryptophan) levels on chromosome 4B. Green represents the number of grains per ear; blue represents 4-indole formaldehyde; red represents tryptophan.

6. Using metabolic data to predict agronomic traits

Based on the BLUP and LASSO models, we combined large-scale metabolic data (1260 metabolite characteristics) to predict 17 agronomic traits. The average prediction value of the BLUP model for 17 agronomic traits is 0.26, and the average prediction value of the LASSO model is 0.27. Among them, the best agronomic traits are PH (plant height) and NGPS (number of grains per panicle), the average prediction of the two models The values ​​are 0.51 and 0.49 respectively (Figure 5). The LASSO method detected 82 and 98 metabolite characteristics, including plant hormone derivatives, sugars and organic acids, among which some have significant effects on NGPS (grains per panicle) and PH (plant height).

Among these metabolites, mr169 (betaine) and S19-0168 (unknown) have the most significant positive effects on PH (plant height) and NGPS (grains per panicle), respectively, while mr355(2′-deoxylnosine-5 ‘-monophosphate) has significant predictive effect values ​​for both traits. Therefore, the metabolome can predict crop agronomic traits and contribute to breeding improvement.

Figure 5 Prediction of wheat plant height (PH) and grains per spike (NGPS) based on LASSO and BULP models. BLUP and LASSO models were used to predict plant height and grains per spike, respectively. Right: BULP prediction results. Left: Lasso prediction result. The x-axis is the predicted value of agronomic traits, and the y-axis is the phenotypic observation value. The image was made using R (http://www.r-project.org/).

Discusses:

The method of combining metabolomics and genomics has been widely used to determine the genetic basis of metabolic diversity. However, most studies so far have only focused on Arabidopsis, tomato, rice and maize. This study combined metabolomics and high-resolution genotyping to analyze the correlation between gene metabolites and metabolic genomics traits in the RIL population.

1. Metabolomics and mQTLs

The detection of metabolites is the basis for studying the genetic variation of metabolites. In this study, 1260 metabolites were obtained through widely targeted LC-MS/MS, and the chemical structures of 467 metabolites were identified. Compared with previous wheat metabolome studies, the results obtained in this study have made considerable progress in the detection of metabolites, including important compound categories, such as polyphenols and flavonoids, which are used in plant biological/non- It is necessary in biological stress and has multiple effects on human health.

The main metabolites usually show strong correlations, such as amino acids, nucleic acids, plant hormones and lipids (Figure 1c), which is consistent with the results of previous studies on rice, wheat and tomatoes. At the same time, we found correlations with metabolites, some of which have strong correlations, such as phenolamides and flavonoids (Figure 1c). The correlation analysis between metabolites not only reflects the relationship between known molecules, but also reflects the relationship between unknown molecules and known molecules, providing important resources for identifying unknown metabolites and pathways in the future.

Based on the linkage analysis of the Wheat660K high-density genetic map, we found that there are 1005 mQTLs randomly distributed in the wheat genome (Figure 2), of which many high-resolution mQTLs have been reported. In addition, we observed that the emergence of mQTLs affected the levels of different metabolites and identified 68 hotspots in the grains, most of which are located on chromosomes 4B and 1B (Figure 2). These hotspots are in Arabidopsis, rice, and tomato. There are also findings in the study of corn and corn that this phenomenon is common and important. These findings indicate that many metabolites can be affected by manipulation of small genome regions, indicating that manipulation of metabolism through breeding is practical.

2. Candidate genes and pathway analysis

Compared with earlier studies, an important advantage of this study is that the wheat hexaploid genome can be obtained, and candidate genes can be identified directly from QTL mapping. In this study, 24 candidate genes were assigned through annotation and research on corresponding genes in model plants (Table 1), and two candidate genes for mQTL mapping were verified through recombinant protein activity determination or mRNA expression analysis (Figure 3).

For the first candidate protein, the protein was confirmed to be a UDP-glycosyltransferase (UGT), which can glycosylate the different oxygen atom positions of the flavonoid A and B rings. According to our enzymatic test, the UGT accepts apigenin, luteolin, capitol and quercetin, but does not accept B-ring methylated flavonoids. When glucose is above 4′-OH, it is more inclined to the 7-OH position. This multi-position glycosylation has been observed in rice before; however, studies have proved that the main position-specific glycosylation, Among them, the two main flavonoids UGTs are responsible for the glycosylation of the 7-OH and 5-OH groups of rice flavone (OsUGT706D1 and OsUGT707A2, respectively). Our verified TaUGTs and other known UGTs have established a phylogenetic tree.

The results show that TaUGTs are classified as the UGT88C subgroup, but they are not well identified. According to the results, this subgroup may mainly play a role in flavanol 7-o-glucosyltransferase, and does not exclude glycosylated 5-OH and 30-OH groups at the same time, depending on the modification of the ring. Different from the first candidate gene, the second verification gene TraesCS2B01G459900 encodes UGT706E7 (identified by the UGT committee), which plays a role in the changes in the expression level of the corresponding metabolites during the grain filling process. The purified protein showed activity against the substrate 3′,4′,5′-trimethoxyflavonoids (contains the glycosyl donor UDP-glucose), and showed less activity against the substrate chlorophyllin.

The protein has good activity on all flavonoids methylated at the 3′, 4′, and 5’positions. The genes in the candidate gene list are related to a variety of metabolic pathways, including flavonoids, phenol amides and amino acids (Table 1). Flavonoids account for the largest proportion of classified metabolites, for example, mapped by mr1120 and mr1112 (Table S4) The TraesCS1D01G020700 has a fairly large PVE, which is about 300 kb away from the confidence interval. Its homologous gene in rice (LOC_Os02g28170) encodes osmat2, which is verified as a flavonoid malonyltransferase by recombinant protein analysis.

Its corresponding homologous gene in maize (GRMZM2G387394) encodes AAT1, which is the first monocotyledonous plant anthocyanin acylase we discovered through mutation phenotype analysis. Based on these findings, we assigned TraesCS1D01G020700 genes. These designated genes have not been reported in common wheat, but their functions need to be further confirmed, as are other genes in the candidate list. The large-scale and high-resolution characteristics of mQTLs in this study benefited from the high coverage, sensitivity and accuracy of the metabolomics methods used, as well as the high density of SNP markers.

In future studies, the hundreds of loci identified in this study will be further verified and identified, which will help to analyze the molecular basis of the metabolic variation of common wheat and clarify new functional proteins and metabolic pathways.

3. Relationship between metabolic traits and agronomic traits

Metabolites are considered to be the bridge connecting the genome and phenotype. Therefore, the study of phenotype and metabolism-related properties reflects the value of this bridge to a large extent. In potato QTL analysis, studies have found that metabolites co-localize with starch and cold sweetener-related traits; other studies have shown that trigonelline has a positive effect on GW by extending the duration of the G2 phase and the entire cell cycle.

Further studies have shown that analysis of metabolites-co-localization of agronomic traits can help infer the genetic link between corn and tomato. In this study, mQTL analysis showed that wm0034 and mr1346 are co-localized, both of which are located in the tryptophan pathway and are involved in the biosynthesis of auxin. Through network analysis (Figure 4a), NGPS is significantly related to these two metabolites.

In addition, we found in pQTL analysis that the corresponding locus of NGPS and the aforementioned mQTL co-localize on chromosome 4B (Figure 4c). We found an auxin inhibitor in the wheat genome annotation

4. Prediction of agronomic traits

In molecular breeding, genomic selection (GS) is more effective than traditional molecular marker-assisted selection (MAS). With the development of high-throughput sequencing technology, transcriptome and metabonomic technology, multi-omics data is used to predict complex agronomic traits, and great progress has been made in crop research. In this study, we used BLUP and LASSO methods to prove that the predictability of yield-related traits (PH and NGPS) reached 0.56 and 0.51, respectively (Figure 5). This result is comparable to previous studies.

Some studies use 1,000 metabolomics feature data from 210 RILs to effectively predict KGW and other traits. Using BLUP and LASSO, the average predictable KGW is 0.55. The LASSO model can effectively screen more than 1,000 metabolites, and a limited number of metabolites that have an important impact on phenotype prediction, as shown in this study. In order to compare predictions using metabolic data and genotype data, we performed the same prediction using genotype data.

Compared with the predicted values ​​of genotype data (0.47 and 0.44), LASSO has higher predicted values ​​of NGPS and PH, which are 0.51 and 0.46, respectively (Figure 5). However, under the BLUP model, these values ​​are opposite. When the number of objects increases to thousands, or when combined with other omics data such as transcriptome and genomic data, the ability to predict should be improved. Therefore, we speculate that these high-efficiency metabolite characteristics are important in biomarker-assisted breeding, and may accelerate plant breeding by providing earlier generation selection.

Conclusion:

Plants produce a large number of metabolites that are very important for their development and growth. However, the genetic structure of the wheat metabolome has not been well studied.

In this study, a high-density genetic map was used to conduct a comprehensive metabonomic study on wheat grain metabolism through widely targeted LC-MS/MS; further combined with agronomic traits, the genetic relationship between metabolites and agronomic traits was analyzed, and a total of 1,260 were detected. Metabolic characteristics.

Through linkage analysis, we found a total of 1,005 metabolic quantitative trait loci (mQTLs) that are unevenly distributed. As a result, we found that 24 candidate genes regulate the levels of different metabolites, and 2 of them are functionally annotated by in vitro analysis to participate in the synthesis and synthesis of flavonoids. Retouch.

Through correlation analysis of agronomic traits of metabolites, co-localization of methylation quantitative trait loci and phenotypic QTL, we revealed the genetic relationship between metabolites and agronomic traits, for example, using correlation and co-localization analysis to identify a candidate Variety, this variety may manage the accumulation of auxin, which affects the number of grains per panicle (NGPS).

We used metabolomics data to predict the performance of wheat agronomic traits, and found that metabolites have a strong predictive ability for NGPS and plant height. This study uses metabolomics and association analysis methods to better understand the genetic basis of wheat metabolism and provide a scientific basis for wheat breeding.

(source:internet, reference only)

Disclaimer of medicaltrend.org