Perspectives of Molecular Design Breeding in Rice

(Perspectives of Molecular Design Breeding in Rice)

The selection and promotion of new varieties of high-quality, multi-resistant, stress-resistant and high-yielding rice crops is an important way to realize food security in China. At present, most of the breeding work is still based on phenotypic selection and breeder's experience, and the breeding efficiency is low; on the other hand, the amount of data accumulated in the bioinformatics database is extremely large, but due to the lack of the necessary data integration technology, the information that can be utilized by the breeders is very limited. Rice molecular design breeding will be at a multi-level level to study all components of rice network interactions and in the process of growth and development of the kinetic behavior of the environmental response; and then use a variety of "genomics" data, in the computer platform on the growth, development and external response behavior of rice prediction; and then according to the specific breeding goals , to build the blueprint for variety design; and then according to the specific breeding goals , to build the blueprint for variety design; and then according to the specific breeding goals , to build the blueprint for variety design. Then, according to the specific breeding objectives, the blueprint of variety design is constructed; finally, new rice varieties that meet the design requirements are bred in combination with the breeding practice.

The core of design breeding is the establishment of molecular design as the goal of the breeding theory and technology system, through the integration and integration of various technologies, the organism from the gene (molecular) to the whole (system) at different levels of design and operation, in the laboratory on the breeding program of the various factors for simulation, screening and optimization, to put forward the best parental selection and progeny selection strategy to achieve the traditional "experience breeding" from the "experience breeding". The design and operation of the program is based on the simulation, screening and optimization of various factors in the breeding program in the laboratory, and the optimal parental selection and progeny selection strategies are proposed to realize the transformation from the traditional "empirical breeding" to the directional and highly efficient "precision breeding" to substantially improve the breeding efficiency.

1 The current status and development trend of basic research related to crop molecular design breeding,

1.1 Bioinformatics Genetic Information Database (GID) data is "exploding"

1.2 The development of molecular labeling technology is rapidly changing

1.3 The research on gene and QTL localization is widely and y developed

1.4 The development of molecular labeling technology is rapidly expanding. QTL localization research is extensive and in-depth

1.4 Gene e-localization and e-elongation have been applied

2 The time is ripe to carry out molecular design breeding in China

The completion of whole genome sequencing of Arabidopsis thaliana and rice has led to a rapid development of plant genomics from structural genomes to functional genomes and other types of "genomics". Genomics and proteomics have developed rapidly. Genomics and proteomics with the power of bioinformatics allows people from the molecular level to understand the physiological activities of plant sub-cells and eukaryotic organisms, how the multi-cellular organization and the realization of its complex functions, a variety of "genomics" to the traditional biology rapidly into the new era of systems biology, this revolutionary change This revolutionary change has given birth to the concept of molecular design. At present, there are many research organizations are doing preparatory work, towards this direction. The U.S. Department of Agriculture has invested in more than a dozen research units to establish

a variety of crop databases, the integration of these databases will become an important basis for future molecular design breeding. Other influential research institutions, such as Pioneer Corporation in the United States, the University of Queensland and CSIRO in Australia, as well as the International Maize and Wheat Improvement Center (IMWIC), have carried out research in the areas of genotype-to-phenotype modeling, analysis of genotype-environment interactions, and simulation of breeding

. China Rice Research Institute in 2004 put forward the concept of rice gene design breeding, that is, after the completion of the whole genome sequencing of rice, in the main agronomic traits on the basis of clear gene function, through the favorable gene shearing, aggregation, breeding in the yield, rice quality and resistance and other aspects of the breakthrough in the new super rice varieties.

Currently in China to carry out molecular design breeding time is ripe, its main performance has the following four aspects. (1) China has bioinformatics research power and technology. China is the world's first rice genome sequencing and rice chromosome 4 fine sequencing of one of the countries. In the process of sequencing, bioinformatics tools are the key to complete sequence assembly and analysis. Being able to accomplish these massive sequencing tasks itself indicates that China has a high level of bioinformatics research. In addition, some progress has been made in the discovery of new markers and genes from genome sequences, EST information and full-length cDNA sequences. (2) Virtual molecular breeding has been carried out. China's use of molecular quantitative genetics and computer technology to study the QTL mapping, QTL and the relationship between the environment is located in the international level, the National High Technology Research and Development Program (863 Program) has been funded to carry out virtual breeding research in major crops, backcrossing breeding, polymerization breeding, hybrid dominance prediction and parental selection computer simulation research has made some progress. (3) We have the technology and experience of establishing large-scale data collection and processing system. China's national crop germplasm resources information system has been established for many years, and the data stored in the system has amounted to tens of millions of items, which has accumulated rich experience in the establishment, improvement and maintenance of large-scale database. (4) We have key technologies such as gene mapping, comparative genomics research and allelic diversity research. China's genetic mapping of crops has been carried out later than that of foreign countries, but it has made great progress in recent years and has covered most of the important traits of various important crops. The mapping of specific genes or genotypes using DNA and protein sequence information is also developing rapidly in crops other than rice. In addition, comparative genomics studies among species within the wheat family and among gramineous crops have been carried out in China and have made some progress, and the ongoing study on allelic diversity has also achieved some milestones. Compared with similar research in foreign countries, the gap between us mainly exists in the following three aspects. (1) The main agronomic traits gene discovery and functional research is insufficient. In the past decade, China's use of molecular markers, in rice, wheat, maize and other major crops have carried out a large number of genes (especially QTL) localization research, accumulated a large amount of genetic information. However, this information is still in a fragmented state, lack of concentration, generalization and summarization; on different genetic backgrounds and environmental conditions under the QTL effect, QTL allelism and different QTL interactions are not systematic and comprehensive, is not conducive to the transformation of the results of the QTL localization of the actual benefits of breeding; important agronomic traits of the genetic basis of the formation of the mechanism and the metabolic network is still very lacking, and these are precisely the important information base for molecular design breeding. The research on genetic basis, formation mechanism and metabolic network of important agronomic traits is still lacking, which is the important information basis for molecular design breeding. At the same time, the lack of computer software with independent intellectual property rights limits the application of existing genes or QTL information to actual breeding. (2) The information system related to molecular design breeding is not perfect. In the National High Technology Research and Development Program (863 Program) and the National Key Basic Research and Development Program (973 Program) and other projects under the strong support of the genetic research on major economic traits of major crops has made great progress, the country has launched a comprehensive study of the functional genome of rice and other major food crops and other major economic traits, but from a comprehensive understanding of the genetic basis of all traits of the crop is still relatively far away. China's existing bioinformatics database, has a clear function and expression of the regulatory mechanism of gene information is relatively scarce; in transcriptomics, proteomics, metabolomics and phenomics and other aspects of the research and the international community there is a large gap in the information system of crop germplasm resources, can be molecularly designed to apply the breeding of information is still very limited. (3) molecular design breeding theory research is relatively lagging behind. At present, the country has begun to realize that molecular design breeding will become the future direction of crop breeding, but most of them are still stuck in the concept, has not really carried out molecular design breeding theory modeling and software development work.

3 China's crop molecular design breeding research focus

China's crop molecular design breeding research should focus on the following three aspects. To efficiently discover genes Π QTL for important agronomic traits by constructing high-generation backcross introgression line populations of rice, wheat, maize, soybean and cotton, etc. Through large-scale backcrossing of introgression lines combined with directional selection, to eliminate the adverse effects of complex genetic backgrounds on the localization accuracy of genes Π QTL, and to efficiently discover genes Π QTL for important agronomic traits in germplasm resources. By analyzing and comparing the localization results of different parents and donor parents in different rounds of high generation backcrossing combinations, the information on the multiple effects of genes Π QTL, multiple effects of genes Π QTL, double allelicity of the same genes Π QTL loci, epistatic interactions among genes Π QTL, interaction between genes Π QTL and genetic backgrounds, and interactions between genes Π QTL and environments were explored. The genetic background imported by high generation backcrossing is highly purified, which is convenient for fine localization of gene ΠQTL with large main effect and stable expression.

Establishing genetic information linkage between core germplasm and backbone parentsCore germplasm represents the greatest genetic diversity with the smallest number of resources, i.e., retaining the smallest possible population and the greatest possible genetic diversity; backbone parents are the breeding materials that have been widely used in current crop breeding and have achieved good breeding results, which contain a large number of favorable genetic resources. By exploring the genetic information in these two types of materials and establishing their molecular design breeding information system and links, we can quickly obtain information on the genes carried by the parents and their interactions with the environment, and provide information support for the molecular design breeding model to accurately predict the performances of the progeny of crosses between different parents in different ecological environments. The GP (Genotype to phenotype) model describes how different genes and genotypes, as well as genes and the environment interact with each other to produce the phenotypes of different traits, so as to identify the target genotypes that can meet the needs of different breeding objectives and ecological conditions, and therefore the GP model is a key component of molecular design breeding. Using the information provided by the genetic information discovered, the core germplasm and the genetic information linkage of the backbone parents, combined with the biological characteristics of different crops and the breeding objectives of different ecological regions, the GP model can simulate and optimize the various indexes of the breeding process, predict the probability of the hybrid progeny of different parents to produce the ideal genotypes and breed good varieties, and greatly improve the efficiency of the breeding process.

4 Examples of molecular design breeding

Peleman and van der Voort trademarked the term "Breedingby design" (Breedingby design), and they believe that molecular design breeding should be carried out in three steps: (1) locate all the QTLs for the agronomic traits; (2) identify the QTLs for the agronomic traits; (3) identify the QTLs for the agronomic traits; and (4) identify the QTLs for the agronomic traits. They believe that molecular design breeding should be carried out in three steps: (1) locate all QTL for agronomic traits; (2) evaluate allelic variation at these loci; and (3) carry out design breeding. The process of molecular design breeding is illustrated here with the results of some of our studies in rice

.

Research on QTL for breeding target traits

This process includes the steps of constructing mapping populations, screening polymorphic markers, constructing marker linkage maps, evaluating quantitative trait phenotypes, and QTL analysis. Here, a population of 65 chromosome segment substitution lines (CSSL) was generated from two parents, japonica Asominori (background or rotating parent) and indica IR24 (donor or non-rotating parent). Each CSSL contained one or several chromosome fragments from IR24, and the rest of the chromosomes were from the background parent Asominori. all donor chromosome fragments covered the whole genome of IR24, and different chromosome fragments were indicated by different RFLP markers. Based on the observed values of grain length, it was possible to analyze the significance of the differences in grain length between genotypes of different markers to determine which segments carried QTLs affecting grain length, and the likelihood of the presence of QTLs was commonly measured by the magnitude of the LOD values (Fig. 12A), and Fig. 12A clearly showed that the chromosome segments represented by markers M23 and M34 contained QTLs controlling grain length, and explained 3619 % and 819 % of the variance in the phenotypic variation of grain length respectively. They explained 3619% and 819% of the phenotypic variation in grain length, respectively, and could be regarded as dominant QTLs, especially on the M23 chromosome segment. However, the additive effects of these two QTLs were in the opposite direction (Fig. 22 B), i.e., for the QTL on the M23 marker, the allele from IR24 increased the grain length, and the allele from Asominori decreased the grain length; for the QTL on the M34 marker, the allele from IR24 increased grain length; for the M34 marker, the allele from IR24 increased grain length; for the M34 marker, the allele from IR24 increased grain length; for the M34 marker, the allele from IR24 increased grain length. For the QTL on marker M34, the allele from IR24 decreased the grain length and the allele from Asominori increased the grain length.

In practice, the LOD threshold can be chosen to determine the existence of QTL according to different research purposes. If the research purpose is to analyze the cloning and function of QTL, a higher LOD threshold, such as 310 or higher, should be chosen to avoid false positives; if the purpose is to predict the genotypes, false positives will not have a large negative impact on the results, and a lower LOD threshold should be chosen to avoid false positives. If the purpose is to predict the genotype, the false positives will not have a large negative impact on the results, then a lower LOD threshold, such as 110, to ensure that the QTL with a small effect can also be identified. In the above example, when using the critical value of 0183 (corresponding to the significance level of 0105), we identified 13 QTL for grain length, 8 QTL for grain width, and some epistatic QTL.

The process of designing target genotypes in relation to breeding goals utilizes the information of QTL for a variety of important breeding traits that have already been identified, including the location of QTL on the chromosome. This process utilizes the information of the identified QTL for various important breeding traits, including the location of the QTL on the chromosome, the genetic effect, the interactions among the QTL, the interactions between the QTL and the background parents and the environment, etc., to simulate and predict the expression of various possible genotypes, and to select the genotypes that meet the specific breeding goals. In our example, Asominori is a short-grain and wide-grain variety, and IR24 is a long-grain and narrow-grain variety, but their CSSL progeny showed superparental segregation for both traits in four directions, so the QTL for enhancement and de-enhancement of the two traits should be distributed among the two parents. By mapping the QTL for grain length, we found that five QTLs on chromosome segments M6, M12, M14, M23 and M25 had positive effects, i.e., for the QTLs on these segments, the allele from IR24 increased the grain length, while for the grain width, there was only one QTL that had a positive effect, which indicated that most of the genes that increased the grain width were from Asominori. In addition, some chromosome segments, such as M10, M12, M14 and M23, were found to carry QTLs for both grain length and width; in segment M10, the QTLs had positive effects on both grain length and width, while in the other segments, the QTLs had opposite effects on both grain length and width. This is consistent with the correlation coefficient r =- 0.34 estimated from the phenotypes.

Based on the above information, the phenotypes of various possible genotypes could be predicted (Fig. 2), and it was found that the minimum and maximum grain length genotypes had grain lengths of 4.20 mm and 6.21 mm, respectively, and the minimum and maximum grain width genotypes had grain widths of 2.12 mm and 3.07 mm, respectively. It is impossible to obtain a genotype with both the maximum grain length and the maximum grain width as shown in Fig. 2, due to the negative multiplicity of some QTLs for the two traits. After simulation, we found a genotype with grain length of 6.05 mm and grain width of 3.00 mm, which is close to the maximum grain length of 6.21 mm and maximum grain width of 3.07 mm (Figure 2). Thus, we have designed a genotype that best meets the breeding objective of long and wide grains.

Analysis of the pathways to achieve the target genotypes

To obtain the designed genotypes in Fig. 2, four chromosome segments of IR24, i.e., M1, M6, M23, and M25, were required. 65 CSSLs were available, including CSSL5 containing fragment M6; CSSL16 containing fragments M1 and M23; and CSSL19 containing fragment M25; and thus could be used as parental materials for generating the designed genotypes. CSSL19 contains fragment M25; therefore, it can be used as parental material to generate the designed genotypes. However, CSSL19 contains fragment M12, which we do not need, and we need to replace it with the fragment of Asominori during the selection process.

Triple-crosses (also known as top-crosses) among the three parents have the potential to aggregate the chromosome segments we need, and there are three ways to generate triple-crosses, namely, triple-cross 1: (CSSL5 × CSSL16) × CSSL19; triple-cross 2: (CSSL5 × CSSL19) × CSSL16 and triple-cross 3: (CSSL16 × CSSL19) × CSSL19; triple-cross 3: (CSSL16 × CSSL16) × CSSL19; triple-cross 2: (CSSL5 × CSSL19) × CSSL19; triple-cross 3: (CSSL16 × CSSL19) × CSSL16). Assuming that marker-assisted selection of target genotypes is adopted, there are many options available, only two of which will be considered here, Marker selection option 1: Generate 100 triple-crossed F1 individuals, each with 30 F2 individuals, and generate 3,000 F8 lines using single-seeded passages***, and then select the target genotypes from them. In marker selection scheme 2, 100 three-cross F1 individuals were generated, and only individuals containing the target chromosome segments were retained through marker-assisted selection. 30 F2 individuals were generated from each selected individual, and F8 lines were generated by single-seeded transmission, from which the target genotypes were selected. The above process was realized with the help of QuCim, a genetic breeding simulation tool. For each three-cross combination, the F8 families obtained from the two marker selection schemes were equal (Table 1). An average of 716 F8 families of the target genotypes were obtained from triple-cross combination 1, 2318 from triple-cross combination 2 and 1118 from triple-cross combination 3 (Table 1). However, the number of DNA samples to be tested for each selected F8 line was 395 for the two marker selection schemes 1 and only 60 for the marker selection scheme 2. Therefore, the scheme 2, which includes two stages of marker selection, significantly reduces the cost of laboratory determination of the markers during the gene polymerization process. The chances of obtaining the target genotype varied significantly among the triple-cross combinations

. Triple-cross combination 2 had the highest probability of 0181 %, while combination 1 had the lowest probability of 0125 %. Therefore, triple-cross combination 2 combined with marker selection scheme 2 is the best way to realize the target genotypes.5 Outlook of molecular design breedingCrop molecular design breeding is a new concept, which is based on bioinformatics as a platform, based on several databases such as genomics and proteomics, and integrating the useful information from all disciplines of crop genetics, physiology, biochemistry, cultivation, biostatistics and other subjects in the process of crop breeding, according to the specific breeding objectives and growth of the crop, and the genotypic characteristics of the crop. Based on the breeding objectives and growing environment of specific crops, the best program is designed on the computer, and then the molecular breeding method is carried out for crop breeding experiments. Compared with conventional breeding methods, crop molecular design breeding is first simulated on the computer to implement, consider more factors, more thorough, and therefore the selected parental combinations, selection pathways, etc. more effective, more able to meet the needs of breeding, can greatly improve the efficiency of breeding. It is worth pointing out that molecular design breeding in the future implementation process will be a combination of molecular biology, bioinformatics, computer science, crop genetics, breeding, cultivation, plant protection, biostatistics, soils, ecology and other multidisciplinary systematic engineering. Crop molecular design breeding is a comprehensive emerging research field, which will have a far-reaching impact on the future development of crop breeding theory and technology. Therefore, we should grasp the opportunity to take full advantage of plant genomics and bioinformatics and other cutting-edge disciplines of the major achievements, and timely development of varieties of molecular design of basic theory research and technology platform construction. To achieve the goal of molecular design breeding, will greatly improve the theoretical and technical level of crop breeding, drive the traditional breeding to high efficiency, oriented development.