Sveriges lantbruksuniversitet

Development of methods and technologies

Here you can read summaries of some of the publications from our research on development of methods and technologies. You can read more about the research projects here

Does genomic selection have a future in plant breeding?

Selection based on genomic information, often called genomic selection, is a recently developed concept that aims to use genetic markers across the whole genome as a selection tool in breeding programmes. Genomic selection is a so-called “black box” approach because associations rather than knowledge of the biological functions of genes are used. This technique has been developed from previous marker-based techniques and became useful when genotyping became more affordable. Today this method has been integrated into dairy cattle breeding programmes and is being tested in other livestock and plant species.

In the process of crop breeding few favourable plants are being selected among numerous lines. Selection starts at early stage of the development and is based on measurable traits during those stages, often visible information such as height or colour. However, the genetic progress for traits measurable at later stages (for example, increases in yield) is predicted to be low, compared to more accurate assessments of phenotypes and information showing high correlation with the final traits, such as the breeding value. Therefore, many plant-breeding companies aim to integrate genomic information into their programmes. Furthermore, numerous studies using simulated and empirical data have been published, often promising high reliabilities when using breeding values based on genetic markers, so-called “genomic breeding values”.

But how realistic are those assessments? In a paper by Elisabeth Jonas and Dirk Jan de Koning, they conclude that in many approaches genetic markers can be used as an additional tool with a better predictive value compared to some of the current selection steps in a breeding cycle. Genomic selection has been proposed to both provide a more accurate estimate of genetic merit as well as to shorten the generation interval. At present, however, strategies to shorten the breeding cycle are rarely discussed.

Jonas and de Koning emphasise that a functioning collaboration between geneticists, farmers, breeders, and seed producing companies is needed to integrate novel approaches for improved breeding into realistic breeding schemes. Also, because there is a huge difference between cattle breeding and plant breeding, methods successfully applied in the former have to be carefully considered for the latter.

Jonas, E., & de Koning, D.J. 2013. Does genomic selection have a future in plant breeding? Trends in Biotechnology 31: 497–504

Protein navigation in plant breeding

By the development of methods for analyzing plant proteins, it is possible to find out which plants are the best to cultivate. The researchers have found proteins in potato typical for high yield, and resistance against late blight. They suggest a new workflow, including both DNA- and protein analyses, to accelerate the breeding for important crop traits.

Senior lecturer Fredrik Levander, at the Department of Immunotechnology at Lund University, is one of the researchers behind a study showing that it is possible to determine which potato plants are high-yielding, and which are resistant to late blight, by analyzing the potato plants' proteins.

– It can be tricky to predict the cultivation characteristics of, for example, a potato by just looking at genes. Proteins bring us closer to what actually happens when the plant grows. Although, at the same time, it is experimentally easier to measure DNA variations, he explains.

Using mass spectrometry, more specifically the technology selected reaction monitoring, it is possible to find out the amounts of selected peptides and proteins in a plant. First of all, it is important to find out which of these molecules are specific to the plants with good properties. Then, plant breeders can use the peptides and proteins as markers for the properties such as resistance to disease, drought tolerance or adaptation to a particular region.

The markers can thus tell the plant breeder what properties the plants have, without having to wait and identify the properties per se, among the grown up plants. This saves energy, time and money.

There are already methods for making selections based on the genome, using DNA markers. But the researchers see benefits of supplementing DNA markers with protein markers. The proteins say more about which biological processes are "going on" in the plant, compared to DNA markers. Some genes are not expressed, and some genes give rise to several different proteins.

– There are often several copies of genes present in plant genomes, and several variants of genes that are very similar to each other. That makes it difficult to predict plant characteristics from DNA. Especially in plants that are tetraploid, hexaploid, and so forth, says Fredrik Levander.

You get closer to the truth with protein analyzes, compared to DNA analyzes, but it's trickier to analyze the proteins. DNA sequences consist of four different bases, and there are easy methods for amplifying large amounts of DNA. The proteins consist of many more constituents (amino acids), and the researchers are limited by how much protein they manage to get from a leaf or a potato tuber.

- Different proteins behave very differently, therefore it is more complicated to analyze them, compared to analyzing DNA.

In the current study, the researchers selected 104 protein markers. Some of these could be used to predict higher potato yield, and resistance to the oomycete Phytophthora infestans causing late blight in foliage and tubers. In other words, they found protein markers for properties for which there are no commercial DNA markers yet.

Chawade, A., Alexandersson, E., Bengtsson, T., Andreasson, E., & Levander, F. 2016. Targeted proteomics approach for precision plant breedingJournal of Proteome Research 15(2): 638-646

Goals and hurdles in genomic selection in crops

In a review paper Elisabeth Jonas and Dirk-Jan de Koning have summarized the goals and hurdles for a successful implementation of genomic selection in breeding programmes for rice, maize, wheat, barley, and forage grass, i. e. both annual and perennial crops.

Further studies on methods and applications of genomic selection are required in order to allow highly accurate predictions across the variety of crop populations. Such studies should focus on population specific requirements (e. g. self- or cross pollination). They should also include non-additive effects, especially genotype-by-environmental interactions. Each crop species has its own specific setting, and the application of strategies based on genomic selection is often far from easy.

A similar conclusion was drawn from a literature study of the application of genomic selection in different livestock populations. Genomic selection is already being used as a selection tool in dairy cattle, and the tool is also part of the planning of breeding programmes in many livestock populations (for example pig, beef cattle and chicken). Genomic selection needs to be carefully assessed to meet specific requirements in both crop and livestock breeding programmes.

Collaboration with the industry is needed in applications of genomic selection, since relevant findings can only come to use via the work of industry partners. Future studies should be more targeted towards breeding programmes and should be specific for different populations. Crop breeding can also be improved by further advancement of methods which include environmental factors. They finally conclude that an open exchange on the status of research results and achievements in breeding programmes is required to achieve significant successes.

Jonas, E., & de Koning, D.J. 2016. Goals and hurdles for a successful implementation of genomic selection in breeding programme for selected annual and perennial cropsBiotechnology and Genetic Engineering Reviews 32: 18-42

Proteins can tell the story on the course of the disease

Phytophthora infestans, an oomycete, is the most harmful pathogen of potato. It causes the disease late blight, which generates increased yearly costs of up to one billion euro in the EU alone, and it is tough to control. Svante Resjö and colleagues tried to find out more about how the late blight pathogen P. infestans acts during the disease infection, and managed to identify proteins that seem to be of high importance for the pathogen at different life stages.

Among 10 000 peptides, from over 2000 proteins, they found 59 interesting ones that were highly abundant in the pathogen at pre-infectious life stages, i. e. in its germinated cysts and the cells that penetrate into the potato plant. A large majority of these proteins have not been recognized as being part of this infection process before, but based on their similarity to other proteins, with known function, the researchers could predict that they play roles in transport, amino acid metabolism, pathogenicity and cell wall structure modification.

The researchers also analyzed the expression of the genes encoding nine of these proteins and found an increased level during disease progression, in agreement with the hypothesis that these proteins are important for the infection to happen. Among the nine proteins was a group involved in the pathogen’s struggle to modify and hold on to the cell wall structure. Silencing of these genes resulted in reduced severity of the infection, additionally indicating that these proteins are important for pathogenicity.

Resjö, S., Brus, M., Ali, A., Meijer, H.G.J., Sandin, M., Govers, F., Levander, F., Grenville-Briggs, L., Andreasson, E. 2017. Proteomic analysis of Phytophthora infestans reveals the importance of cell wall proteins in pathogenicityMolecular and Cellular Proteomics 16:1958-1971

A new way of gene editing potatoes

Mariette Andersson and her colleagues have found a new way of using the so-called genetic scissor Crispr/Cas9 so that the DNA encoding it does not end up in the genome of the edited potato plant. Thereby they have further developed the method of genome editing of potato.

Instead of allowing the DNA, coding for the scissor, to code for the RNA and protein that make up the Crispr/Cas9-complex inside the plant, temporarily, the RNA and the protein complex are produced outside the plant before allowing the complex to do the job in the cell.

The researchers took cells from the potato, removed the cell walls and added the Crispr/Cas9 complex (i.e. the gene scissor and not the DNA that encodes for the scissor). Then the cell cultures were grown into plants. The researchers checked that the mutations were in the right places, and examined whether any DNA was accidentally inserted into the genome. With this new "DNA-free" method, the researchers produced two potato clones with mutations in all four alleles of the gene (as most cultivated potatoes this was a tetraploid, thus having four sets of chromosomes) without any accidentally added DNA.

Different labs all over the world are working to adapt the Crispr/Cas9 method to work efficiently and with great precision in various plants and animals. This study is a step along the way to refine the potato genome editing technology.

The gene the researchers chose to edit in the potato is involved in the biosynthesis of the amylose starch molecule. The gene got a mutation that prevented the production of amylose in the plant, and instead more of the amylopectin starch molecule was produced.

Andersson, M., Turesson, H., Olsson, N., Fält, A.S., Ohlsson, P., Gonzalez, M.N., Samuelsson, M. & Hofvander, P. 2018. Genome editing in potato via CRISPR‐Cas9 ribonucleoprotein deliveryPhysiologica Plantarum doi: 10.1111/ppl.12731

Better prediction of desired traits by improved genomic selection model

In both animal and plant breeding, the main goal is to select individuals that will improve future generations with respect to desired traits. For that purpose, statistical methods are applied to predict the genetic values of the individuals, and to rank the candidate progenitors of the future generation. In the last decade genomic selection, which is a selection method based on the availability of hundreds of thousands of DNA markers, has been used more and more. Today this method is applied routinely for selection in several livestock species like dairy cattle, pig, chicken and beef cattle. Nonetheless, the applications of strategies based on genomic selection are not always straightforward for all species and populations, and studies are needed to determine the specific requirements for each case.

The most common statistical method used to predict the genomic breeding values (GBLUP) assumes that the markers are independent and all have some effect on the analysed trait. However, this is rarely the case. As an example, there are studies indicating that markers located within or near regions rich in genes on a chromosome can explain more variance than markers located in intergenic regions.

Presently, knowledge of the genetic architecture of complex traits is available, and the amount of information on the DNA markers is continuously increasing in the form of accurately annotated genomes, Quantitative Trait Loci (QTLs) databases, gene expression and gene pathway studies. Now attempt has been made to include these types of information in the prediction models in order to improve the accuracy of the genomic breeding values.

The researchers have looked into the possibilities to include explanatory variables for marker-specific variances in a general BLUP analysis method. Starting with simulated data they created different scenarios of the genetic architecture of a trait which were then analysed using the traditional GBLUP model and the alternative model that includes the extra information on the markers. The models were evaluated in terms of their predictive accuracy. Their results indicate that the alternative model tends to perform better compared to the traditional GBLUP, and yield higher accuracies in most of the scenarios tested. Moreover, the researchers identified two factors that influence how well the alternative model will work: 1. The genetic architecture of the trait. The alternative model performed better than GBLUP when the number of genes controlling the trait was low, but its predictive ability decreased with an increasing number of genes involved. 2. The external information. In the study the researchers assumed that the location of the genes was known, and they defined windows of markers around the genes to have higher weights than the rest of the markers. The results show that larger windows tend to decrease the accuracy because the information provided to the model was vaguer in this case. Including biological information into the model is beneficial as it can increase the accuracy compared to the standard GBLUP. Nonetheless, this benefit depends upon the underlying genetic architecture of the trait and on the quality of the external information.

Mouresan, E.F., Selle, M. & Rönnegård, L. 2019. Genomic prediction including SNP-specific variance predictorsG3: Genes, Genomes, Genetics 9:3333-3343 DOI: 10.1534/g3.119.400381

Published: 02 December 2019 - Page editor: