MAS-MABC Outline

Xavier Delannay

November 2009

Marker-assisted selection (MAS) and marker-assisted backcrossing (MABC) have been the first successful wide-scale implementations of marker technology in breeding programs. With MAS, the breeders use markers linked to specific traits to quickly and effectively select plants carrying those traits in segregating populations. This process can be done for a single trait or for a few traits at the same time. The main benefit of MAS is to provide quick and accurate genotyping of progenies for traits that are typically hard or laborious to phenotype accurately. It is also very useful to follow specific genes involved in the expression of a trait, as in the case of some disease resistances that are controlled by a number of different resistance genes. Those multiple resistance sources would be very hard or impossible to distinguish phenotypically. In the case of MABC, the goal is to quickly introgress a mapped trait from a donor parent (typically a poorly adapted line) to a recurrent parent (typically an elite line) via multiple backcrossing steps in order to convert the recurrent parent with the new trait. The use of markers in MABC has two purposes: allowing a clean transfer of the trait with a minimum amount of donor parent genetic material being carried along (foreground selection), and facilitating the quick recovery of the recurrent parent background in a reduced number of backcrossing generations (background selection). Once a trait or a gene has been introgressed into a breeding operation via MABC, it can be easily followed up in subsequent crosses using MAS.

In contrast to applications such as MARS which use de novo QTL mapping as part of their process, the use of MAS or MABC implies prior knowledge of mapping information for the targeted traits. This mapping information will typically have been obtained through earlier mapping experiments conducted with crosses between parents contrasting for the trait. Ideally, the linkage between the traits and the markers should be as close as possible but more loosely linked markers (10 cM distance or even more) will still be useful as long as the genotyping error rate is better than the one observed with the original phenotyping method or as long as the overall process is more efficient. Having closely linked flanking markers will be very useful, especially for MABC. Markers that reside directly on the gene of interest are of course the best option and no flanking markers will be needed for MAS in those cases. A premise of MAS and MABC is that the phenotypic effect associated with the presence of the selected marker allele will hold true across diverse backgrounds and environments. The main benefit of MAS is to replace laborious or inaccurate phenotyping and to select for multiple traits at one time. In traits that are very easy to phenotype (such as spraying herbicide-tolerant progenies with the corresponding herbicide), MAS will not necessarily offer a benefit and accurate "visual" phenotyping may still be done more quickly and more cheaply.

The bulk of MAS and MABC applications have focused on the selection of qualitative traits in segregating populations. Those are traits that are typically conditioned by a single gene that can be closely mapped with markers, and that have significant effects that remain generally consistent across germplasm and environments. Examples of those traits are most disease resistance genes, some abiotic factor resistance traits (e.g. AltSB gene for aluminium tolerance in sorghum, Saltol gene for salt tolerance in rice), and some grain quality traits (e.g. opaque-2 gene for endosperm quality in maize, PinA and PinB genes for grain texture in wheat). More recently, some breeders have started using MAS for the selection of more quantitative traits controlled by multiple loci (QTLs), focusing primarily on the most significant of those loci. Examples of such traits are disease tolerance traits controlled by multiple genes acting in a quantitative way and abiotic traits such as drought tolerance. The successful implementation of a MAS or MABC approach with those quantitative traits will depend in a large part on the consistency of the observed QTL effects across a wide range of genetic backgrounds and environments.

GENERAL OUTLINE OF A MAS PROJECT

Qualitative traits:

Once previous mapping work has helped elucidate the genetics of a trait and has identified marker loci that are closely linked to a particular gene conditioning that trait, that information can be transferred to breeders for use in their programs. If the trait is already present in the local breeding germplasm, MAS can be used directly to select desirable progenies in populations segregating for that trait. If the trait needs to be introgressed from an exotic source, MABC should first be used to transfer the trait into some elite germplasm (see below for the outline of a MABC project). Once this has been accomplished, regular MAS can be used to follow the trait through the subsequent breeding operations.

Timing:

One of the big advantages of MAS is the ability to select plants carrying the trait at an early stage, before phenotyping for the trait would even be possible. Using MAS early in the growing season will allow the early elimination of unwanted plants so that the bulk of the work (crosses, phenotyping for other traits, etc.) can be focused on the plants of interest. Seed chipping has been implemented in many commercial breeding programs and allows for the genetic selection of only desired seed for planting, thereby reducing significantly the amount of field or greenhouse space needed.

Sampling:

Critical attention needs to be placed on accurate labeling of the samples, both in the field plants and in the corresponding leaf samples to be sent to the genotyping lab. Ideally, bar-coding linked to an electronic field book and to a lab LIMS system should be in place, especially for large scale MAS operations.

Genotyping:

In most MAS applications, timely genotyping is very important as the breeders will need the results on time to make crosses or to remove unwanted plants from rows while they are still young. It will be important to use a genotyping services provider that can quickly handle large quantities of samples for DNA extraction and genotyping and that can quickly provide accurate genotypic data in a format that is directly usable by the breeder in the field. The configuration used in MAS applications generally involve large numbers of samples X small numbers of markers (1-10) so the marker platform used will need to be amenable to handle this format at low cost.

Confirmation:

Typically, the exclusive use of MAS will be sufficient for selection for mapped target traits during the breeding process as long as the linkage between the marker and the trait is good enough. However, a final phenotypic confirmation for the presence of the target trait(s) should be done before release of the final varieties. This would typically be done during the final cycle of field testing of advanced lines prior to the selection of the final varieties to commercialize.

Quantitative traits:

Most of the issues raised for the use of MAS for qualitative trait will apply for quantitative traits, with the following caveats:

• Since the precision of the linkage information will typically be lower for quantitative traits, it will be important to use flanking markers and to use markers that bracket the whole significant interval of the QTL region (not just the peak) so that the QTL is not inadvertently lost through the MAB process.
• It will be necessary to conduct phenotypic evaluations at various stages of the process to confirm that the QTL activity holds true while the QTL region is moved into different genetic backgrounds and is tested in different environments.
• Moving multiple pre-mapped QTL regions into different genetic backgrounds, especially combining regions from different sources, may be particularly tricky due to the higher probability of epistatic interactions between those regions and with various regions of the new genetic backgrounds.

GENERAL OUTLINE OF A MABC PROJECT

A MABC project consists of two main parts being performed in tandem:

• Selection of plants carrying the target trait from the donor parent through the successive backcrossing generations (foreground selection)
• Among those plants carrying the target trait, identification at each generation of individuals carrying the largest representation of the genome of the recurrent parent (background selection)

A lot of the considerations raised for the MAS process (timing, sampling, genotyping) will apply just as well for the MABC process. However, while MAS generally follows the normal breeding cycle (main growing season followed by off-season nursery), MABC is best performed in specialized accelerated nurseries where as many generations as possible can be raised quickly in a given year. This is because MABC focuses on a single trait at a time and is relying entirely on genotyping during its process, so it does not need any normal growing season for usual plant evaluation. The goal of MABC is to transfer a trait from one genetic background to another in the minimum amount of time. This is obtained by a combination of a quick generation time and an aggressive use of markers for background selection, which can easily cut down the number of backcrossing generations needed from five to two or three (BC2 or BC3) depending on the targeted quality of the final product. In some crops, the germination of immature seed or embryo rescue can help accelerate the crossing generation cycle. A gain of a few weeks at each generation can make a difference in the year the finished product can reach the full field season for final evaluation. Commercial maize breeding programs use dedicated MABC nurseries in areas such as Hawaii where up to four crossing generations can be obtained per year. Those efforts are focused primarily on the quick conversion of commercial maize inbreds with a variety of transgenic traits (singly or in combination). In those cases, new inbred development is done using nontransgenic germplasm and the final selected inbreds then receive the targeted transgenic traits quickly via MABC.

Foreground selection:

Foreground selection is the part of MABC that is the most similar to MAS. In this case, however, one of the goals besides the selection of the target trait at each generation is to minimize the amount of linked genomic region from the donor parent that end up being transferred along with the trait. In traditional backcrossing, the linked regions from the donor parent can cover a very large span of the chromosome on either side of the introgressed gene even after many generations of backcrossing. This can lead to linkage drag, where deleterious traits from the donor parent are inadvertently transferred to the recipient parent along with the target trait.

Ensuring the cleanest transfer of the target trait includes the following steps:

• The availability of several closely linked markers on each side of the target trait. This is easy for transgenic traits in crops where a dense set of mapped markers is available but could be harder to achieve if the marker-trait linkage is not strong, and especially in the case of quantitative traits where the region to introgress may be quite large.
• Enough plants are screened for the linked markers at each generation to increase the chances of recombination close to the target region. This is done typically in two successive steps:
o In the BC1 generation, the focus is on finding the closest possible recombinations on one side of the target trait (besides ensuring that the proper alleles on the other side are still present). Enough plants are selected at this stage to still allow for background selection (see below).
o In the BC2 generation, the same takes place for the other side of the target trait.
• Selfing will then be needed to fix the introgressed region. That will be done at the end of the background selection process, which may take an additional generation.

This selection of a very clean introgression can thus be done quickly in two generations of backcrossing. One caveat, however, is that the size of the final donor region surrounding the introgressed gene will depend on the intensity of the effort, especially in terms of number of BC1 and BC2 plants that are screened. Enough plants need to be screened not only to find a close recombination at each step, but also to have enough plants remaining for a sufficient background selection.

Background selection:

The background selection is focused on recovering as much as possible of the genome of the recurrent parent on the chromosomes not carrying the target trait (that particular chromosome is primarily handled as part of the foreground selection). The concept is to use a set of well-spaced markers that cover all those chromosomes. At each backcross generation, the plants preselected from the foreground selection step are genotyped for this array of markers and scored for their similarity to the genome of the recurrent parent. At each generation, the plants that have recovered the most of the recurrent parent are used for the next generation of backcrossing. Plants with more than 95% recovery of the recurrent parent's genome can be obtained by the BC2 or BC3 generation depending on the intensity of the work done.

Stacking of introgressed traits (gene pyramiding):

In many cases, the breeder's goal will not be to introgress a single trait but potentially to introgress several traits at the same time, possibly from different sources. Instead of trying to handle all those traits together in the backcrossing process, the best approach usually is to perform all those conversions into the same background individually in parallel, and then to intercross the final single conversions to combine the traits together. In that case, only MAS is needed at the end since the narrowing of the introgressed regions through foreground selection and the recovery of the recurrent parent through background selection have already been done for each individual trait.

Final considerations:

The quality of the final product in terms of size of the donor introgression and the extent of recovery of the recurrent parent genome will be highly dependent on the amount of resources devoted to the project. Obtaining the highest quality conversion may be too expensive for some programs due to the number of plants and datapoints needed. In most cases, the breeder will need to find the right balance between the resources available and his intended final product. Doing introgression work to feed into an ongoing breeding program (as for most nontransgenic trait introgression projects) does not necessitate a final MABC product of quality as high as for the conversion of an elite commercial maize inbred with a new transgene. Whatever the choice of the breeder, the use of MABC instead of conventional backcrossing will have a huge impact on the speed, quality and final cost of a trait introgression effort.