Recent developments in molecular biology and statistics have opened the possibility of identifying and using genomic variation and major genes for the genetic improvement of livestock. Information concerning these techniques and their application to the genetic improvement of animals is reviewed. Application of marker assisted selection can be more effective in characteristics that are expressed late in the life of the animal, or controlled by a few genes. The first example corresponds to the longevity and carcass characteristics in meat producing animals, the second, to the resistance to certain diseases or defects of simple inheritance. The detection of major genes using mixture models with segregation analysis can direct the work of identification of DNA marker genotypes towards populations and characteristics with greater probability of detecting a QTL. The present trend indicates that molecular, pedigree and phenotypic information will be integrated in the future through mixture models of segregation analysis that might contain QTL effects through the markers, polygenic inheritance and uses powerful and flexible methods of statistical estimation such as Gibbs Sampling.

Introduction

During the last five decades, the application of methods based on population genetics and statistics allowed the development of animals with a high productive efficiency. These systems are based on simplified models of genic action that assumes a large number of or genes with small individual effects in the expression of the phenotype (polygenes) and emphasises the average genic effects (additive effects) over their interactions. These procedures allow obtaining breeding values of the animals using phenotypic and genealogical information. Properties of the predictions are equivalent to the levels of correlated random effects of a mixed linear model or best linear unbiased predictors (BLUP) which is based to a large extent on the work of C. Henderson (Henderson, 1984).

Important advances to some of the economically important characters in several species of livestock has been achieved based on phenotypic performance, however, several limitations of these methods of improvement based on population genetics alone are becoming evident with time. Their efficiency decreases when the characteristics are difficult to measure or have a low heritability. Additionally, selection has been generally limited to those characteristics that can properly be measured in large numbers of animals. Some characteristics such as the rate of survival are expressed very late in the life to serve as useful criteria of selection. Also, the traditional selection within populations has not been very efficient when the selection objective involves several characteristics with unfavourable genetic correlation, for example, milk production and protein content of milk (Schwerin et al., 1995).
The use of molecular techniques could help to solve some of the limitations of the current methods. The capacity to generate dense genetic maps in each species can in principle allow the complete genome to be evaluated for quantitative trait loci (QTL) with a major effect on the phenotype. This information can then be used in genetic programs (Kinghorn et al.,1994). Also, methods of segregation analysis have been developed to detect the presence of major genes from the analysis of pedigreed data in absence of molecular information (Bovenhuis et al., 1997).

The objective of this paper is to review some advances on molecular and statistical methodology to identify and estimate the effects of major gene effects in animal populations, and its potential use in the genetic improvement of livestock.

Genetic markers

Sax (1923) showed that an observable gene with simple Mendelian inheritance could act as a marker for the segregation of a gene involved in the expression of a quantitative trait. Early work on detection of genome variation of potential markers focused on the analysis of proteins and blood type variation, but this was found to be impractical to be used as genome markers. Protein systems lack adequate polymorphism, genome coverage was low and there was a requirement for the gene to be expressed at
the phenotypic level to make detection possible (Drinkwater and Hetzel, 1991).

RFLP. Beginning in 1980, an accelerated understanding of the structure and function of the genome has been obtained as animals (Schimenti, 1998). At first, techniques were developed to visualise the differences at the level of the structure of the DNA based on the use of bacterial restriction enzymes that cut the DNA at sites with specific nucleotide sequences. With this basis, the technique denominated restriction fragment length polymorphisms (RFLPs) was developed. The identification of RFLPs requires the use of gel electrophoresis to separates the DNA fragments of differing sizes followed by transfer of the fragments to a nylon membrane (Southern blot) and visualisation of specific DNA sequences using radioactive or chemiluminescent probes exposed to an X-ray film (Drinkwater and Hetzel, 1991).

Microsatellites. Microsatellites systems are composed of DNA repeats in tandem at each locus. The tandem repeats are usually simple dinucleotides (such as (TG)_n with each dinucleotide repeated about ten times. Its high degree of polymorphism in the number of repeats (n) allow its use as location markers in genome mapping. The length of each allele is determined by Polymerase Chain Reaction (PCR) using specific oligonucleotides primers flanking the repeat sequence. The DNA products are visualised after electrophoresis. PCR based Microsatellites techniques facilitated the construction of genome maps in most livestock species because its abundance in the genome, the specificity of the primers, its high degree of polymorphism which several alleles and their easy detection (Albert et al., 1994; Lewin, 1994; Bishop et al., 1995; Smith and Smith, 1993).

Direct v/s indirect markers

There are two main categories of genomic information that can be used in genetic improvement of livestock:

1. Genes with known on the expression of certain protein (direct markers)
2. Genes with effects detected on the characteristic in statistical terms. These markers are based on polymorphic sequences of the DNA that is in the same chromosome (linked) to a gene of quantitative effect (QTL). (indirect markers)
   

The second type (2), corresponds to genes linked to the QTL with detectable variation by means of RFLPs, microsatellites or other similar molecular systems. Several studies have demonstrated relationships between molecular variants and the phenotypic expression in several characteristics in several animal species (Beever et al., 1990; Andersson et al., 1994; Georges et al., 1995; Haley, 1995; Ashwell et al., 1997). This has stimulated the idea to add the genomic to the phenotypic information to increase or speed up the selection response to the "traditional methods" in which it is known as marker-assisted selection or MAS.

Marker assisted selection (MAS)

It is currently possible to incorporate this information in the present systems of selection with BLUP, that is to say, to add the information corresponding to direct or indirect markers to the systems of equations of the mixed model (Goddard, 1992; Kinghorn and Clarke, 1997).

The identification of direct markers avoids the ambiguity caused by the possibility of recombination between the marker and the QTL. In the later case we should strictly define this kind of selection as QTL assisted selection. For simplicity we treat this option as a special case of MAS without recombination among the marker and the QTL. Current marker use in MAS with the availability of relatively dense genome maps and multiple microsatellite markers available, will be more probably groups of allelic configurations of haplotypes, usually flanking the QTL, than a single marker. These haplotypes could in some cases have very low recombination rates with the QTL.

Several studies of simulation have evaluated the consequences of MAS in populations with a QTL segregating by comparing the use of a purely polygenic animal-BLUP model with a mixed model that incorporates both the polygenic effects and the haplotype identities. An extreme case is the result of Meuwissen and Goddard (1996). They found possible increases until of 30 until a 64% in the genetic response to selection of different characteristics in the first five generations from selection when the QTL explains a 33% of the genetic variation in the base population. However, Ruane and Colleau (1995) found possible increases in selection response of only 0.2 to 1% in six generations using a single marker. This wide variation suggests to maintain an attitude of caution in absence of greater experimental evidence and a greater amount of simulation studies covering more specific situations, that allow to reach certain consensus on the value of the MAS in diverse characters and species.
Meuwissen and Goddard (1996) concluded the following from their study on the use of the MAS:

1) MAS only can increase the rate of genetic gain in the long term when there is a continuous advantage of new identified QTL (obviously this has a biological limit determined by the Maximum proportion of the genetic variance that can be explained by the segregation of QTL).

2) The extra genetic gain due to the MAS decreases very quickly with the number of generations of selection for a same QTL. The rate of identification of new QTL is difficult to predict.

3) The gain due to MAS for a certain QTL is higher when the characteristic is measured after the selection, as it happens with the fertility and carcass characteristics.

Detection of major genes without markers

In the last ten years statistical methodologies of detection of major genes based on pedigree and phenotypic information on populations have been developed for animal populations. These methods are based on the use of mixed models and segregation analysis to fit the data to a mixture genetic model that includes in addition to the polygenic effects, the effects of a biallelic major gene. Calculation is performed in two stages; firstly genotype probabilities are obtained, then major gene, fixed effects and polygenic effects are fitted and used to recalculate new parameters by regressing phenotypes on estimated probabilities. Calculation is iterated upon convergence (Kinghorn et al., 1993). Segregation analysis allows inferring the unknown genotypes from the probabilities of transmission of the gene given the phenotype of the individual and their relatives.

In mixture models, regression and Gibbs sampling estimation approaches have been implemented to obtain estimates of the major gene effects and allelic frequency (Bovenhuis et al., 1997).

These methods that make use of information currently available in many animal populations, are an option for a preliminary screening for major genes that can contribute to rationalise the use of expensive QTL-marker linkage estimation experiments.

Discussion

A fundamental aspect for the efficient use of the markers for the MAS resides in the correct detection of the location of the QTL in the chromosomes, as well as of the magnitude of its allelic effect and its allelic frequency. The accumulation of studies within the same chromosomal region can help to distinguish between the cases of real or false QTL detection. Meuwissen and Goddard (1996) demonstrated that the incorporation of a false QTL to MAS, could reduce the genetic gain until a 14% in comparison with a usual polygenic BLUP model.

Most studies so far have studied the effect of MAS using rather simplified assumptions and a single trait affected by one QTL and polygenes. Studies using more realistic models such as multiple estimated QTL effects and multiple trait selection could help to make better decisions regarding the use of MAS in animal improvement.

While molecular techniques offer an important series of possibilities for the genetic improvement of livestock, the materialisation of these expectations requires of the solution of a number important of technical problems to take advantage of all the information available an efficient way, to reduce the costs of generating genomic information and of obtaining reliable estimations of the effects of the QTL and the application of the MAS and the genomic information in general for the improvement animal. A common problem related with QTL estimates is inconsistency, which means that a QTL effect is not expressed similarly in several years, or when its used in a different population (Mayo and Franklin, 1998).

Many epistatic effects have been found associated to QTL effects. In order to use molecular information in selection, reliable QTL effects should be incorporated in the models of analysis from population genetics in a way, which is consistent with observed variation. Evidences of widespread epistasis affecting QTL effects may restrict use and prediction of QTL effects outside the populations used for detection. This is particularly important when breed crosses instead of sire families in outbreed populations are used as design to detecting the QTL effects (Mayo and Franklin, 1998).

Conclusions

A rational use of the molecular methodologies requires the simultaneous optimisation of selection on all the genes affecting important traits in the population. The maximum benefit can be obtained when these techniques are used integrated with reproductive technologies like the artificial insemination, and collection and production in vitro of embryos to accelerate the genetic change (Bishop et al., 1995; Montaldo, 1993). There is a danger associated with a potentially inadequate use of QTL information giving an excessively high emphasis to simple molecular information in detriment of the overall economic gain through all traits and their polygenic effects in the population. Dissemination of the information to the industry is therefore a complex issue concerning QTL effects and molecular markers.

The characteristics on which the application of the MAS can be effective are those that are expressed late in the life of the animal, or that are controlled by a few pairs of alleles. The first example corresponds to the longevity and carcass characteristics in meat producing animals, the second, to the resistance to certain diseases or defects of simple inheritance. The use of MAS could be justified, by virtue of its high cost, in animal nuclei that allow dilution of the costs when germplasm of the nucleus is extensively used towards the commercial population. Also in those characteristics in which the procedures of conventional selection have reached their limits in efficiency or the results have been not satisfactory.

The main advantage of including molecular information over pure segregation analysis, is the possibility of evaluating the simultaneous effect of several QTL on the characteristics of economic importance, and in the future increasing its precision and the complexity of the involved models of genic action, for example QTL with multiple alleles. The present trend indicates that molecular, pedigree and phenotypic information will be integrated in the future through mixture models of segregation analysis that might contain QTL effects through the markers, polygenic inheritance and uses powerful and flexible methods of estimation such as Gibbs Sampling.

Before the molecular information on the QTL which control the characteristics of economic interest is generated, the detection of major genes using segregation analysis could direct the work of identification of genotypes towards populations and characteristics with greater probability of detecting a QTL using molecular markers.

Many questions remains on the nature and action of the QTL involved on the variation of complex traits and about the nature and definition of QTL effects. The use of molecular techniques involves new opportunities and new challenges for building and using more predictive and efficient statistical models for livestock improvement.

Acknowledgements

The authors wish to thank Professor Brian Kinghorn and Dr. Julius van Der Werf from the University of New England at Armidale, Australia for inspiring discussions.

References

Albert, B., Bray, D., Lewis, J., Raff, M., Roberts, K. and Watson, J. D. (1994). Molecular biology of the cell. Garland Publishing, Inc.

Andersson, L., Haley, C.S., Ellegren, H., Knott, S.A., Johansson, M., Andersson, K., Andersson-Eklund, L., Edfors-Lilja, I., Fredholm, M., Hansson, I., Hakansson, J., and Lundstrom, K. (1994). Genetic mapping of quantitative trait loci for growth and fatness in the pig. Science 263:1771.

Ashwell, M. S., Rexroad Jr., C.E., Miller, R.H., Van Raden, P. M. and Da, Y. (1997). Detection of loci affecting milk production and health traits in an elite US Holstein population using microsatellite markers. Animal Genetics 28:216-222.

Beever, J. E., George, P. D., Fernando, R. L., Stormont, C. J. and Lewin, H. A. (1990). Associations between genetic markers and growth and carcass traits in a parental halfsib family of Angus cattle. Journal of Animal Science 68:337.

Bishop, M. D., Hawkins, G. A. and Keeler, C. L. (1995). Use of DNA markers in animal selection Theriogenology 43: 61-70.

Bovenhuis, H., van Arendonk, J. A. M., Davis, G., Elsen, J.-M., Haley, C. S., Hill, W. G., Baret, P. V., Hetzel, D. J. S. and Nicholas, F. W. (1997). Detection and mapping of quantitative trait loci in farm animals. Livestock Production Science 52:135-144.

Drinkwater, R. D. and Hetzel, D. J. S. (1991). Application of molecular biology to understanding genotype-environment interactions in livestock production. In: Proc. of an International Symposium on Nuclear Techniques in Animal Production and Health. IAEA, FAO. Vienna, 15-19 April, 1991:437-452.

Georges, M., Nielsen, D., Mackinon, M., Mishra, A., Okimoto, R., Pasquino, A. T., Seargeant, L. S. , Sorensen, A, Steele, M. R., Zhao, X., Womack, J. E. and Hoeschele, I. (1995). Mapping quantitative trait loci controlling milk production in dairy cattle by exploiting progeny testing. Genetics 139:907-920.

Goddard, M. E. (1992). A mixed model for analyses of data on multiple genetic markers. Theoretical and Applied Genetics 83:878-886.

Haley, C. S. (1995). Livestock QTL - bringing home the bacon? Trends in Genetics 11:488-492.

Henderson, C. R. (1984). Applications of linear models in animal breeding. University of Guelph, Guelp, Canada.

Kinghorn, B. P., Kennedy, B. W. and Smith, C. (1993). A method for screening for genes of major effect. Genetics 134:351-360.

Kinghorn, B. P., van Arendonk, J. A. M. and Hetzel, J. (1994). Detection and use of major genes in animal breeding. AgBiotech News and Information 6:297N-302N.

Kinghorn, B. P., and Clarke, B. E. (1997). Genetic evaluation at individual QTL. Animal Biotechnology 8:63-68.

Lewin, B. (1994). Genes V. Oxford University Press.

Mayo, O. (1996). The application of QTL methodology to Merino breeding. Wool Technology and Sheep Breeding 44:281-289.

Mayo, O. and Franklin, I. R. (1998). The place of QTL in the basis of quantitative genetics. I. General considerations. In: Proc. of the 6^th Word Congress on Genetics Applied to Livestock Production, Armidale, and NSW 26:77-80.

Meuwissen, T. H. E., and Goddard, M. E. (1996). The use of marker-haplotypes in animal breeding schemes. Genetics Selection Evolution 28:161-176.

Montaldo, H. H. (1993). Biotecnología y mejoramiento genético animal en México (Biotechnology and genetic improvement of animals in Mexico). In: (M. Arenas, L. F. Bojalil y L. Hernández Comp. Las profesiones en México: Agronomía, Medicina Veterinaria y Zootecnia, Universidad Autónoma Metropolitana-Xochimilco, Universidad de Colima, México, D. F. pp 90-108).

Ruane, J. and Colleau, J. J. (1995). Marker assisted selection for genetic improvement of animal populations when a single QTL is marked. Genetical Research 66:71-83.

Sax, K. (1923). The association of size differences with seed-coat pattern and pigmentation in Phaseolus vulgaris. Genetics 8:522-560.

Schwerin, M., Brockmann, G., Vanselow, J., and Seyfert, H. M. (1995). Perspectives of molecular genome analysis in livestock improvement. Archiv fur Tiernzucht-Archives of Animal Breeding Dummerstorf 38:21-31.

Smith, C., and Smith, D. B. (1993). The need for close linkages in marked-assisted selection for economic merit in livestock. Animal Breeding Abstracts 61:197-204.

Schimenti, J. (1998). Global analysis of gene function in mammals: integration of physical, mutational and expression strategies. EJB Electronic Journal of Biotechnology at http://www.ejbiotechnology.info/content/vol1/issue1/full/2/.