Status of Genome and QTL Mapping in Pigs-Data of Hohenheim

Three informative F2 families (by use of European Wild boar (W), Pietrain (P) and Meishan (M)), each with more than 300 animals, were genotyped for evenly spaced marker loci and recorded for more than 100 quantitative traits. Linkage and QTL mapping data for 8 chromosomes are presented (74, 76 and 75 mapped loci in families WxP, MxP and WxM, resp). Linkage maps gave evidence of heterogeneity in recombination between sexes and families. The male to female recombination ratio were 1.19 (WxP), 1.35 (MxP) and 1.27 (WxM). Several QTLs were mapped for performance traits of growth, carcass and meat quality. These QTLs are located on different chromosomes and influenced by families. Larger effects were found on chromosome 6 and 7, and e.g. up to 60 % of the phenotypic F2 variance for meat quality traits was associated with chromosome 6. Candidate genes are proposed for some of the QTL intervals. The subsequent QTL mapping use a combined strategy of genome-wide marker mapping with a positional candidate gene approach in order to identify genes which are significant for breeding.


Introduction
Main tasks of genome research in farm animals are orientated to identify genes affecting economically relevant traits.Values of those traits are often influenced by many, in detail unknown factors and thus quantitatively distributed in populations (quantitative traits).For those traits, LEUTHOLD (1972) and his co-workers used parameters of metabolism which are closely related to the regulation and expression of the genes involved.Another approach for genetic analysis of quantitative traits is the diagnosis of effects arising from chromosome regions.Within families the transfer of chromosome intervals can be traced by Polymorphie loci and subsequently used for the analysis of gene effects on quantitative traits (Quantitative Trait Loci, QTL;GELDERMANN, 1975).Among farm animal species this approach was first used in pigs by ANDERSSON et al. (1994).Its success is partly due to some advantages of this species for genetic analysis, e.g. the short generation interval, multiparity, low number of morphologically different chromosomes (GUSTAVSSON, 1988), simple and standardized housing of test animals, numerous and well characterized criteria of produetion traits and similarity to human physiology.Mapping of the porcine genome has been stimulated and supported by multinational programmes (e.g.PiGMaP -ARCHIBALD et al. (1991) which enabled construction ofa middle density linkage map and cytogenetic map needed for QTL mapping).At present these maps encompass 1774 loci, of which 507 are designated genes (Livestock Animal Genome Database -Roslin Institute, Edinburgh; stage 08/1998; http://www.ri.bbsrc.ac.ukY).The majority of mapped loci are microsatellites, i.e. short repetitive DNA motifs (e.g.LITT and LUTY, 1989;TAUTZ, 1989) which occur interspersed in non repetitive DNA and cover the whole genome with about 60.000 to 100.000 loci.
At Hohenheim, three informative porcine F 2 families have been generated for QTL analysis on growth, carcass and meat quality traits.Objectives of this programme are a detailed recording of quantitative traits, providing of data and samples for Joint research, efficient genotyping of DNA loci, genome-wide QTL mapping for a large number of traits as well as comparison of families for intrachromosomal recombination and QTL effects.In this report we give a short review on results so far obtained.

Material and Methods
Genetically diverse resources of pigs (European Wild Boar, Meishan and Pietrain) were used for the generation of three F 2 families (Tab.1).All pigs were kept in one experimental Station under standardized housing and feeding.At ten weeks of age, the piglets were taken into single pens for growth and fattening tests.The pigs were slaughtered at a defined age (210 days).Samples have been collected for measuring additional phenotypic criteria and isolation of genomic DNA.
The F 2 families were optimized for QTL analysis.Based on some marker loci, individuals were selected in the founder generation according to the expected degree of heterozygosity in the F, offspring.Then, F, animals with the highest information values for mapping were mated for the produetion of the F 2 generation.Only one to three males and a few females were used in the founder and F, generation in order to reduce the number of segregating haplotypes in the F 2 .More than 300 F 2 animals have been generated per family.Comparison groups were kept for animals from the founder and F, generation.All individuals were housed in one Station in order to have uniform environmental effects.
Table 2 gives a review of the traits included.The performance traits belong to stress reaction, fattening, carcass, and meat quality.Moreover, additional traits are included which are highly heritable, and some of them correspond to performance traits.Up to now, 8 chromosomes are mapped in all three pedigrees.Table 3 lists the numbers of markers considered for linkage and QTL mapping in the work presented.74, 76 and 75 loci were genotyped in the families WxP, MxP and WxM (KNORR, 1996;YUE, 1998;BEECKMANN, 1998).The Type II loci are microsatellites, the Type I loci include biochemical polymorphisms, blood groups, allotypes and DNA variants for potential candidate genes (e.g.CRC, GH).For example, the growth hormone (GH) gene variants were analyzed according to the RFLP method of LARSEN and NIELSEN (1993) by using the restriction enzymes Apal and HinPI (KNORR, 1996) for which restriction sites are in the promoter and signal peptide coding regions.Marker loci were chosen to be evenly distributed over all chromosomes.Especially microsatellite loci were selected so that they were variable between founder sources (informative in at least two families), cover the chromosomes with distances less than 20 cM, maximally different in their allelic fragment lengths (2 bp), and suitable for multiplex PCR (similar PCR conditions, different length of PCR products).The large scale genotyping of microsatellites was performed in an automated approach (Fig. 1).Genotype data were verified with help of a Computer programme by comparing the observed data with known alleles and with pedigree information (for details see YUE, 1998 andBEECKMANN, 1998).Linkage analyses were performed using the Software CRIMAP (GREEN et al., 1990) according to the guidelines in KEATS et al. (1991).Sex-averaged as well as sexspecific maps were constructed.Data were analyzed using the analytical methods and approach developed by HALEY et al. (1994).The analysis for each chromosome included background genetic effects on other linkage groups as suggested by ZENG (1994) andJANSEN (1993).Background genetic effects were included as covariates using a stepwise selection procedure.Chromosome-specific empirical threshold values of the F test statistic were estimated via permutation tests (CHURCHILL and DOERGE, 1994).Genome-wide thresholds were calculated from the data by applying a Bonferroni correction for the number of chromosomes not included in the current analysis.The 10%, 5% and 1% genome-wide thresholds were estimated as 7.3, 8.1, 10.0 in WxP; 7.3, 8.0, 9.8 in MxP and 7.6, 8.4 and 9.9 in WxM respectively!Associations between GH gene variants (combined alleles of two RFLPs) and trait values were analysed using the Statistical package LSMLMW of HARVEY (1987).

Results
Linkage Maps.The linkage data of the chromosomes so far mapped are shown in Table 3. Between 74 and 76 markers were typed on the 8 chromosomes with an average marker distance of about 18 cM.The total genetic length covered per family was between 1317 cM and 1345 cM.On average higher recombination frequencies were observed in females than in males.The male to female recombination ratio was 1.19 (WxP), 1.35 (MxP) and 1.27 (WxM).For chromosome 1 male recombination was higher in all three families.Large differences in the length for some linkage groups were mainly due to markers located at the end of chromosomes that were not informative in all families (e.g.chromosome 6).Map positions and effects of QTLs.The estimates for the most likely position and effects of QTLs detected at a highly significant level of linkage (P < 0.01; F ratio > 10.0) are given in Table 4.The F values for all investigated traits of different trait complexes are shown in Figure 2. First it can be seen, that QTLs differ in the size of their effects on traits ofthe same trait complex between pedigrees, Moreover, for the same traits QTLs were not mapped in all families.For example, major QTLs for fattening traits were only identified in WxP.For carcass traits highly significant QTLs were mapped in all three families on chromosomes 1 and 4, on chromosome 6 in WxP and MxP and on chromosome 7 in MxP and WxM.Smaller effects were found on chromosomes 3 (WxP; MxP), 8 (WxP; MxP) and 12 (MxP; WxM).QTLs for meat quality traits were located on chromosome 6 in WxP and MxP and on chromosomes 3 and 13 in WxM.In the families WxP and MxP, QTLs for carcass composition and meat quality were located in the region ofthe CRC gene, which had segregating alleles at position +1843 bp in these families.In WxM the CRC locus had a fixed allele (C) at position +1843 bp, but the microsatellita-ETH5001 within the CRC gene was informative.As given in Table 4, the gene action of the QTLs was largely additive.Surprisingly, for back fat traits, the alleles of the QTLs on chromosome 7 inherited from Meishan had allelic effects opposite of what would be expected based on the phenotype of this breed and were associated with a lower fatness ofthe carcass.The direction of the QTL effects on other chromosomes was in the expected direction.The locations of QTLs significant at least at the 10% genome-wide thresholds are shown in Figure 3, where the three pedigrees were visually merged on the map of one family.In general, QTLs were localized at few positions.The positions of QTL showed some similarities between families, as demonstrated e.g. for chromosomes 1, 4, 6, 7 and 8.
A. Fattening traits Associations between GH genotypes and performance traits.In the F 2 families, RFLPs of the GH gene were considered in an association analysis.Associations between GH gene variants and performance traits were detected in pedigrees MxP and WxP.The results given in Table 5 show that the GH locus explained 12% to 18% of the phenotypic variance in MxP and between 7% to 13% in WxM.In both pedigrees  4 and Fig. 3).Therefore, we presume the associations found are caused by variants within the GH gene and not by a linked QTL with fixed alternative alleles in the founder breeds.

Discussion
For comparison and confirmation, evaluation of QTL effects and positions require data from several families.By comparing the data of our three F 2 families with other results (ANDERSSON et al., 1994;ANDERSSON-EKLUND et al, 1998;ROTHSCHILD et al, 1995), data show strong influences of families on QTL profiles.However, some chromosomal regions, especially those with larger effects, carry similar QTLs in several families.In correspondence with other reports (ANDERSSON-EKLUND et al, 1998;KNOTT et al, 1998), we found QTLs affecting a distinct quantitative trait to be located on several chromosomes.The GH gene locus gives evidence that genotypes for potential candidate genes should be analysed by both, QTL mapping as well as association analysis.
The programme for a genome-wide QTL mapping is still under progress.In the frame of a cooperative programme*), marker loci are presently genotyped for all porcine chromosomes.Moreover, further additional phenotypic criteria are prepared to be included in the QTL analysis.After finishing the genome-wide QTL mapping it will follow a candidate gene approach which will make use of mapped QTLs for a preselection of potential candidate genes.Only those genes which are located in chromosomal intervals with major QTL effects are regarded.Potential candidate genes are screened from comparative mapping results, as well as according to their functions and tissue specific expression (EST from cDNA libraries).As far as possible, the potential candidate genes are genotyped by use ofDNA variants in functional sites.Finally, QTLs are mapped by including the markers already typed combined with the newly regarded potential candidate genes.For such a strategy the co-operative Programme can use a number of pre-conditions: many phenotypic criteria of fattening, carcass and meat quality traits, large number of loci already mapped in pig, information on a number of potential candidate genes for growth, carcass composition, muscle structure etc., homology in location of genes between mammalian species, knowledge of QTLs already identified in several F 2 families, application of additional DNA techniques for Screening and genotyping of potential candidate genes (e.g.muscle and fat tissue specific cDNA).QTL analysis and association studies in families are able to resolve effects of chromosomal intervals not much less than +5 cM.Once QTLs have been assigned to an interval between two markers, the next target will be to isolate and identify the  ERNST RITTER, Dummerstorf

Table 2
Quantitative traits measured for the F 2 animals (Erfaßte quantitative Merkmale an den F 2 -Tieren) MÜLLER et al. (1998)rait values are given in detail byMÜLLER et al. (1998)."'Not all are considered in the QTL mapping presented.

Table 4a Summary
of significant QTL effects for produetion traits (WxP) (Zusammenfassung der signifikanten QTL-Effekte für Leistungsmerkmale)Only up to 4 traits with F ratios £ 10 are given per chromosome.Additional traits with F ratios t 10 are given below.SSCl: Back fat depth at 13*/14* vertebra, back fat depth at back, back fat weight, bacon external fat weight, Shoulder meat weight, chops meat weight, bacon meat weight, bacon meat weight / half carcass weight, live weight at slaughter, carcass weight.SSC4: Chops meat weight, weight of liver.SSC6: pH 45 min M. long, dorsi, pH 45 min M. semimembranosus, conductivity 45 min M. long, dorsi, conductivity 45 min M. semimembranosus, conductivity 24 h M. semimembranosus, stiffness oSM.semiinenbranostts, meat colour, fat cuts, bacon meat weight / half carcass weight, bacon meat weight, Shoulder meat weight, bacon weight." The most likely positions are given in centimorgans from the proximal end of the chromosome. 11Estimates are given as mean ± SE. 11 The reduction ofthe residual variance in the F, generation by including a QTL at the most likely position.

The reduction ofthe residual variance in the Fi generation by including a QTL at the most likely position.
Only up to 4 traits with F ratios k 10 are given per chromosome.Addiüonal traits with ,F ratios S 10 are given below.SSC6: pH 45 min M. long, dorsi, pH 45 min M. semimembranosus.conductivity45 min M. long, dorsi, conductivity 45 min M semimembranosus, conducliv.ty24h M. semimembranosus.mealcolour, stiffness of Usemimenbranosus, back fat depth at I3°>/14» vertebra, average back fat depth, Shoulder external fat weight, fat cuts, bacon meal weight / lean cuts weight, dressing percentage, bacon weight/ half carcass weight,, mealifat ratio, bacon meat weight / half carcass weighL SSC7: Average back fat depth, abdominal fat weight, back fat depth at hip, back fat depth at 13 u '/14 u ' vertebraThe most likely positions are given in cenlimorgans from Ihe proximal end of the chromosome." Estimates are given ns mean ± SE. »

Table 4c
Summary of significant QTL effects for produetion traits (WxM) (Zusammenfassung der signifikanten QTL-Effekte für Leistungsmerkmale) Bacon meat weight, chops meat weight."The most likely positions are given in centtmorgans from the proximal end of Ihe chromosome.2} Estimates are given as mean ± SE. " The reduction ofthe residual variance in the F 2 generation by including a QTL at the most likely position.