The aim of this study was to confirm the existence of two QTL associated with BFT on SSC2p and to further dissect the imprinting effect of the IGF2 gene, in order to resolve contradictory results published for the French and Dutch MS × European pedigrees [5, 12]. The two designs were based on similar founder breeds (MS were crossed with LW and/or LR) and they contributed equally (considering the number of F2) to the so-called "combined pedigree". However, two major differences should be noted. First, the two pedigrees are reciprocal: the MS breed was used as the sire breed in the Dutch pedigree and as the dam breed in the French one. The second discrepancy lies in the pedigree structure: a limited number of large F1 sire families were produced in the French pedigree, whereas the Dutch pedigree consisted in many small F1 sire families.
Our objectives were to (1) show that another QTL on SSC2p is associated with BFT in addition to the IGF2 gene and (2) determine the most likely hypothesis to explain the discrepancies regarding observed imprinting effects in the two pedigrees. Two hypotheses were proposed: a) the imprinting of IGF2 gene is not complete and b) the mendelian effect detected at 0 cM (at the position of the IGF2 locus) in the French pedigree  is in fact an artefact due to genetic linkage between IGF2 and a second QTL.
Detection of QTL underlying BFT on the short arm of SSC2
Analysis of the combined pedigree with the single-QTL model gave a most likely position of the QTL at 0 cM, but LRT values over the 5% threshold were obtained for all positions in the first 50 cM of SSC2 (Figure 1). When the two pedigrees were analysed independently, two different most likely QTL positions were obtained, in accordance with previously published results on the two pedigrees, i.e. at 0 cM in the French pedigree [11, 17] and in an interval between positions 20 and 50 cM in the Dutch pedigree [22, 23]. After genotyping the Dutch pedigree for the IGF2-intron3-G3072A mutation, it was concluded that the IGF2 QTN, localised at 0 cM, explained most of the observed paternally expressed QTL for BFT on SSC2 . However, the presence of an additional QTL around 30 cM could not be excluded. In the present work, nearly every position on the short arm of SSC2 was significant, which is consistent with the large variability of positions found in the Dutch pedigree, and tends to confirm the hypothesis that more than one QTL associated with BFT is segregating in this region. However, this hypothesis had not been validated before for these two pedigrees.
With the combined pedigree, the multiple-QTL model gave no significant result, so that the hypothesis of two co-segregating QTL could not be validated. These results indicate that the F1 sires in general are not heterozygous for both loci and/or that the dataset does not provide enough information to validate the alternative hypothesis, which might be due to the proportion of homozygous sires for the IGF2 mutation in the combined pedigree (0.5).
The separate analyses of the A/G and G/G F1 fathers provided evidence for the segregation of two QTL, the IGF2 mutation and a second one most likely positioned at 44 cM, which is close to the position initially reported in the Dutch pedigree . Segregation of a QTL affecting BFT around 40 cM was also reported by Lee et al.  in a Wild Boar × MS pedigree in which all founders were A/A for the IGF2-intron3-G3072A mutation. Therefore, our results confirm the hypothesis of a second QTL segregating in the Dutch LW × MS cross, as suggested by Jungerius et al. , and extend this observation to the French cross.
Is the IGF2 gene only paternally expressed?
The IGF2 gene has been studied in detail in several species and, in most cases, paternal expression has been described. Nevertheless, modifications of the imprinted status of genes have been reported in humans (most often associated with diseases) . For the IGF2 gene, such pathological modifications have already been described [26, 27]. However, a study on the loss of imprinting of the IGF2 gene in colorectal cancers has also shown a loss of imprinting in normal mucosa and peripheral blood leukocytes . Moreover, Sakatani et al.  have reported maternal expression of IGF2 in a healthy human population. Li et al.  have demonstrated that IGF2 P1 transcripts are bi-allelically expressed in all studied organs from adult healthy pigs. In addition, several studies have reported that the imprinting pattern of a locus can be variable over ontogenetic time  or under different environmental effects .
One present objective was to evaluate whether the differences of the imprinting status previously reported for the French  and the Dutch  pedigrees could be clarified. Significant effects of the allele inherited from the dam at the IGF2 mutation were obtained when analysing the progeny of A/G mothers, which is not in accordance with exclusive paternal expression of the IGF2 gene. Simulations showed that segregation of a second QTL at 40 cM from IGF2 can lead to the false detection of expression of the maternally inherited IGF2-allele in the French pedigree, whereas this was almost never observed in simulations of the Dutch or the combined pedigree. The discrepancies between simulation results can be explained by differences in the pedigree structures. The Dutch pedigree is based on 104 matings between F0 animals that produced many small half- and full-sib families (24 F1 sires and 174 dams), whereas the French pedigree is based on six matings only between the F0 animals that produced six large half- and full-sib families (six sires and 20 dams). Another difference between the two designs was the allele frequencies at the IGF2-intron3-G3072A mutation: in the French design, 75% of the F1 dams were heterozygous at the mutation whereas only 47% dams where heterozygous in the Dutch design. Since the simulations were performed using the real genotype data at the IGF2-intron3-G3072A mutation, we could not estimate the influence of the percentage of heterogeneous dams. The relative impact of the different family designs and allele frequencies at the mutation could not be differentiated with these simulations. Nevertheless, the ANOVA and simulation analyses suggest that the differences in the effect of the maternal allele at the IGF2 locus can be caused by the segregation of an additional QTL at 44 cM. The genetic linkage between IGF2 and a second QTL with an effect of 0.32 s.d. is high enough to create this artefactual maternal effect, even if the two loci are relatively distant.