The first purpose of this study is to fine map the SSCX QTL for fatness and muscling traits. From the primary scan in the INRA population, the CI of the QTL for BFT and HW were approximately 15 cM wide . In this study, they were dramatically reduced to merely 7 cM (Table 1) through joint analysis of both INRA and JXAU populations with new information of high-density markers commonly genotyped for them. However, the current QTL intervals still contain a large recombination coldspot which spans from SW259 to UMNP1218 (~36.4 Mb vs. 0 cM), therefore further refinement of these QTL by linkage analysis seems impossible.
Fortunately, the distribution of the chromosomes in the INRA population permitted to study two groups of related F1 dams. Despite unbalanced number of progeny in the two groups, we conclude that they have contrasted genotypes for the QTL. Indeed, the power estimation for the detection of the interaction between the QTL effect and the groups indicates that it is highly unlikely that the QTL is not detected in “homozygous dams” families due to a lack of power in this group. From this conclusion, the haplotype analysis of these two groups with different QTL genotypes enabled us to estimate the most likely position of the QTL. The full sisters 910002 and 910013 carried the same LW chromosome X but different MS X chromosomes bearing different QTL alleles. Because 910013 is homozygous for the SSCX QTL, its MS and LW chromosomes X share a similar QTL allele “q” associated with low BFT and high HW traits. In contrast, the heterozygote 910002’s MS chromosome X contains another QTL allele “Q”. In the coldspot region between MCSE3F14 and UMNP1218 (about 34.5 Mb in length), 910013 carries a haplotype defined as Hap1, associated thus to a “q” allele, whereas 910002 carries a Hap2 haplotype associated with a “Q” allele. Noticeably, Hap1 and Hap2 MS haplotypes over most part of the coldspot are quite similar except for alleles at a microsatellite locus MCSE58H4 (Figure 2). Interestingly, 20 out of 23 F1 sows of the French families inherited the MS Hap1 (“q” allele) and the BFT QTL explained about 40% of phenotypic variance in the whole INRA F2 male population . Even if the limited numbers of their offspring of most F1 sows do not allow individually determining their QTL genotype, a large part of F1 sows are thus supposed to be heterozygous for the QTL, i.e. having a “Q” MS allele, despite harboring a Hap1 haplotype. Globally, this region of coldspot of recombination is thus extremely poorly polymorphic, and the two closely related haplotypes do not co-segregate with the QTL haplotypes. It is thus very likely that the causal mutation affecting BFT traits is located outside of the region MCSE3F14-UMNP1218 corresponding to the coldspot of recombination.
In addition, another IBS haplotype spanning SWR1861-SW259 interval ahead of the coldspot was also found on the MS chromosomes X of the segregating (910002) and non-segregating (910097 and 910018) sows. If the QTL was located in this interval, 910097 and 910018 sows would share the same “Q” allele on their MS chromosomes as the 910002. As 910097 and 910018 are homozygous at the QTL, their LW chromosomes should then harbor the “Q” allele, which is not likely. Thus, we can conclude from the haplotype analysis that it is not so likely to have the QTL just at the upper boundary of the coldspot (in blue on Figure 2). However, we cannot exclude that the QTL can be farer on the upper left area, where no common haplotypes with the used microsatellites was seen between individuals from the groups A and B.
In the region SW1426-UMNP93, two different MS haplotypes were observed between the full sisters 910002 and 910013 who had different MS QTL alleles, as well as between their belonged groups (A and B). More importantly, we previously evidenced that the three related F1 dams (910013, 910097 and 910018) inherited the same MS haplotype over the region from their recent ancestor; that is, their MS haplotypes are identical by descent (IBD) rather than mere IBS to each other . Similarly, the MS haplotypes carried by the full sisters 910002, 91009 and 910010 are also IBD. Because of the perfect match between allelic and IBD-haplotypic distribution among these MS chromosomes, we believe that the region downstream the coldspot (in yellow on Figure 2) is most likely to contain the QTL, which is also in agreement with the likely location of the QTL detected in the INRA population (Figure 1).
Following the study of Pérez-Enciso et al. , we managed to find some clues for supporting either the hypothesis of single pleiotropic QTL or the hypothesis of multiple linked QTL for the investigated traits. However, in our study the only linked QTL test significant was for IMF, suggesting that two positions separated by only 5 cM have joint and opposite effects on the trait. Despite the presence of two highly informative markers between these positions, the power to discriminate between these two positions is not high in our study due to limited number of recombination events. This result needs confirmation to validate that it is not an artifact . Čepica et al. reported a genome-wide significant QTL for CW that co-localized with QTL for BFT at the centromeric region of SSCX in a Wild Boar × MS cross. However, QTL for CW detected in the JXAU F2 males only reached a suggestive level and was mapped at 0 cM, far away from QTL for other traits. Moreover, there was an absence of QTL for CW in the INRA F2 males. These results indicate that the present QTL for fatness or HW has probably negligible effect on CW.
Pérez-Enciso et al. reported that at least two distinct regions segregate on SSCX in different populations, one in the neighborhood of SW259/SW1994 markers, with an effect on ham weight and carcass length, and another one between markers SW2476 and SW1943, with primary effects on fatness and shoulder weight. It must be noted that the marker order SW259/SW1994 (74.4 cM) - SW2476 (77.6 cM) - SW1943 (87.4 cM) on the USDA-MARC porcine genetic map was inconsistent with the order SW2476 - SW259-SW1994 - SW1943 on RH map and pig clone map [24, 27]. The present result of QTL analyses in the INRA F2 males showed that the HW QTL peak was located only 5 cM upstream of the BFT QTL peak (Table 1), which is in agreement with Pérez-Enciso et al. and Čepica et al.. Even so, we couldn’t discriminate the HW QTL from the BFT QTL because their CI overlapped. Indeed, as shown in Figure 1, the peak in the test statistics curve for the BFT QTL was much broader than that for the HW QTL detected in the INRA F2 males, and the latter was within the former. Moreover, the QTL for HW and BFT found in the JXAU F2 males were located at the same position (71 cM), very close to the location (74 cM) of the HW QTL detected in the INRA F2 males. Thus, although CI overlapped and tests for 2 linked QTL were not significant, our results, on one hand, are in agreement with the previous suggestions that two QTL exist: one proximal to UMNP1174 and another proximal to SW1426 (Figure 1), and on the other hand, they indicate the former QTL influence both BFT and HW rather than only HW.
It is interesting to compare the sizes of QTL effects on the same trait between the two populations. We found that the SSCX QTL for average BFT could explain 5.7% of phenotypic variation in the JXAU F2 population, which is lower than 35.9% and 6.2% of those explained by QTL mapped on SSC7 and SSC4, respectively . In contrast, the SSCX QTL detected in the INRA population showed markedly stronger effect on BFT than the SSC4 and SSC7 QTL . Nevertheless, high significance and similar map location of the BFT QTL on SSCX were found in the two populations, suggesting the existence of common QTL between them. As to the QTL for HW and LEA on SSCX, their significance levels differed largely between populations (Table 1). Given the same QTL shared by the two populations, these discrepancies were probably due to the differences between them in epistatic QTL , QTL allele segregation pattern of the founder breeds, environment effect and/or trait measurements. Despite these discrepancies, we detected no interaction between QTL effects and population, and the estimates of substitution allele effects in the two populations were close (Table 1).
The results of the QTL analysis and haplotype analysis in two combined F1 dam families indicate QTL segregation within the MS breed rather than the LW breed, which is expected since the LW instead of the MS has been selected for lean growth over decades as a commercial line. The Chinese MS pig with excess body fat is often used in breeding programs in order to take advantage of its prolificacy, while during the process, how to avoid the disadvantages of its excessive fatness and low growth rate have to be considered. The segregation of SSCX QTL for BFT and HW within Meishan breed provides an opportunity for us to make effective use of Meishan chromosome X in crossbreeding and to increase the frequency of the favorable alleles in the purebred by marker assisted selection.
ACSL4 is located at 80.5 cM, in close proximity to the most likely position of QTL for BFT identified in the INRA population. This gene showed consistent and multiple significant associations at the single SNP (data not shown) and haplotype levels in the two populations (Table 4). However it is obvious that the two ACSL4 SNPs are not the causal mutation(s) because the segregating and non-segregating F1 sows had the same SNP genotype. But their haplotypes should be linked with the causal mutation(s) because the QTL disappear when accounting for the gene haplotypes. Therefore, we cannot preclude that a polymorphism in ACSL4 mRNA sequence or its cis-acting elements may result in the QTL effects. Mercadé et al. sequenced most of the region of the ACSL4 mRNA in multiple breeds, and identified 10 polymorphisms within the 3’-UTR region, all of which formed only two haplotypes. Further, they found that the haplotype 1 (DQ144454:g.2274A-2645G-2782G-2933delete-2934delete-3272-C-3590G-3591T-3862T-4074A) fixed in the MS breed was at high frequency (0.95) in the LW breed. As the MS and LW pigs were used as founders in the INRA population, these polymorphisms or haplotypes could not be responsible for our observed effects on BFT.
The candidate genes IRS4 and SERPINA7 are also within our refined QTL region. Previous studies [16–18] reported significant associations between the IRS4 SNPs (FN424076:g.96C > G and FN424076:g.1829T > C) and BFT, as well as between a missense mutation p.245N > H in the SERPINA7 gene and BFT. In this study, the SNP FN424076:g.96C > G can be firstly excluded as a causal mutation, because its “C” allele was the major allele in both the MS and LW founders that likely carry different QTL alleles. Furthermore, the other SNPs, like the two SNPs in ACSL4, were not co-segregating with QTL alleles between the combined families A and B, so they are unlikely to be causal mutation either.
Despite the lack of supporting evidence for the polymorphisms in the three candidate genes underlying the target QTL, further research is needed to identify their potential variations in DNA sequence (e.g. copy number variation), DNA methylation and gene expression levels.