To map the QTL on SSC7, we used the allele type in the QTL (Q or wt), not the phenotype (vertebral number: 20, 21, 22, or 23). By using half-sib analysis with approximately 1100 progeny animals, we defined the 30 alleles of 15 parent animals. The amount of data seems small, but it is worthwhile because it is qualitative, not quantitative. These 15 animals were from a closed breeding population, the AY population. This population was derived from 10 sires and 65 dams as founder individuals and was bred for seven generations (1987 to 1993), in which average daily gain and backfat thickness were major objectives of breeding but not the number of vertebrae. After 1993, the population was maintained with 10 sires and 35 dams. We think that a large degree of recombination was accumulated in this population, making the population favorable for IBD analysis. On the other hand, unrelated individuals were used as the founders, and we expected that the Q alleles of some of these pigs would have common sequences in a limited region near the QTL, enabling high resolution for fine mapping similar to those used in association studies.
We started fine mapping of the QTL over a region of approximately 9 Mb to give a 95% confidence interval in the QTL analysis. In the first step of haplotype analysis for the 19 Q and 11 wt alleles (Figure 2), the candidate region was fortunately narrowed to approximately 300 kb (between SJ7088 and SJ7040), but a conserved region contained only two markers in the 19 Q alleles. The second step of the analysis, using densely located markers, revealed the conserved region with seven markers (no more than 200 kb) in the Q alleles of the AY population.
In our previous study , we characterized the QTL alleles of parental pigs in the F2 families, and we used these animals for further analysis. The 14 Q alleles of these pigs (five Landrace, two Berkshire, two Duroc and five Large White) presented a conserved haplotype from SJ7088 to SJ7114. As a final mapping result, the candidate region for the QTL was an approximately 41-kb region between SJ7126 and SJ7099, which included the common haplotype within all the Q alleles in the present study.
The result of the fine mapping was confirmed by linkage disequilibrium analysis and an association study in independent meat animals produced with Duroc sires and F1 (Landrace × Large White) dams. The 41-kb region was included in a haplotype block, and the SNPs developed in this region were highly associated with the number of vertebrae. We therefore concluded that the QTL was located in the 41-kb region.
In the 41-kb region, transcripts were detected in pigs as EST in the embryonic stages. They encoded parts of a putative uncharacterized protein, which is not a member of any other known gene family. We cloned the cDNA covering the putative open reading frame from the swine embryos and we named it vertnin (VRTN) as the gene responsible for the QTL for vertebral number in pigs. A domain search of the swine homolog (C14orf115) in Ensemble (http://uswest.ensembl.org/index.html) revealed that it had a similar motif to the helix-turn-helix domain of Transposase IS3/IS911 . Although this domain is reported to be unique to bacteria, vertnin is expected to have DNA-binding activities. In our preliminary study the green fluorescent protein (GFP)-fused vertnin protein was expressed in the nuclei of cultured cells (Additional file 8 Figure S4). The QTL on SSC7 in this study affected the number of thoracic vertebrae but not the number of lumbar vertebrae, whereas the QTL on SSC1 affected the numbers of both thoracic and lumbar vertebrae . This evidence suggested that the number-increase allele of swine VRTN added an additional thoracic segment in the animal. We suspect that the expression pattern of Hox genes (e.g. Hoxa-9 or Hoxc-9, which are expressed in the terminal region of the thoracic segment and more posterior portion of mouse embryos; [17, 18]) could be altered. A series of genome sequencing projects (funded by National Human Genome Research Institute; http://genome.gov/) suggested that orthologs exist not only in other mammals but also in birds and fish, and encode conserved proteins (Additional file 9 Figure S5). Vertnin is therefore likely to be an essential factor for development of the embryo in a wide range of organisms. Its functional analysis may provide novel findings in the areas of developmental biology (e.g., somitogenesis or morphogenesis), so we are planning to perform a functional analysis with model organisms such as the mouse or chicken.
Polymorphism analysis of the 41-kb region showed that the AY population had only two haplotypes (Q and q), which consisted of 42 polymorphic sites. Another haplotype (q') was detected in the wt allele of a Landrace sow, in which the q' allele had no significant effect (calculated as 0.05) and the other allele (Q) increased 0.55 of vertebral number . In our preliminary study it was also found in the sequence of the BAC clone L261J7  (Additional file 2 Table S1). There were nine candidate polymorphic sites, and they were highly associated with each other in European commercial-breed pigs. We cannot define the causative polymorphic sites genetically, but these nine sites were associated with the number of vertebrae in commercial-breed pigs, and this information will be useful for genetic diagnosis in breeding populations.
In the swine VRTN gene, we found one nonsynonymous substitution, NV064 (G→A at nucleotide 1247 of AB550854; Gly365Asp), but it was excluded from the candidate sites because the same nucleotide G was located in the Q and q' alleles (Table 3). We detected a change in the transcription of VRTN with changes in embryonic stage. It is possible that this change in expression with stage is the origin of the QTL. Among the polymorphisms of the swine VRTN gene, an insertion of a PRE1 element (one of the swine SINE elements) into the intron of the Q allele, along with the SNPs in the promoter region, may be a cause of the changes in expression of this allele with embryonic stage. In mouse organogenesis, expression of some genes is regulated by a mechanism that includes the activation of a SINE B2 repeat . In this case, the SINE B2 acts as a boundary between the heterochromatin and euchromatin, partly by the activation of pol III and pol II RNA polymerase. In another case, SINE elements have been reported to act as enhancer elements, as is the case with the FGF8 gene . In the swine VRTN gene, transcripts around the PRE1 element were detected on the Q allele by RT-PCR in our preliminary study (data not shown). To further explore the causative mutation as well as the biological changes induced by it, we are using molecular biological techniques to study the regulation of swine VRTN expression.