Vkorc1 genetic polymorphisms in mouse priority strains
We identified 10 haplotypes that were defined by 84 SNPs and 153 indels [see Additional file 1]. Six haplotypes were identified when the coding region was considered (Figure 1b). Five non-synonymous SNPs altering five amino acids (R12W, A21T, A26S, A48T and R61L) and four synonymous SNPs were identified (Figure 1b). The C132C polymorphism unique to M. m. musculus mapped onto the evolutionary conserved redox center C132-X-X-C135 motif of the Vkorc1 [24, 25].
Consideration of SNPs and indels in the 5' region increased the number of haplotypes to nine (Figure 1b), and inclusion of intronic polymorphisms increased the number of haplotypes to ten. All M. m. domesticus strains were identical in the coding region but genetic polymorphisms in the 5' region separated M. m. domesticus into two groups (Figure 1b). However, our comprehensive survey uncovered only two previously known Vkorc1 haplotypes (1 and 7) and two novel haplotypes (2 and 8) in M. m. domesticus. Considering the coding and 5' regions, only synonymous SNPs and 5' polymorphisms distinguished M. m. domesticus and M. m. musculus. A single non-synonymous SNP distinguished M. m. domesticus and M. m. musculus from M. m. molossiunus and M. m. castaneus. Four non-synonymous SNPs and at least eight 5' variants were unique to M. m. spretus.
Phylogenetic analysis of Vkorc1 haplotypes
The analysis of Vkorc1 haplotypes using the NJ method resulted in four main clades that were supported by bootstrap values of 93% or higher (Figure 2a). All 30 inbred strains and one wild-derived M. m. domesticus strain fell into clade I. One inbred and one wild-derived M. m. domesticus strain together formed clade II. Wild-derived M. m. musculus strains formed clade III and wild-derived M. m. molossinus and M. m. castaneus strains formed clade IV. The overall topology of the Vkorc1 haplotype tree is consistent with those obtained during previous phylogenetic studies of mice [19, 26]. It was also supported by bootstrap values of 89% or higher when MP was used to analyze the Vkorc1 haplotype data.
Relationship between Vkorc1 polymorphisms and prothrombin time
Variation in PT within and between clades I-IV was substantial. However, variation in PT within M. m. domesticus (clades I and II) overlapped with PT variation in the other subspecies (Figure 2b). In particular, the mean (and standard deviation) of PT was 10.318 (0.514) for clade I, 10.772 (0.857) for clade II, and 10.345 (0.541) for clades (I + II) (i.e. within M. m. domesticus). The corresponding values for clades III and IV, respectively, were 10.850 (0.451) and 11.075 (0.729).
None of the clusters identified based on PT values for each strain was comprised out of groups of related Vkorc1 haplotypes as defined by the phylogenetic analysis [see Additional file 2]. Most notably, haplotype 1 was observed in conjunction with the entire range of PT values present in the data, and it clustered with a wide range of genetically different haplotypes from various distinct clades. For example, the strains CBA/J (haplotype 1) and JF1/Ms (haplotype 5), which differ by 5' polymorphisms as well as one non-synonymous and two synonymous polymorphisms, clustered together based on their PT values (Figure 1b). Similarly, A/J (haplotype 1) and PWK/PhJ (haplotype 9), which differ in at least 10 polymorphisms in the 5' region and 3 synonymous SNPs, clustered together.
Statistical analysis confirmed this suggested lack of association between PT and Vkorc1 haplotypes within each of the clades I-IV (Mann-Whitney U tests, all N.S. at α = 0.05). Moreover, no significant association was found between haplotypes identified in M. m. domesticus and PT (clades I vs. II) (P = 0.389), suggesting 5' polymorphisms have little if any effect. Even clades differing by a non-synonymous SNP (A21T) as well as 5' polymorphisms did not significantly differ in PT (clades III vs. IV) (P = 0.733). Thus, association between the vitamin K-dependent phenotype PT and haplotype variation of the Vkorc1 was not statistically supported within subspecies and between sister clades of subspecies as defined by the Vkorc1 haplotype tree.
However, the genetic distance between Vkorc1 haplotypes accrued between more divergent subspecies was correlated with differences in PT (clades I vs. III; P = 0.022) and (clades I vs. IV; P = 0.006). However, given the large number of mutations across the genome separating these clades, this association is, in our view, not an indication of an association between Vkorc1 haplotypes and variation in PT between clades.
Relationship between Vkorc1 polymorphisms and bone mineralization
Similar to the results for PT variation in BMD within and between clades I-IV was substantial (Figure 2b), with the entire range of variation present in M. m. domesticus (clades 1+II). In particular, the mean (and standard deviation) of BMD was 0.055 (0.005) for clade I, 0.053 (0.005) for clade II, and 0.054 (0.005) for clades (I + II) (i.e. within M. m. domesticus). The corresponding values for clades III and IV, respectively, were 0.046 (0.001) and 0.046 (0.002). A similar pattern was observed for BMC (Figure 2b).
Analysis of BMD and BMC values resulted in two major clusters [c.f. Additional file 2]. One major cluster was comprised out of the M. m. domesticus haplotypes 1, 2, and 7. The other major cluster included a wide range of genetically different haplotypes from all subspecies of Mus, suggesting that there is no clear association of Vkorc1 haplotypes with BMD and BMC values. For instance, haplotype 1, which was found in conjunction with the entire range of BMD and BMC values, also clustered together with SPRET/EiJ (haplotype 10) based on BMD. However, these two haplotypes differ by as many as 4 non-synonymous SNPs and 21 polymorphisms in the 5' region (Figure 1b).
Statistical analysis confirmed this suggested lack of association between BMD and BMC and Vkorc1 haplotypes within each of the clades I-IV (Mann-Whitney U tests, all N.S. at α = 0.05). Moreover, no significant association was found between haplotypes identified in M. m. domesticus and BMD and BMC (clade I vs. II) (P = 0.510 for BMD and P = 0.582 for BMC) suggesting 5' polymorphisms have little if any effect on the phenotypes. Even clades differing by a non-synonymous SNP (A21T) as well as 5' polymorphisms did not significantly differ in BMD and BMC (clade III vs. IV) (P = 0.671 for BMD and P = 0.497 for BMC). Thus, within subspecies or between closely related subspecies no association between the vitamin K dependent phenotypes BMD and BMC and haplotype variation of the Vkorc1 was found.
However, when genetically more divergent clades were compared an association of Vkorc1 haplotypes with BMD and BMC was observed (clade I vs. III; P < 0.001 for BMD and P = 0.001 for BMC and clade I vs. IV; P < 0.001 for BMD and P = 0.001 for BMC). However, as discussed above for PT, this association should not be interpreted as direct evidence for the role of the Vkorc1 in determining BMD and BMC.