Limits...
Genes Regulated by Vitamin D in Bone Cells Are Positively Selected in East Asians.

Arciero E, Biagini SA, Chen Y, Xue Y, Luiselli D, Tyler-Smith C, Pagani L, Ayub Q - PLoS ONE (2015)

Bottom Line: Comparing allele frequency-spectrum-based summary statistics between these gene sets and matched control genes, we observed a selection signal specific to East Asians for a gene set associated with vitamin D action in bones.Examination of population differentiation and haplotypes allowed us to identify several candidate causal regulatory variants in each gene.We also observed haplotype sharing between East Asians, Finns and an extinct archaic human (Denisovan) sample at the CXXC1 locus, which is best explained by incomplete lineage sorting.

View Article: PubMed Central - PubMed

Affiliation: The Wellcome Trust Sanger Institute, Wellcome Genome Campus, Hinxton, CB10 1SA, United Kingdom.

ABSTRACT
Vitamin D and folate are activated and degraded by sunlight, respectively, and the physiological processes they control are likely to have been targets of selection as humans expanded from Africa into Eurasia. We investigated signals of positive selection in gene sets involved in the metabolism, regulation and action of these two vitamins in worldwide populations sequenced by Phase I of the 1000 Genomes Project. Comparing allele frequency-spectrum-based summary statistics between these gene sets and matched control genes, we observed a selection signal specific to East Asians for a gene set associated with vitamin D action in bones. The selection signal was mainly driven by three genes CXXC finger protein 1 (CXXC1), low density lipoprotein receptor-related protein 5 (LRP5) and runt-related transcription factor 2 (RUNX2). Examination of population differentiation and haplotypes allowed us to identify several candidate causal regulatory variants in each gene. Four of these candidate variants (one each in CXXC1 and RUNX2 and two in LRP5) had a >70% derived allele frequency in East Asians, but were present at lower (20-60%) frequency in Europeans as well, suggesting that the adaptation might have been part of a common response to climatic and dietary changes as humans expanded out of Africa, with implications for their role in vitamin D-dependent bone mineralization and osteoporosis insurgence. We also observed haplotype sharing between East Asians, Finns and an extinct archaic human (Denisovan) sample at the CXXC1 locus, which is best explained by incomplete lineage sorting.

Show MeSH

Related in: MedlinePlus

Positive selection at the LRP5 locus.(A). A 140 kb region on chromosome 6 that spans LRP5 showing GENCODE (Version 19) transcript annotation. Positions of 11 candidate regulatory variants and DNase I Hypersensitivity Clusters are shown along with the −log10 of the combined p-values from frequency-spectrum-based analysis in three continental populations. The significance threshold is indicated by the dashed line and two non-overlapping 10 kb windows have a significant combined p-value in CHB. (B). A closer look at the 3’ selected region in East Asians (highlighted in blue). The region contains both variants with the highest derived allele frequency in East Asians (rs649772 and rs671494) that lie in a DNase I hypersensitivity cluster and show ENCODE chromatin state segmentation associated with enhancer binding in several cell lines. In osteoblasts the variants lie within the histone sequence variant, H2A.Z, that determines accessibility of the transcription start site (dark bands) and there are additional H3K4me1 and H3K27ac histone modifications upstream of the variant. The candidate regulatory variant site also shows binding for many transcription factors. The lower part of the panel shows median joining haplotype networks in a ~20 kb region that is in high LD (r2 ≥ 0.95) in CHB. Phased haplotypes generated by the 1000 Genomes Project were used to construct this network. The derived alleles for the regulatory variants rs649772 and rs671494 lie on the branch leading towards the most frequent haplotype found in East Asians and shows a star like expansion typical of a selection signal. The non-synonymous variant rs3736228 (red line) that is associated with bone mineral density in genome wide association studies lies on a separate branch.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4697808&req=5

pone.0146072.g004: Positive selection at the LRP5 locus.(A). A 140 kb region on chromosome 6 that spans LRP5 showing GENCODE (Version 19) transcript annotation. Positions of 11 candidate regulatory variants and DNase I Hypersensitivity Clusters are shown along with the −log10 of the combined p-values from frequency-spectrum-based analysis in three continental populations. The significance threshold is indicated by the dashed line and two non-overlapping 10 kb windows have a significant combined p-value in CHB. (B). A closer look at the 3’ selected region in East Asians (highlighted in blue). The region contains both variants with the highest derived allele frequency in East Asians (rs649772 and rs671494) that lie in a DNase I hypersensitivity cluster and show ENCODE chromatin state segmentation associated with enhancer binding in several cell lines. In osteoblasts the variants lie within the histone sequence variant, H2A.Z, that determines accessibility of the transcription start site (dark bands) and there are additional H3K4me1 and H3K27ac histone modifications upstream of the variant. The candidate regulatory variant site also shows binding for many transcription factors. The lower part of the panel shows median joining haplotype networks in a ~20 kb region that is in high LD (r2 ≥ 0.95) in CHB. Phased haplotypes generated by the 1000 Genomes Project were used to construct this network. The derived alleles for the regulatory variants rs649772 and rs671494 lie on the branch leading towards the most frequent haplotype found in East Asians and shows a star like expansion typical of a selection signal. The non-synonymous variant rs3736228 (red line) that is associated with bone mineral density in genome wide association studies lies on a separate branch.

Mentions: We used functional annotation, population differentiation and extent of linkage disequilibrium (LD) to identify several candidate regulatory variants in each gene that were most likely to be responsible for the adaptation signal we observed. All the candidate variants that we identified had the highest derived allele frequency in East Asians and lay in regions showing DNAse I hypersensitivity or ENCODE functional annotation in osteoblasts, supporting our conclusion that these could be responsible for the selection signal (Table 1). Simulations have shown that the window with the most significant frequency spectrum-based combined p-value lies within the 40 kb region surrounding the selected allele and that the Nielsen et al.’s CLR is better at localizing the selection signal due to selective sweeps and is robust to “assumptions regarding recombination rates and demography” [46]. All our candidate regulatory variants lay either within the selected 10 kb window (CXXC1) or were within 500–850 bp of the 10 kb window with the most significant combined p-value and were in high LD (r2 ≥ 0.8) with SNPs within the selected window (LRP5 and RUNX2). These included one variant each in CXXC1 (rs59393148) and RUNX2 (rs2677100) and two in LRP5 (rs671494 and rs649772). The two LRP5 variants were next to each other and in perfect LD suggestive of a multi-nucleotide polymorphism that might have arisen by a single mutation. A benign non-synonymous SNP (rs3736228) in LRP5, which is also in LD with our candidate variants, is another attractive candidate for the selection signal in East Asians. The derived T allele of this variant (rs3736228) has been associated with low bone mineral density and risk of osteoporosis. However, the derived T allele frequency in East Asians is ~ 20% and it seems unlikely to be the signal picked up by the frequency-spectrum-based tests. In addition, the two candidate regulatory variants in LRP5 lie on the branch leading towards the selected haplotype in the median-joining haplotype network (Fig 4).


Genes Regulated by Vitamin D in Bone Cells Are Positively Selected in East Asians.

Arciero E, Biagini SA, Chen Y, Xue Y, Luiselli D, Tyler-Smith C, Pagani L, Ayub Q - PLoS ONE (2015)

Positive selection at the LRP5 locus.(A). A 140 kb region on chromosome 6 that spans LRP5 showing GENCODE (Version 19) transcript annotation. Positions of 11 candidate regulatory variants and DNase I Hypersensitivity Clusters are shown along with the −log10 of the combined p-values from frequency-spectrum-based analysis in three continental populations. The significance threshold is indicated by the dashed line and two non-overlapping 10 kb windows have a significant combined p-value in CHB. (B). A closer look at the 3’ selected region in East Asians (highlighted in blue). The region contains both variants with the highest derived allele frequency in East Asians (rs649772 and rs671494) that lie in a DNase I hypersensitivity cluster and show ENCODE chromatin state segmentation associated with enhancer binding in several cell lines. In osteoblasts the variants lie within the histone sequence variant, H2A.Z, that determines accessibility of the transcription start site (dark bands) and there are additional H3K4me1 and H3K27ac histone modifications upstream of the variant. The candidate regulatory variant site also shows binding for many transcription factors. The lower part of the panel shows median joining haplotype networks in a ~20 kb region that is in high LD (r2 ≥ 0.95) in CHB. Phased haplotypes generated by the 1000 Genomes Project were used to construct this network. The derived alleles for the regulatory variants rs649772 and rs671494 lie on the branch leading towards the most frequent haplotype found in East Asians and shows a star like expansion typical of a selection signal. The non-synonymous variant rs3736228 (red line) that is associated with bone mineral density in genome wide association studies lies on a separate branch.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4697808&req=5

pone.0146072.g004: Positive selection at the LRP5 locus.(A). A 140 kb region on chromosome 6 that spans LRP5 showing GENCODE (Version 19) transcript annotation. Positions of 11 candidate regulatory variants and DNase I Hypersensitivity Clusters are shown along with the −log10 of the combined p-values from frequency-spectrum-based analysis in three continental populations. The significance threshold is indicated by the dashed line and two non-overlapping 10 kb windows have a significant combined p-value in CHB. (B). A closer look at the 3’ selected region in East Asians (highlighted in blue). The region contains both variants with the highest derived allele frequency in East Asians (rs649772 and rs671494) that lie in a DNase I hypersensitivity cluster and show ENCODE chromatin state segmentation associated with enhancer binding in several cell lines. In osteoblasts the variants lie within the histone sequence variant, H2A.Z, that determines accessibility of the transcription start site (dark bands) and there are additional H3K4me1 and H3K27ac histone modifications upstream of the variant. The candidate regulatory variant site also shows binding for many transcription factors. The lower part of the panel shows median joining haplotype networks in a ~20 kb region that is in high LD (r2 ≥ 0.95) in CHB. Phased haplotypes generated by the 1000 Genomes Project were used to construct this network. The derived alleles for the regulatory variants rs649772 and rs671494 lie on the branch leading towards the most frequent haplotype found in East Asians and shows a star like expansion typical of a selection signal. The non-synonymous variant rs3736228 (red line) that is associated with bone mineral density in genome wide association studies lies on a separate branch.
Mentions: We used functional annotation, population differentiation and extent of linkage disequilibrium (LD) to identify several candidate regulatory variants in each gene that were most likely to be responsible for the adaptation signal we observed. All the candidate variants that we identified had the highest derived allele frequency in East Asians and lay in regions showing DNAse I hypersensitivity or ENCODE functional annotation in osteoblasts, supporting our conclusion that these could be responsible for the selection signal (Table 1). Simulations have shown that the window with the most significant frequency spectrum-based combined p-value lies within the 40 kb region surrounding the selected allele and that the Nielsen et al.’s CLR is better at localizing the selection signal due to selective sweeps and is robust to “assumptions regarding recombination rates and demography” [46]. All our candidate regulatory variants lay either within the selected 10 kb window (CXXC1) or were within 500–850 bp of the 10 kb window with the most significant combined p-value and were in high LD (r2 ≥ 0.8) with SNPs within the selected window (LRP5 and RUNX2). These included one variant each in CXXC1 (rs59393148) and RUNX2 (rs2677100) and two in LRP5 (rs671494 and rs649772). The two LRP5 variants were next to each other and in perfect LD suggestive of a multi-nucleotide polymorphism that might have arisen by a single mutation. A benign non-synonymous SNP (rs3736228) in LRP5, which is also in LD with our candidate variants, is another attractive candidate for the selection signal in East Asians. The derived T allele of this variant (rs3736228) has been associated with low bone mineral density and risk of osteoporosis. However, the derived T allele frequency in East Asians is ~ 20% and it seems unlikely to be the signal picked up by the frequency-spectrum-based tests. In addition, the two candidate regulatory variants in LRP5 lie on the branch leading towards the selected haplotype in the median-joining haplotype network (Fig 4).

Bottom Line: Comparing allele frequency-spectrum-based summary statistics between these gene sets and matched control genes, we observed a selection signal specific to East Asians for a gene set associated with vitamin D action in bones.Examination of population differentiation and haplotypes allowed us to identify several candidate causal regulatory variants in each gene.We also observed haplotype sharing between East Asians, Finns and an extinct archaic human (Denisovan) sample at the CXXC1 locus, which is best explained by incomplete lineage sorting.

View Article: PubMed Central - PubMed

Affiliation: The Wellcome Trust Sanger Institute, Wellcome Genome Campus, Hinxton, CB10 1SA, United Kingdom.

ABSTRACT
Vitamin D and folate are activated and degraded by sunlight, respectively, and the physiological processes they control are likely to have been targets of selection as humans expanded from Africa into Eurasia. We investigated signals of positive selection in gene sets involved in the metabolism, regulation and action of these two vitamins in worldwide populations sequenced by Phase I of the 1000 Genomes Project. Comparing allele frequency-spectrum-based summary statistics between these gene sets and matched control genes, we observed a selection signal specific to East Asians for a gene set associated with vitamin D action in bones. The selection signal was mainly driven by three genes CXXC finger protein 1 (CXXC1), low density lipoprotein receptor-related protein 5 (LRP5) and runt-related transcription factor 2 (RUNX2). Examination of population differentiation and haplotypes allowed us to identify several candidate causal regulatory variants in each gene. Four of these candidate variants (one each in CXXC1 and RUNX2 and two in LRP5) had a >70% derived allele frequency in East Asians, but were present at lower (20-60%) frequency in Europeans as well, suggesting that the adaptation might have been part of a common response to climatic and dietary changes as humans expanded out of Africa, with implications for their role in vitamin D-dependent bone mineralization and osteoporosis insurgence. We also observed haplotype sharing between East Asians, Finns and an extinct archaic human (Denisovan) sample at the CXXC1 locus, which is best explained by incomplete lineage sorting.

Show MeSH
Related in: MedlinePlus