Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Jul 2;36(11):2591-2603.
doi: 10.1093/molbev/msz158. Online ahead of print.

EPAS1 Gain-Of-Function Mutation Contributes to High-Altitude Adaptation in Tibetan Horses

Free PMC article

EPAS1 Gain-Of-Function Mutation Contributes to High-Altitude Adaptation in Tibetan Horses

Xuexue Liu et al. Mol Biol Evol. .
Free PMC article


High altitude represents some of the most extreme environments worldwide. The genetic changes underlying adaptation to such environments have been recently identified in multiple animals but remain unknown in horses. Here, we sequence the complete genome of 138 domestic horses encompassing a whole altitudinal range across China to uncover the genetic basis for adaptation to high-altitude hypoxia. Our genome dataset includes 65 lowland animals across ten Chinese native breeds, 61 horses living at least 3,300 meters above sea level across seven locations along Qinghai-Tibet Plateau, as well as 7 Thoroughbred and 5 Przewalski's horses added for comparison. We find that Tibetan horses do not descend from Przewalski's horses but were most likely introduced from a distinct horse lineage, following the emergence of pastoral nomadism in Northwestern China ∼3,700 years ago. We identify that the endothelial PAS domain protein 1 gene (EPAS1, alsoHIF2A) shows the strongest signature for positive selection in the Tibetan horse genome. Two missense mutations at this locus appear strongly associated with blood physiological parameters facilitating blood circulation as well as oxygen transportation and consumption in hypoxic conditions. Functional validation through protein mutagenesis shows that these mutations increase EPAS1 stability and its hetero dimerization affinity to ARNT (HIF1B). Our study demonstrates that missense mutations in the EPAS1 gene provided key evolutionary molecular adaptation to Tibetan horses living in high-altitude hypoxic environments. It reveals possible targets for genomic selection programs aimed at increasing hypoxia tolerance in livestock and provides a textbook example of evolutionary convergence across independent mammal lineages.

Keywords: EPAS1; Tibetan horse; convergence; hypoxia adaptation; metabolism; respiration.


<sc>Fig</sc>. 1.
Fig. 1.
Geographic distribution, genetic structure and evolutionary relationships of Chinese native horse breeds. (A) The geographic distribution of 15 Chinese native horse populations and one Przewalski’s horse population (PrZ, black color). The blue color (hollow shapes) represents Tibetan horses (JiZi, Jiangzi; LKZ, Langkazi; MZH, Mozhu; NiMu, Nimu) living at least 4,000 m above the sea level (m.a.s.l.) whereas the blue color (solid shapes) represents Qinghai horses (CDM, Chaidamu; HeQu, Hequ; DaTo, Datong) living at an altitude of at least 3,000 m. The orange color represents the Southwestern horses (DeBa, Debao pony from Guangxi; JiCh, Jianchang pony from Sichuan; NiQi, Ningqiang pony from Shannxi). The green represents the Northern horses (MoGo, Mongolian horse from Inner Mongolia; WMG, Mongolian horse from Mongolia; ELC, ErlunChun horse from Heilongjiang; YaQi, Yanqi horse from Xinjiang; Yili horse from Xinjiang). The colors of symbols that indicate the geographic regions are the same as those in the PCA plots and phylogenetic trees. (B) PCA plots of the first two components of all horse samples (inner plot) and all Chinese native horses (outer plot). The fraction of the total variance explained is reported on each individual axis between parentheses. (C) The ML-TreeMix tree of all horses, with PrZ as outgroup, assuming four migration events. Four migration events that are most consistent with known events, because they increased the explained variance to 99.8% (Pickrell and Pritchard 2012), and does not increase afterwards (supplementary table S25, Supplementary Material online). Migration arrows are colored according to their weights. Horizontal branch lengths are proportional to the amount of genetic drift parameter that has occurred on the branch. The drift parameter measures the variance in allele frequency estimated along each branch of the tree. The yellow and orange lines indicate the instantaneous admixtures, whereas arrows denote continuous (unidirectional) gene flow. (D) The Neighbor-Joining tree of the horse breeds, with PrZ as outgroup. Bootstrap reported was close to 100%. (E) ∂a∂i best-supported population model depicting the evolutionary trajectories of the main three clusters of Chinese native horses. The light blue, green, dark blue and orange rectangles represent the ancestral, Northern China (NC), Qinghai-Tibetan (QT) and Southwestern (SW) populations, respectively. The numbers within the rectangles represent the effective size (individual horses) for the corresponding population. The average number of migrants per year between the different groups is shown between the black arrows. Ne = effective size (individuals). T = time of divergence (years).
<sc>Fig</sc>. 2.
Fig. 2.
Positive selection scans for adaptation to high altitude hypoxia. Horses living at high altitude (QT) are compared with lowland controls (NC and SW). The population genetic differentiation FST values (A), the nucleotide diversity θπ ratios (θπ-LL/θπ-QT) (B) and the transformed heterozygosity score ZHP (C) are calculated within 100 kb sliding windows (step size = 15 kb). The significance threshold of selection signature was arbitrarily set to top 5% percentile outliers for each individual test and is indicated with red horizontal dashed lines. The black horizontal dashed lines display the top 1% quantile.
<sc>Fig</sc>. 3.
Fig. 3.
The strongest positive selection signatures around the EPAS1 peak. The θπ ratio (θπ-LL/θπ-QT) (A), Tajima’s D (B) and FST value (C) are plotted against the peak position from 51.5 Mb to 53.5 Mb on chromosome 15. Both π ratio and Tajima’s D values were based on a 20 kb window and a 20 kb step. The black and the red lines represent the Tajima’s D values for high-altitude and lowland horses, respectively. The gray columns represent the strongest positive selection signatures in the region considered. The small black boxes and short lines represent the gene structure of EPAS1, which is the only gene within the strongest selective signal. The red dot represents the significant threshold of FST value per SNP > 0.3 and the red dashed line represents the FST threshold. The top two SNPs are noted by black arrows.
<sc>Fig</sc>. 4.
Fig. 4.
Annotation and validation of the EPAS1 missense SNPs showing positive selection signatures. (A) EPAS1 protein sequence analysis. The protein coordinates are based on the ENSECAT00000015683.1Ensembl protein. The upper panel shows the Pfam domains of the EPAS1 protein, including the Per-Arnt-Sim (PAS) domains (red), the basic helix-loop-helix (HLH) domain (yellow), the hypoxia-inducible (HIF) domain (blue) and the C-terminal transactivation (CTAD) domain (orange). Protein sequence polymorphisms present in Tibetan horses and 22 vertebrates are provided. (B) Genotypes were determined using the KASP technology in Przewalski’s horses, Thoroughbred horses and 27 Chinese native horse breeds (N = 908 horses). The region and altitude information are indicated by the red line. (C) Western-blotting analysis of the A549 cell lysates transfected with the GFP-tagged recombinant plasmid of WT, R144C and E263D or the empty vector. WT, R144C and E263D represent the plasmid overexpressing the wild-type EPAS1 protein, SNP1 mutant and SNP2 mutant protein, respectively, REF represents the empty vector. The antiGFP and antiTubulin antibodies were used to measure the protein expression of EPAS1 and the internal reference protein Tubulin, respectively. (D) Validation of EPAS1 and ARNT interactions. A549 cells were cotransfected with myc-tagged EPAS1 and flag-tagged ARNT for 48 h, followed by immunoprecipitation against myc tag and immunoblotting against flag and myc. (E) The qPCR gene expression of the EPAS1 downstream genes, including VEGFA, VHL and LDHA in the transfected A549 cells. EPO expression levels were measured in HepG2 cells. * and ** displayed the statistical significances of P-values <0.05 and 0.01, respectively.
<sc>Fig</sc>. 5.
Fig. 5.
Genotype–phenotype association of the EPAS1-R144C mutation. Association analysis of the EPAS1-R144C SNP (SNP1: mutant allele, A; reference allele, G) with the individual hemoglobin level and blood physiological values in Hequ (HeQu) and Guanzhong horses (GZg). We found a significant association between the genotypes and HCT (A), HMG (B), MCHC (C), HBDH (D), LDH (E) and CK (F). ** Displays the statistical significance of P-value <0.01 according to the ANOVA F-test.

Similar articles

See all similar articles

Cited by 1 article


    1. Alexander DH, Novembre J, Lange K. 2009. Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 199:1655–1664. - PMC - PubMed
    1. Beall CM, Cavalleri GL, Deng L, Elston RC, Gao Y, Knight J, Li C, Li JC, Liang Y, McCormack M., et al. 2010. Natural selection on EPAS1 (HIF2α) associated with low hemoglobin concentration in Tibetan highlanders. Proc Natl Acad Sci U S A. 10725:11459–11464. - PMC - PubMed
    1. Beall CM, Decker MJ, Brittenham GM, Kushner I, Gebremedhin A, Strohl KP. 2002. An Ethiopian pattern of human adaptation to high-altitude hypoxia. Proc Natl Acad Sci U S A. 9926:17215–17218. - PMC - PubMed
    1. Bielejec F, Rambaut A, Suchard MA, Lemey P. 2011. SPREAD: spatial phylogenetic reconstruction of evolutionary dynamics. Bioinformatics 2720:2910–2912. - PMC - PubMed
    1. Bigham AW, Lee FS. 2014. Human high-altitude adaptation: forward genetics meets the HIF pathway. Genes Dev. 2820:2189–2204. - PMC - PubMed