A genetic mechanism for Tibetan high-altitude adaptation

Abstract

Tibetans do not exhibit increased hemoglobin concentration at high altitude. We describe a high-frequency missense mutation in the EGLN1 gene, which encodes prolyl hydroxylase 2 (PHD2), that contributes to this adaptive response. We show that a variant in EGLN1, c.[12C>G; 380G>C], contributes functionally to the Tibetan high-altitude phenotype. PHD2 triggers the degradation of hypoxia-inducible factors (HIFs), which mediate many physiological responses to hypoxia, including erythropoiesis. The PHD2 p.[Asp4Glu; Cys127Ser] variant exhibits a lower K(m) value for oxygen, suggesting that it promotes increased HIF degradation under hypoxic conditions. Whereas hypoxia stimulates the proliferation of wild-type erythroid progenitors, the proliferation of progenitors with the c.[12C>G; 380G>C] mutation in EGLN1 is significantly impaired under hypoxic culture conditions. We show that the c.[12C>G; 380G>C] mutation originated ∼8,000 years ago on the same haplotype previously associated with adaptation to high altitude. The c.[12C>G; 380G>C] mutation abrogates hypoxia-induced and HIF-mediated augmentation of erythropoiesis, which provides a molecular mechanism for the observed protection of Tibetans from polycythemia at high altitude.

Figure 1

Genome-wide allele frequency differentiation between Tibetans, Mongolians and Europeans. To measure genome-wide differentiation in allele frequency between Tibetans and Mongolians, we calculated the PBS for the c.12C>G and c.380G>C variants in EGLN1 for each Affymetrix SNP. We included all unrelated Tibetan, Mongolian and European individuals in the analysis for whom we had Affymetrix 6.0, c.12C>G and c.380G>C genotype data. The final set included 16 Tibetans, 11 Mongolians and 14 Europeans. The PBS for c.12C>G was 1.02 (empirical P value = 9 × 10⁻⁶). The four SNPs with PBS greater than 1.02 were all located in the EGLN1 region.

Figure 2

The p.[Asp4Glu; Cys127Ser] PHD2 mutant shows gain of function under hypoxia. (a) The K_m value for O₂ for the p.[Asp4Glu; Cys127Ser] PHD2 mutant is lower than that for wild-type PHD2. Shown is the effect of O₂ concentration on the reaction velocity of wild-type PHD2 (red line) and the p.[Asp4Glu; Cys127Ser] mutant (blue line), using the HIF-2α ODDD as a substrate. The inset shows the corresponding double-reciprocal plot. The data presented are the average ± s.e.m. of eight independent assays comparing wild-type PHD2 and the p.[Asp4Glu; Cys127Ser] mutant in the same assay. Data from independent assays were normalized relative to the disintegrations per minute (d.p.m.) obtained with wild-type PHD2 at an O₂ concentration of 212 μM. (b) Decreased HIF-2α accumulation in hypoxic cells expressing p.[Asp4Glu; Cys127Ser] PHD2 in knockdown-rescue experiments. Shown is a representative immunoblot analysis of Hep3B cells expressing Venus fluorescent protein (V), wild-type (WT) PHD2 or p.[Asp4Glu; Cys127Ser] PHD2. Cells were stably coinfected with an shRNA targeting the 3′ UTR of endogenous EGLN1 (not present in the rescue constructs) or a negative control shRNA and were grown in the presence of 21% or 1% O₂ for 12 h. Decreased HIF-2α levels in cells expressing the p.[Asp4Glu; Cys127Ser] variant were observed in three independent experiments. The electrophoretic mobility of the exogenous PHD2 protein was slightly decreased owing to the presence of the Flag epitope. Tubulin was used as a loading control. (c) RT-PCR demonstrates knockdown of endogenous EGLN1 in Hep3B cells by the EGLN1-targeted shRNA and equal levels of transgene transcripts in cells expressing the transgenes for wild-type and p.[Asp4Glu; Cys127Ser] PHD2. Shown are representative data from three independent experiments.

Figure 3

Erythroid colony (BFU-E) assays. Erythroid colonies were enumerated after stimulation with various concentrations of EPO. (a) BFU-Es homozygous for the c.[12C>G; 380G>C] variant in EGLN1 are hypersensitive to EPO under normoxia. BFU-E colony assays in Tibetans homozygous (n = 2 biological replicates) or heterozygous (n = 1) for the c.[12C>G; 380G>C] variant in comparison to normal controls (n = 10) at different EPO concentrations. Values are normalized to the number of BFU-Es grown at ambient oxygen tension at 3,000 mU/ml EPO, which was expressed as 100%. Designated error bars in controls and for cells homozygous for c.[12C>G; 380G>C] represent ±s.d. (b) Sensitivity of BFU-Es to low EPO concentration (15 mU/ml). Hypersensitive BFU-Es from Tibetans with the c.[12C>G; 380G>C] variant (n = 6) lose their hypersensitivity to EPO as the oxygen tension decreases to 5% and fail to form colonies at 1% O₂, whereas control colonies (n = 3) become hypersensitive to EPO with increasing hypoxia. (c) Colony growth of progenitor cells from a Tibetan with the c.[12C>G; 380G>C] mutation and controls at 3,000 mU/ml EPO. Tibetan BFU-E colonies with the c.[12C>G; 380G>C] mutation have decreased colony formation with ambient O₂ levels and 5% O₂, and they fail to grow at 1% O₂. In contrast, BFU-Es from control subjects (n = 3) form more colonies as the oxygen tension decreases. (d) Size of erythroid colonies increases in BFU-Es from controls but decreases for progenitor cells from a Tibetan with the c.[12C>G; 380G>C] mutation. Average BFU-E colony size represented in pixels (y axis) is greater in BFU-E colonies from the control subject under hypoxia (5% O₂), whereas, in the Tibetan subject with the c.[12C>G; 380G>C] mutation, hypoxia decreases BFU-E colony size. Error bars in Tibetans and controls represent ±s.d. (Fig. 3a is an independent experiment performed with different individuals and at a different time than the experiments in Fig. 3b–d, which consist of 6 biological replicates (5 Tibetans homozygous and 1 Tibetan heterozygous for c.[12C>G; 380G>C]) together with 3 controls.)

Figure 4

Hypoxia increases the proliferation of control BFU-E colonies but decreases that of colonies with the c.[12C>G; 380G>C] mutation in EGLN1. BFU-Es were grown in 3,000 mU/ml EPO, and all images were acquired at 40× magnification (scale bars, 1 mm). (a,b) Representative colonies from a control (wild type) subject and (c,d) a Tibetan subject homozygous for the c.[12C>G; 380G>C] mutation. Note the larger colony sizes for the control BFU-E colonies under hypoxia (5% O₂) (b) relative to ambient oxygen tension (a). BFU-Es with the c.[12C>G; 380G>C] mutation exhibit smaller colony sizes under normoxia (c) in comparison to control cells (a), and the colonies are paler than the controls, reflecting the decreased hemoglobinization with the c.[12C>G; 380G>C] erythroid progenitors in comparison to the controls. (d) The decrease in colony size and hemoglobinization for Tibetan BFU-Es with the c.[12C>G; 380G>C] mutation is even more pronounced under 5% O₂ (in comparison to b).