EGLN1 [encoding HIF (hypoxia-inducible factor)-prolyl hydroxylase 2] plays a pivotal role in the HIF pathway and has emerged as one of the most intriguing genes with respect to physiology at HA (high altitude). EGLN1, being an actual oxygen sensor, appears to have a potential role in the functional adaptation to the hypobaric hypoxic environment. In the present study, we screened 30 polymorphisms of EGLN1, evaluated its gene expression and performed association analyses. In addition, the role of allelic variants in altering TF (transcription factor)-binding sites and consequently the replacement of TFs at these loci was also investigated. The study was performed in 250 HAPE-p [HAPE (HA pulmonary oedema)-patients], 210 HAPE-f (HAPE-free controls) and 430 HLs (healthy Ladakhi highland natives). The genotypes of seven polymorphisms, rs1538664, rs479200, rs2486729, rs2790879, rs480902, rs2486736 and rs973252, differed significantly between HAPE-p and HAPE-f (P<0.008). The genotypes AA, TT, AA, GG, CC, AA and GG of rs1538664, rs479200, rs2486729, rs2790879, rs480902, rs2486736 and rs973252, prevalent in HAPE-p, were identified as risk genotypes and their counterpart homozygotes, prevalent in HLs, were identified as protective. EGLN1 expression was up-regulated 4.56-fold in HAPE-p (P=0.0084). The risk genotypes, their haplotypes and interacting genotypes were associated with up-regulated EGLN1 expression (P<0.05). Similarly, regression analysis showed that the risk alleles and susceptible haplotypes were associated with decreased SaO2 (arterial oxygen saturation) levels in the three groups. The significant inverse correlation of SaO2 levels with PASP (pulmonary artery systolic pressure) and EGLN1 expression and the association of these polymorphisms with SaO2 levels and EGLN1 expression contributed to uncovering the molecular mechanism underlying hypobaric hypoxic adaptation and maladaptation.

CLINICAL PERSPECTIVES

  • One of the clinical features of HAPE is low blood SaO2 levels, and EGLN1, an oxygen-sensing pathway gene, has been shown be naturally selected at HA.

  • To provide more insights into the variants within the EGLN1 gene towards susceptibility to HAPE, we screened several polymorphisms, evaluated gene expression and performed various association analyses. In addition, the role of allele variants in altering TF-binding sites was also investigated. The genotypes of seven polymorphisms differed significantly between HAPE-p and HAPE-f, with HAPE-p having significantly up-regulated EGLN1 expression, which correlated with the risk genotypes and haplotypes. These genetic markers were even associated with decreased SaO2 levels, which indicated their potential functional consequences.

  • These results add to our understanding of the mechanisms underlying hypobaric hypoxia-associated disorders, such as HAPE, and also to numerous respiratory and cardiovascular diseases where hypoxia is explicitly involved.

INTRODUCTION

Exposure to HA (high altitude) results in reduced blood SaO2 (arterial oxygen saturation) that stimulates an array of physiological responses in the body [1,2]. HIF (hypoxia-inducible factor), a basic helix–loop–helix PER/ARNT/SIM TF (transcription factor), helps maintain cellular oxygen homoeostasis during hypoxic stress [3]. It mediates transcriptional responses to hypoxia by regulating a number of genes pertaining to different pathways, such as erythropoiesis, angiogenesis and glycolysis, thereby regulating the cellular oxygen content of the body [25]. EGLN1 (HIF-prolyl hydroxylase 2), a member of the ubiquitous Fe(II) and 2-oxoglutarate-dependent oxygenase superfamily, plays a pivotal role in HIF-1α regulation. EGLN1 destabilizes HIF-1α in normoxic conditions via the VHL (von Hippel–Lindau) ubiquitination complex. It catalyses the post-translational modification of HIF-1α by hydroxylating proline at positions 402 and 564 of the ODD (oxygen-dependent degradation) domain of HIF-1α [4]. The VHL tumour suppressor recognizes this modified HIF for polyubiquitination and proteosomal degradation [57]. Interestingly, EGLN1 also possesses hypoxia-responsive element that makes it a HIF-inducible gene [7]. This pathway, being an actual oxygen sensor, appears to have a potential role in the functional adaptation to the hypobaric hypoxia environment of HA.

In recent years, EGLN1 has been shown to have a strong natural selection in HA populations [815]. This has strongly favoured genetic adaptation in highland populations. However, the same may not be applicable to sojourners because, among these, there are subjects who perform physical activities without showing signs of discomfort. On the contrary, there are susceptible subjects who experience various levels of discomfort upon physical exertion and develop various HA disorders such as HAPE (HA pulmonary oedema). The latter is characterized by a non-cardiogenic pulmonary oedema with vasoconstriction, endothelial dysfunction and intravascular overperfusion [1618]. HAPE is a rare life-threatening condition that occurs in susceptible sojourners upon rapid ascent to or physical exertion at altitudes above 2500 m [19]. Hence investigating the variants within this gene may explain the significance of inter-individual deviations.

Therefore, in the present study, we genotyped 30 polymorphisms of EGLN1, evaluated its gene expression and performed association analyses. In addition, we undertook a TFSEARCH to find the alterations in the TF-binding sites due to allelic changes and as a consequence the replacement of TFs, if any. The study was performed in HAPE-p (HAPE-patients) and HAPE-f (HAPE-free controls) for susceptibility to HAPE and HLs (healthy Ladakhi highland natives) for adaptation. The study detected a number of significant polymorphisms; moreover, their association with both SaO2 levels and EGLN1 expression reflected their functional relevance in the hypobaric hypoxia environment of HA.

MATERIALS AND METHODS

Subject selection

The three groups comprised 250 HAPE-p, 210 HAPE-f and 430 HLs. The HAPE-p and HAPE-f, residing at an altitude of <200 m were unrelated sojourners of Indo–Aryan origin. They all reached Leh, Ladakh (3524 m) by airplane and from there they ascended or descended together to different heights (3500–5600 m). A few of them suffered mountain disorders and they were immediately transferred to Sonam Norboo Memorial Hospital, Leh, Ladakh (~3500 m) for confirmation of diagnosis and treatment. All of the subjects were recruited by Sonam Norboo Memorial Hospital. Each subject was examined for any previous history of cardiopulmonary and infectious diseases through a detailed questionnaire. HAPE-f were healthy individuals, who carried out routine physical activities without suffering from the disorder. The HLs, residing at ≥3500 m, consisted of the Ladakhi population of Tibetan origin.

Human ethical committees of Institute of Genomics and Integrative Biology and Sonam Norboo Memorial Hospital approved the experimental protocol. Each subject was apprised of the study prior to obtaining a written informed consent to participate.

Clinical assessment

The diagnosis of HAPE was based on SaO2 levels, the presence of pulmonary rales, cyanosis, chest X-ray and PASP (pulmonary artery systolic pressure). The clinical symptoms included cough, dyspnoea at rest, breathlessness, absence of infection and reduced exercise performance [20]. Clinical characteristics, such as systolic and diastolic blood pressure, SaO2 levels, PASP, pulse rate and BMI (body mass index) were measured. SaO2 levels were measured using a Finger-Pulse Oximeter 503 (Criticare Systems). PASP was measured using a Hewlett-Packard HP-Sonos 5500 echocardiography machine in a left lateral decubitus position with a 20° upper-body tilt without prior oxygen inhalation by a single experienced trained clinician.

A venous blood sample (10 ml) was drawn in acid/citrate/dextrose anticoagulant tubes. The blood sample from the patients was drawn and clinical assessments were carried out after diagnosis but prior to treatment; whereas, in the healthy groups, the blood sample was drawn after an overnight fast. Plasma and peripheral blood leucocytes were separated; the latter were processed for DNA and RNA. Plasma and RNA were stored at −80°C and DNA at −20°C.

Polymorphism selection and genotyping

After extensive database and literature search for polymorphisms of EGLN1, we restricted our selection to 30 polymorphisms of this gene. The polymorphisms were genotyped by sequenome MS-based genotype assay using iPLEX Gold technology (Supplementary Table S1 at http://www.clinsci.org/cs/124/cs1240479add.htm). These SNPs (single nucleotide polymorphisms) were also evaluated for their tagging efficiency from the genotype data available in the International HapMap Consortium 2005 (Supplementary Table S2 at http://www.clinsci.org/cs/124/cs1240479add.htm). The haplotypes and interacting genotypes were evaluated and compared in the three groups.

EGLN1 expression

Gene expression was performed on 15 samples each from HAPE-p, HAPE-f and HLs. Total RNA was extracted from 2 ml of whole blood using TRI Reagent® RT blood (Molecular Research Centre). The quantity and quality of RNA were determined on a NanoDrop ND-1000 spectrophotometer, and integrity was checked on a 1.5% agarose gel. Total RNA (1.0 μg) was used to generate cDNA by using the EZ-first strand cDNA synthesis kit for RT (reverse transcription)–PCR (Biological Industries). Primers for real-time PCR were designed for EGLN1 and RN18S1 (18S rRNA) using the Pearl Primer software (see Supplementary Table S3 at http://www.clinsci.org/cs/124/cs1240479add.htm). Real-time PCR was performed on an ABI Prism 7300 Sequence Detection System (Applied Biosystems) using a SYBR Green PCR Master Mix (Applied Biosystems). PCR was performed in duplicate and was repeated three times for each gene and each sample. Relative transcript quantities were calculated using the ΔΔCt method with RN18S1 as the endogenous reference gene.

Evaluation of TF-binding sites

The effect on TF-binding sites due to presence of the major or minor allele at a given locus was evaluated. TF-binding sites were identified with a cut-off affinity-binding score of 85 by TFSEARCH-v1.3 (http://www.cbrc.jp/research/db/TFSEARCH.html). The nucleotide sequences with major and minor alleles were retrieved from the SNP database (http://www.ncbi.nlm.nih.gov/projects/SNP).

Statistical analyses

The SPSS-15.0 and EPIINFO-6.0 software were used for analyses. Genotype and allele distributions, ORs (odds ratios) and 95% CIs (confidence intervals) were calculated by multivariate logistic regression. The covariates used were age, gender and BMI. HWE (Hardy–Weinberg equilibrium) was checked using a χ2 goodness-of-fit test. Permutation analysis for haplotype evaluation was performed using the algorithm (Phase-2.1) developed by Stephens et al. [21]. LD (linkage disequilibrium) was measured by Haploview-4.0 [22] and textile plot (http://www.stat.math.keio.ac.jp/TextilePlot/genetics/) [23,24]. The polymorphisms were checked for their tagging efficiency with a cut-off r2 of ≥ 0.8 using the software Tagger (Haploview-4.0; http://www.broadinstitute.org/haploview/haploview) (Supplementary Table 2). The genotype interactions were evaluated by MDR (multi-dimensional reduction) (version 1.2.2) [25]. The quantitative RT–PCR results were analysed by one-way ANOVA. Correlation of genotypes, haplotypes and interacting genotypes with EGLN1 expression was performed using ANCOVA (analysis of covariance). The Pearson's correlation (r) were used to assess the correlation of SaO2 levels with PASP and EGLN1 expression. The association of major genotypes and haplotypes with SaO2 levels was performed using binary logistic regression analysis. An unpaired Student's t test was used for the comparisons between the groups. Values are represented as means±S.D. A P value of <0.05 after adjustment with confounders and Bonferroni's multiple correction was considered statistically significant.

RESULTS

Clinical characteristics

The clinical characteristics of the subjects are shown in Table 1. There was no significant difference between the ages of the three groups. PASP was significantly higher in HAPE-p compared with HAPE-f and HLs (both P<0.0001; Figure 1A and Table 1), whereas the SaO2 levels were significantly lower in HAPE-p compared with HAPE-f and HLs (P<0.0001; Figure 1B and Table 1).

PASP, SaO2 levels and EGLN1 expression in the three groups

Figure 1
PASP, SaO2 levels and EGLN1 expression in the three groups

(A) PASP (mmHg) in HAPE-f, HAPE-p and HLs. (B) SaO2 levels in HAPE-f, HAPE-p and HLs. (C) The relative expression of EGLN1 by real-time PCR expressed as the fold change. EGLN1 was up-regulated 4.55- and 1.5-fold in HAPE-p and HLs respectively, when compared with HAPE-f. Values are means±S.D. (D) Clinical correlations in the three groups. A significant inverse correlation was found between SaO2 levels and PASP in HAPE-p and HLs. (E) Correlation between EGLN1 expression and SaO2 levels. An inverse correlation was obtained.

Figure 1
PASP, SaO2 levels and EGLN1 expression in the three groups

(A) PASP (mmHg) in HAPE-f, HAPE-p and HLs. (B) SaO2 levels in HAPE-f, HAPE-p and HLs. (C) The relative expression of EGLN1 by real-time PCR expressed as the fold change. EGLN1 was up-regulated 4.55- and 1.5-fold in HAPE-p and HLs respectively, when compared with HAPE-f. Values are means±S.D. (D) Clinical correlations in the three groups. A significant inverse correlation was found between SaO2 levels and PASP in HAPE-p and HLs. (E) Correlation between EGLN1 expression and SaO2 levels. An inverse correlation was obtained.

Table 1
Clinical characteristics of the HAPE-p, HAPE-f and HLs

Results are means±S.D. and are compared using an unpaired Student's t test. SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; PR, pulse rate.

    P values 
Characteristic HAPE-p (nHAPE-f (nHLs (nHAPE-p compared with HAPE-f HAPE-p compared with HLs HAPE-f compared with HLs 
Age (years) 29.1±13.86 (250) 27.8±10.4 (210) 39.05±7.44 (430) 0.263 <0.0001 <0.0001 
SBP (mmHg) 124.6±21.1 (250) 119.8±3.5 (210) 121±8.0 (430) 0.001 0.001 0.04 
DBP (mmHg) 82.7±17.0 (250) 80.4±10.0 (210) 81.1±10.0 (430) 0.085 0.122 0.406 
MAP (mmHg) 96.7±16.0 (250) 93.5±10.0 (210) 94.4±10.0 (430) 0.012 0.021 0.285 
PR (rate/min) 92.2±22.9 (250) 80.1±19.9 (210) 84.0±15.7 (430) <0.0001 <0.0001 0.007 
SaO2 levels (%) 71.7±12.5 (250) 91.2±3.7 (210) 89.6±3.2 (430) <0.0001 <0.0001 <0.0001 
PASP (mmHg) 50.1±9.00 (90) 29.2±3.77 (35) 30.33±6.31 (55) <0.0001 <0.0001 0.34 
    P values 
Characteristic HAPE-p (nHAPE-f (nHLs (nHAPE-p compared with HAPE-f HAPE-p compared with HLs HAPE-f compared with HLs 
Age (years) 29.1±13.86 (250) 27.8±10.4 (210) 39.05±7.44 (430) 0.263 <0.0001 <0.0001 
SBP (mmHg) 124.6±21.1 (250) 119.8±3.5 (210) 121±8.0 (430) 0.001 0.001 0.04 
DBP (mmHg) 82.7±17.0 (250) 80.4±10.0 (210) 81.1±10.0 (430) 0.085 0.122 0.406 
MAP (mmHg) 96.7±16.0 (250) 93.5±10.0 (210) 94.4±10.0 (430) 0.012 0.021 0.285 
PR (rate/min) 92.2±22.9 (250) 80.1±19.9 (210) 84.0±15.7 (430) <0.0001 <0.0001 0.007 
SaO2 levels (%) 71.7±12.5 (250) 91.2±3.7 (210) 89.6±3.2 (430) <0.0001 <0.0001 <0.0001 
PASP (mmHg) 50.1±9.00 (90) 29.2±3.77 (35) 30.33±6.31 (55) <0.0001 <0.0001 0.34 

Genotype and allele distribution

The genotype and allele distributions are shown in Supplementary Table S4 (at http://www.clinsci.org/cs/124/cs1240479add.htm). The three groups were in HWE for the studied SNPs. The SNPs rs1538664 G/A, rs479200 C/T, rs2486729 G/A, rs2790879 T/G, rs480902 T/C, rs2486736 G/A and rs973252 A/G differed significantly between HAPE-p and HAPE-f after adjustment with age, gender and BMI (P=0.00048, 0.008, 0.000038, 0.00017, 0.00011, 0.000035 and 0.000038 respectively). Multivariate logistic regression analysis revealed a significantly higher risk of HAPE in homozygotes rs1538664 AA, rs479200 TT, rs2486729 AA, rs2790879 GG, rs480902 CC, rs2486736 AA and rs973252 GG, with the adjusted risk odds being above 2.0 for these genotypes. As a consequence, the respective alleles were over-represented in HAPE-p (P=0.049, 0.004, 0.02, 0.019, 0.006, 0.022 and 0.013 respectively) and hence were recognized as risk alleles, whereas the alleles that were over-represented in HLs were recognized as protective alleles. A comparison of data for HLs with that of their lowland counterpart HapMap-CHB (Han Chinese in Beijing, China), retrieved from NCBI (National Center for Biotechnology Information), revealed a significant selection of the protective homozygotes rs1538664 GG, rs479200 CC, rs2486729 GG, rs2790879 TT, rs480902 TT, rs2486736 GG and rs973252 AA in HLs (P=7.25×10−12, 1.00×10−6, 4.67×10−11, 3.17×10−12, 2.23×10−12, 2.82×10−12 and 8.15×10−14 respectively).

LD pattern and haplotype distribution

Supplementary Figures S1(a)–S1(c) (at http://www.clinsci.org/cs/124/cs1240479add.htm) show the THM (triangular heat map) examined by Haploview, with Supplementary Figures S1(d)–S1(f) showing the parallel co-ordinate display by textile plot. THM inferred one LD block each in HAPE-f (Supplementary Figure S1a) and HLs (Supplementary Figure S1c) and two blocks in HAPE-p (Supplementary Figure S1b). The textile plot also provided similar results (Supplementary Figures S1d–S1f).

A maximum likelihood analysis by Phase suggested 24 possible haplotypes. Table 2, however, represents the haplotypes which exceeded the cut-off frequency of >2%. Three haplotypes A-T-A-G-C-A-G, G-T-A-G-C-A-G and G-T-G-T-C-G-A were prevalent in HAPE-p with ORs of 1.43, 1.96 and 3.10 when compared with HAPE-f (P=0.042, 0.445 and 0.006 respectively) and with ORs of 5.01, 3.43 and 4.47 when compared with HLs (P=4.72×10−36, 0.002 and 0.00012 respectively), and hence were recognized as risk haplotypes. Similarly, the haplotype G-C-G-T-T-G-A was prevalent in HAPE-f and HLs when compared with HAPE-p, with ORs of 0.65 and 0.15 (P=0.002 and 1.3×10−49 respectively) and hence was recognized as protective haplotypes.

Table 2
Distribution of significant haplotypes of the polymorphisms rs1538664, rs479200, rs2486729, rs2790879, rs480902, rs2486736 and rs973252 of EGLN1 in HAPE-p, HAPE-f and HLs

P values were obtained by multivariate logistic regression analysis and Bonferroni's multiple correction test (P<0.05) after adjustment with age, gender and BMI using SPSS 15.0 software.

 Distribution HAPE-p compared with HAPE-f HAPE-p compared with HLs HAPE-f compared with HLs 
Haplotype HAPE-p (%) HAPE-f (%) HLs (%) OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value 
A-T-A-G-C-A-G 59.0 50.0 22.0 1.43 (1.11–1.86) 0.042 5.01 (3.96–6.54) 4.72×10−36 3.54 (2.79–4.48) 3.60×10−22 
G-T-A-G-C-A-G 5.0 2.6 1.5 1.96 (0.95–4.03) 0.445 3.43 (1.74–6.76) 0.0019 1.75 (0.78–3.94) 0.17 
G-T-G-T-C-G-A 5.0 1.7 1.1 3.10 (1.33–7.25) 0.006 4.47 (2.13–9.39) 0.00012 1.44 (0.54–3.81) 0.459 
G-C-G-T-T-G-A 27.8 37.4 71.4 0.65 (0.49–0.85) 0.002 0.15 (0.12–0.19) 1.33×10−49 0.24 (0.19–0.31) 4.76×10−29 
 Distribution HAPE-p compared with HAPE-f HAPE-p compared with HLs HAPE-f compared with HLs 
Haplotype HAPE-p (%) HAPE-f (%) HLs (%) OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value 
A-T-A-G-C-A-G 59.0 50.0 22.0 1.43 (1.11–1.86) 0.042 5.01 (3.96–6.54) 4.72×10−36 3.54 (2.79–4.48) 3.60×10−22 
G-T-A-G-C-A-G 5.0 2.6 1.5 1.96 (0.95–4.03) 0.445 3.43 (1.74–6.76) 0.0019 1.75 (0.78–3.94) 0.17 
G-T-G-T-C-G-A 5.0 1.7 1.1 3.10 (1.33–7.25) 0.006 4.47 (2.13–9.39) 0.00012 1.44 (0.54–3.81) 0.459 
G-C-G-T-T-G-A 27.8 37.4 71.4 0.65 (0.49–0.85) 0.002 0.15 (0.12–0.19) 1.33×10−49 0.24 (0.19–0.31) 4.76×10−29 

Multi-dimensional reduction analysis

In HAPE-p compared with HAPE-f, the best model was a two-locus interaction between rs1538664 and rs2486729 (Figure 2A). The GA–GA interaction was categorized as being high-risk (P=0.028) and GG–GG as being low-risk (P=1.8×10−5). In HAPE-p compared with HLs, a four-locus model was the best, which included polymorphisms rs1538664, rs479200, rs2486729 and rs480902 (Figure 2B). The interacting genotypes AA-TT-AA-CC and GA-TT-AA-CC of the same polymorphisms were categorized as high-risk and GG-CC-GG-TT as low-risk (P<0.0002).

MDR analysis

Figure 2
MDR analysis

Dark-grey shading, high risk; light-grey shading, low risk; no shading/white, no combination. The distribution of genotype interactions between (A) two-locus, rs1538664 and rs2486729, selected as the best MDR model in HAPE-f and HAPE-p and (B) four-locus, rs1538664, rs479200, rs2486729 and rs480902, selected as the best MDR model in HAPE-p and HLs.

Figure 2
MDR analysis

Dark-grey shading, high risk; light-grey shading, low risk; no shading/white, no combination. The distribution of genotype interactions between (A) two-locus, rs1538664 and rs2486729, selected as the best MDR model in HAPE-f and HAPE-p and (B) four-locus, rs1538664, rs479200, rs2486729 and rs480902, selected as the best MDR model in HAPE-p and HLs.

Gene expression

The expression of EGLN1 was up-regulated 4.55-fold in HAPE-p (P=0.0084) and 1.50-fold in HLs compared with HAPE-f (Figure 1C).

Influence of genotypes on expression

EGLN1 expression was compared with the genotypes of the seven polymorphisms (Supplementary Figure S2 at http://www.clinsci.org/cs/124/cs1240479add.htm). The results show that the risk genotypes, when compared with the protective genotypes, were associated with elevated EGLN1 expression (P<0.05). Of significance, the risk genotype rs1538664 AA was associated with a 3.5- and 2.0-fold up-regulation of EGLN1 expression in HAPE-p and HLs respectively (P=0.01 and 0.04 respectively). The risk genotype rs479200 TT was associated with a 3.4-, 2.1- and 3.2-fold higher expression of EGLN1 in HAPE-p, HAPE-f and HLs (P=0.01, 0.05 and 0.00029 respectively). Likewise, the rs2486729 AA genotype was significantly associated with a 2.5-, 4.1- and 2.2-fold increase in EGLN1 expression in HAPE-f, HAPE-p and HLs (P=0.03, 0.008 and 0.04 respectively). The risk genotype rs2790879 GG was associated with a 3.1-fold up-regulation of EGLN1 in HAPE-p (P=0.001). However, the risk genotypes rs2486736 AA, rs480902 CC and rs973252 GG were associated with a 3.2-, 2.9- and 3.3-fold increase in EGLN1 expression in HLs (P=0.001, 0.04 and 0.00052 respectively).

Influence of haplotypes on expression

The susceptible haplotypes G-T-G-T-C-G-A, G-T-A-G-C-A-G and A-T-A-G-C-A-G when compared with the protective haplotype G-C-G-T-T-G-A were associated with elevated EGLN1 expression by 1.7-, 1.5- and 2.2-fold in HAPE-f (P=0.23, 0.30 and 0.026 respectively), by 2.6-, 2.5- and 3.9-fold in HAPE-p (P=0.04, 0.04 and 0.003 respectively), and by 2.9-, 1.9- and 3.1-fold in HLs (P=0.031, 0.10, and 0.021 respectively) (Figure 3A).

Relative expression of EGLN1 against haplotypes and interacting genotypes
Figure 3
Relative expression of EGLN1 against haplotypes and interacting genotypes

(A) The A-T-A-G-C-A-G, G-T-A-G-C-A-G and G-T-G-T-C-G-A haplotypes compared with the protective haplotype G-C-G-T-T-G-A in HAPE-f, HAPE-p and HLs associate with up-regulated EGLN1 expression. Association of EGLN1 expression in individuals bearing the (B) two-locus model by MDR between rs1538664 and rs2486729 in HAPE-f and HAPE-p and (C) four-locus model between rs1538664, rs479200, rs2486729 and rs480902 in HAPE-p and HLs. *P≤0.05.

Figure 3
Relative expression of EGLN1 against haplotypes and interacting genotypes

(A) The A-T-A-G-C-A-G, G-T-A-G-C-A-G and G-T-G-T-C-G-A haplotypes compared with the protective haplotype G-C-G-T-T-G-A in HAPE-f, HAPE-p and HLs associate with up-regulated EGLN1 expression. Association of EGLN1 expression in individuals bearing the (B) two-locus model by MDR between rs1538664 and rs2486729 in HAPE-f and HAPE-p and (C) four-locus model between rs1538664, rs479200, rs2486729 and rs480902 in HAPE-p and HLs. *P≤0.05.

Influence of interacting genotypes on expression

The two-locus risk-associated interacting genotype AA–AA of rs1538664 and rs2486729 was observed only in HAPE-p and was associated with 2.2-fold higher EGLN1 expression when compared with its protective interacting genotype GG–GG (P=0.0048). GA–GA of the same polymorphisms, prevalent in HAPE-p, was associated with a 1.7- and 1.2-fold higher EGLN1 expression in HAPE-p and HAPE-f respectively, when compared with GG–GG (Figure 3B). The four-locus risk-associated interacting genotypes GA–TT–AA–CC and AA–TT–AA–CC from the polymorphisms rs1538664, rs479200, rs2486729 and rs480902 respectively, were associated with higher EGLN1 expression in HAPE-p and HLs when compared with the protective interacting genotype GG–CC–GG–TT (P<0.05; Figure 3C).

Correlation of SaO2 levels with PASP and EGLN1 expression

SaO2 levels were inversely correlated with both PASP and EGLN1 expression. However, the correlation between SaO2 levels and PASP was significant in HAPE-p and HLs (P=0.019 and 0.038; Figure 1D), whereas it was significant only in HAPE-p with respect to EGLN1 expression (P 0.027; Figure 1E). The same trend was also observed in the other two groups but did not reach statistical significance.

Contribution of EGLN1 risk alleles and major haplotypes to SaO2 levels

Regression analysis was performed to determine the association of risk alleles and major haplotypes of EGLN1 with SaO2 levels. The results were statistically significant even after Bonferroni multiple correction (Table 3). The regression coefficient showed an association of the risk alleles with decreased SaO2 levels in HAPE-p (P=0.0042, 0.0003, 8.35×10−6, 0.0001, 0.0002, 3.86×10−5 and 8.01×10−5 respectively). Even the two control groups, HAPE-f and HLs, had a significant negative correlation of the risk alleles with SaO2 levels (P<0.021; Table 3). The correlation of SaO2 levels with haplotypes was equally noteworthy as can be seen from Table 3. The susceptible haplotype A-T-A-G-C-A-G had a significant association with decreased SaO2 levels in HAPE-p and HLs (P=0.0026 and 0.0067 respectively). The protective haplotype G-C-G-T-T-G-A was positively and significantly correlated with SaO2 levels in HAPE-p, HAPE-f and HLs (P=0.0022, 0.0024 and 0.037 respectively).

Table 3
Relationship between EGLN1 genotypes and major haplotypes and SaO2 levels

P values were obtained by multinomial logistic regression analysis using SPSS 15.0 and Bonferroni multiple correction test. β, regression coefficient

  HAPE-p HAPE-f HLs 
EGLN1 polymorphism Risk allele/major haplotype β P value β P value β P value 
rs1538664 −0.125 0.0042 −0.167 0.0005 −0.097 0.004 
rs479200 −0.157 0.0003 −0.150 0.0019 −0.082 0.015 
rs2486729 −0.194 8.35×10−6 −0.137 0.0046 −0.094 0.006 
rs2790879 −0.165 0.00015 −0.149 0.002 −0.083 0.013 
rs480902 −0.161 0.00022 −0.111 0.021 −0.078 0.020 
rs2486736 −0.179 3.86×10−5 −0.129 0.0076 −0.081 0.017 
rs973252 −0.172 8.01×10−5 −0.129 0.0076 −0.083 0.013 
        
rs1538664-rs479200- rs2486729-rs2790879- rs480902-rs2486736- rs973252 A-T-A-G-C-A-G −0.155 0.0026 −0.007 0.875 −0.111 0.00675 
 G-T-A-G-C-A-G −0.064 0.142 −0.074 0.124 −0.049 0.155 
 G-T-G-T-C-G-A −0.033 0.444 −0.081 0.09 −0.009 0.770 
 G-C-G-T-T-G-A 0.157 0.0022 0.173 0.0024 0.071 0.037 
  HAPE-p HAPE-f HLs 
EGLN1 polymorphism Risk allele/major haplotype β P value β P value β P value 
rs1538664 −0.125 0.0042 −0.167 0.0005 −0.097 0.004 
rs479200 −0.157 0.0003 −0.150 0.0019 −0.082 0.015 
rs2486729 −0.194 8.35×10−6 −0.137 0.0046 −0.094 0.006 
rs2790879 −0.165 0.00015 −0.149 0.002 −0.083 0.013 
rs480902 −0.161 0.00022 −0.111 0.021 −0.078 0.020 
rs2486736 −0.179 3.86×10−5 −0.129 0.0076 −0.081 0.017 
rs973252 −0.172 8.01×10−5 −0.129 0.0076 −0.083 0.013 
        
rs1538664-rs479200- rs2486729-rs2790879- rs480902-rs2486736- rs973252 A-T-A-G-C-A-G −0.155 0.0026 −0.007 0.875 −0.111 0.00675 
 G-T-A-G-C-A-G −0.064 0.142 −0.074 0.124 −0.049 0.155 
 G-T-G-T-C-G-A −0.033 0.444 −0.081 0.09 −0.009 0.770 
 G-C-G-T-T-G-A 0.157 0.0022 0.173 0.0024 0.071 0.037 

Influence of variants on EGLN1 regulation through TFs

Figure 4(A) is a schematic representation of EGLN1 with the 30 polymorphisms studied, but highlighting only the seven that associated with adaptation and mal-adaptation. Figures 4(B) and 4(C) show the protective and risk alleles respectively, the TF-binding sites and TFs on a given site. Figure 4(D) provides the TF-binding sites, their respective TFs, AS (affinity score) and biological functions of the TFs. The analysis suggested that the appearance of rs1538664 A allele in place of rs1538664 G allele created a site for C-MYB (AS=85.6), whereas the change of allele from C to T of rs479200 and from G to A of rs2486729 obliterated the sites for CREB (cAMP-response-element-binding protein), VBP (VHL-binding protein) and HSF (heat-shock factor) respectively. The susceptible rs2790879 G allele replaced the site for NIT2 (nitrilase family member 2) and created the site allowing the binding of GATA-3, ADR1 (alcohol dehydrogenase II synthesis regulator 1), BCD (bicoid) and GATA-1 (AS=89.1, 87.7, 87.3 and 86.5 respectively). Likewise, the rs2486736 A allele, prevalent in HAPE-p, replaced the sites for AP-1 (activator protein-1) and CAP (adenylate cyclase-associated protein), and formed the sites for GATA-1, CAP and GATA-3 (AS=91.4, 86.1 and 85.6 respectively). However, the replacement of allele from T to C in rs480902 and A to G in rs973252 did not alter the TFs.

Schematic representation of polymorphisms of EGLN1 lying in the TF-binding sites

Figure 4
Schematic representation of polymorphisms of EGLN1 lying in the TF-binding sites

(A) EGLN1 structure with the 30 polymorphisms. (B) The protective alleles of the seven polymorphisms lying in the TF-binding sites of the respective TFs. (C) The risk alleles lying in the TF-binding sites of the respective TFs. (D) The affinity score and function of the TFs that bind to the TF-binding sites in the vicinity of the EGLN1 polymorphisms.

Figure 4
Schematic representation of polymorphisms of EGLN1 lying in the TF-binding sites

(A) EGLN1 structure with the 30 polymorphisms. (B) The protective alleles of the seven polymorphisms lying in the TF-binding sites of the respective TFs. (C) The risk alleles lying in the TF-binding sites of the respective TFs. (D) The affinity score and function of the TFs that bind to the TF-binding sites in the vicinity of the EGLN1 polymorphisms.

DISCUSSION

The results of the present study provide detailed information about EGLN1 under hypobaric hypoxia that might help in uncovering the role of this gene in HAPE and HA adaptation. Multivariate logistic regression analysis revealed the over-representation of the genotypes rs1538664 AA, rs479200 TT, rs2486729 AA, rs2790879 GG, rs480902 CC, rs2486736 AA and rs973252 GG in HAPE. In accordance was the over-representation of the risk alleles bearing haplotypes A-T-A-G-C-A-G, G-T-A-G-C-A-G and G-T-G-T-C-G-A in HAPE-p and the protective alleles bearing haplotype G-C-G-T-T-G-A in HAPE-f and HLs. The prevalence of risk genotypes and haplotypes in HAPE-p and their under-representation in HLs and HAPE-f supported the hypothesis of the involvement of common genetic factors in adaptation and mal-adaptation. As the distribution of genotypes and haplotypes was encouraging, we searched for the interacting genotypes, as these interactions could be key determinants to disease susceptibility. The MDR analysis was in agreement with the results of the individual genotype and haplotype because the interactions of the risk genotypes were over-represented in HAPE-p. Furthermore, the MDR analysis also pointed to the fact that the majority of the loci were associated with susceptibility to the stressful environment, whereas few were involved in the adaptation process. Among these polymorphisms, rs1538664 was omnipresent, rs2486729 and rs479200 found their role in HAPE pathophysiology, whereas rs480902 and rs479200 were involved in HA adaptation. With the exception of our previous study [13], there is a lack of information in the literature on susceptible alleles, as most studies have focused on adaptation [815].

Equally pertinent has been the assessment of EGLN1 expression, PASP and SaO2 levels in the three groups. The more than 4-fold up-regulation of EGLN1 in the HAPE-p strongly suggested that the gene was associated with HAPE pathophysiology. The up-regulation essentially caused the dysfunction of HIF-1α which, in turn, prohibited the downstream genes from maintaining cellular oxygen homoeostasis. The overexpression of EGLN1 has been reported to associate with tumour aggressiveness [26], whereas attenuation of EGLN1 restored tumour oxygenation and hence endothelial normalization [27]. Of interest, a slight increase in EGLN1 expression was also observed in HLs, which is attributed to the fact that this population is continually under stress. Studies on the performance of this gene under the hypobaric hypoxia of HA have not been reported, as the focus has been restricted to genetic studies [815]. Likewise, it is necessary to highlight that the two control groups in the present study had mean PASP levels of ~30 mmHg compared with the generally reported ~25 mmHg in healthy subjects under normoxia [28]. In agreement with our findings, however, elevated PASP levels have also been reported in HA populations [18,29,30]. It is important to add that we have consistently observed elevated levels of circulating biomarkers of oxidative stress and vascular homoeostasis in HLs [19,28]. The overall findings suggest that natural selection, over a period of generations, has incorporated crucial changes in the physiological processes to combat the stressful environment [2830]. Detailed investigations will be needed to validate these findings.

Another important aspect of our present study is the SaO2 level in the three groups. The significantly elevated levels of SaO2 in the two healthy groups, HAPE-f and HLs, compared with HAPE-p suggest higher oxygen availability in the blood of these two groups and would presumably provide these subjects with scope for improved physical performance. As expected, SaO2 levels were negatively correlated with PASP in HAPE-p. SaO2 levels also were negatively correlated with EGLN1 expression in the three groups. As SaO2 levels are a direct indicator of exercise capacity, its negative correlation with EGLN1 expression further strengthens the involvement of up-regulated EGLN1 in HA pathophysiology. As the findings on EGLN1 expression and SaO2 levels were of significance, we investigated their association with the studied genetic variants. We found an association of increased EGLN1 expression and decreased SaO2 levels with the risk genotypes, alleles and haplotypes, indicating the functional consequences of these polymorphisms. Our results suggest that the overall changes and their interactions contribute to HAPE pathophysiology.

Of interest, the associated polymorphisms were located in intron 1 and this suggested that this intron could be involved in the regulation of the gene. Therefore we used TFSEARCH to identify TF-binding sites in intron 1 and also the alterations at these sites due to the presence of variants. TFs regulate the flow of information from DNA to mRNA and hence have a crucial role in maintaining physiological processes. Our analysis revealed that the protective alleles of these polymorphisms had binding sites for TFs such as CREB, VBP, NIT2 and AP-1 [3135]. For example, the rs479200 C allele had the binding site for CREB and VBP. CREB mediates specific responses of the pulmonary circulation to hypoxia [31] and regulates several genes, such as tyrosine hydroxylase that helps in heightening ventilatory response by synthesizing the neurotransmitter dopamine to increase alveolar oxygen [32]. VBP assists in the binding of the VHL tumour suppressor protein to the ODD domain of the EGLN1 protein. This complex helps in proteasomal degradation of EGLN1 [33], ensuring normal functioning of HIF and thereby cellular oxygen homoeostasis [3]. Similarly, the rs2790879 A allele forms a binding site for NIT2, a potential tumour suppressor [34], and the rs2486736 G allele forms the binding site for AP-1, which controls differentiation, proliferation and apoptosis [35]. These TFs potentially influence the growth pattern and angiogenesis in the HLs or even in the sojourners when exposed to the same environment for a longer period.

In contrast with the protective alleles, the appearance of risk alleles in these TF-binding sites either disrupt the existing sites or create new ones. For example, replacement of the rs479200 C allele with T disrupted the VBP site and, as a result, might compromise EGLN1 degradation. This was supported by our observation of significantly up-regulated EGLN1 expression in HAPE patients. Likewise, the variant alleles rs1538664 A, rs2790879 G and rs2486736 A created sites for TFs, including C-MYB, GATA-1, GATA-3, C/EBPδ (CCAAT/enhancer-binding protein δ) and BCD. These TFs are associated with tumour growth, proto-oncogenicity, and inflammatory and oxidative stress responses [3640]. The TFs C-MYB and GATA-1 regulate leukaemogenesis and haematopoiesis respectively [36,37], whereas, C/EBPδ influences the regulation of inflammation and energy metabolism [38]. These TFs can be implicated in HAPE pathophysiology by virtue of increased pulmonary and oxidative stress due to the sudden onset of physiological responses indicated by increased vasoconstrictors [41,28], growth factors [42], free radicals [19] and inflammatory markers [4143] in HAPE.

In conclusion, the results of our present study have provided the selection of seven polymorphisms involved in HA physiology. The prevalence of the risk alleles in HAPE-p and the protective alleles in HLs have provided us with allelic variants at the same locus that are involved in disease and adaptation. EGLN1 expression was increased under the HA environment and was associated with risk genotypes and haplotypes in HAPE. Of note, TFSEARCH has provided further the evidence that the variants in this region altered the TF-binding sites, thereby potentially modifying physiological functions. Similarly, the association of these polymorphisms with SaO2 levels contributed to uncovering the functional relevance of these polymorphisms in hypobaric hypoxic adaptation. The findings of the present study are encouraging; however, they need to be validated through functional studies to determine a causal role. Likewise, the outcome of the TFSEARCH findings may provide insight into gene regulation under a hypobaric hypoxia environment at HA. Replication studies are also needed with larger sample sizes and in different ethnicities to establish the global efficacy of these markers.

FUNDING

This work was supported by the Council of Scientific and Industrial Research, India [Supra-Institutional Project-SIP0006].

AUTHOR CONTRIBUTION

Aastha Mishra performed the study, analysed the data and wrote the paper; Ghulam Mohammad and Tashi Thinlas performed the clinical study, collected the blood samples and discussed the data; and Qadar Pasha designed the study, analysed and discussed the data, and wrote the paper. All authors read and approved the final paper.

We acknowledge the support and encouragement of the Director of the Institute of Genomics and Integrative Biology. We also thank the technical support of the staff at Sonam Norboo Memorial Hospital, Leh and all of the participants.

Abbreviations

     
  • AP-1

    activator protein-1

  •  
  • BCD

    bicoid

  •  
  • BMI

    body mass index

  •  
  • C/EBPδ

    CCAAT/enhancer-binding protein δ

  •  
  • CAP

    adenylate cyclase-associated protein

  •  
  • CI

    confidence interval

  •  
  • CREB

    cAMP-response-element-binding protein

  •  
  • DFD

    Deformed

  •  
  • HA

    high altitude

  •  
  • HAPE

    HA pulmonary oedema

  •  
  • HAPE-f

    HAPE-free controls

  •  
  • HAPE-p

    HAPE-patients

  •  
  • HIF

    hypoxia-inducible factor

  •  
  • EGLN1

    HIF-prolyl hydroxylase 2

  •  
  • HLs

    healthy Ladakhi highland natives

  •  
  • HSF

    heat-shock factor

  •  
  • HWE

    Hardy–Weinberg equilibrium

  •  
  • LD

    linkage disequilibrium

  •  
  • MDR

    multi-dimensional reduction

  •  
  • NIT2

    nitrilase family member 2

  •  
  • ODD

    oxygen-dependent degradation

  •  
  • OR

    odds ratio

  •  
  • PASP

    pulmonary artery systolic pressure

  •  
  • RN18S1

    18S rRNA

  •  
  • RT

    reverse transcription

  •  
  • SaO2,

    arterial oxygen saturation

  •  
  • SNP

    single nucleotide polymorphism

  •  
  • TF

    transcription factor

  •  
  • THM

    triangular heat map

  •  
  • VHL

    von Hippel–Lindau

  •  
  • VBP

    VHL-binding protein

References

References
1
Hohenhaus
E.
Paul
A.
McCullough
R. E.
Kucherer
H.
Bartsch
P.
Ventilatory and pulmonary vascular response to hypoxia and susceptibility to high altitude pulmonary oedema
Eur. Respir. J.
1995
, vol. 
8
 (pg. 
1825
-
1833
)
2
Smith
T. G.
Robbins
P. A.
Ratcliffe
P. J.
The human side of hypoxia-inducible factor
Br. J. Haematol.
2008
, vol. 
141
 (pg. 
325
-
334
)
3
Semenza
G. L.
Regulation of oxygen homeostasis by hypoxia-inducible factor 1
Physiology
2009
, vol. 
24
 (pg. 
97
-
106
)
4
van Patot
M. C.
Gassmann
M.
Hypoxia: adapting to high altitude by mutating EPAS-1, the gene encoding HIF-2α
High Alt. Med. Biol.
2011
, vol. 
12
 (pg. 
157
-
167
)
5
Berra
E
Benizri
E
GinouveÁs
A
Volmat
V
Roux
D.
Pouysseagur
J
HIF prolyl-hydroxylase 2 is the key oxygen sensor setting low steady-state levels of HIF-1a in normoxia
EMBO J.
2003
, vol. 
22
 (pg. 
4082
-
4090
)
6
To
K. K. W.
Huang
L. E.
Suppression of HIF-1α transcriptional activity by the HIF prolyl hydroxylase EGLN1
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
38102
-
38107
)
7
Del Peso
L.
Castellanos
M. C.
Temes
E.
Martin-Puig
S.
Cuevas
Y.
Olmos
G.
Landazuri
M. O.
von Hippel Lindau/hypoxia-inducible factor (HIF) pathway regulates the transcription of the HIF-proline hydroxylase genes in response to low oxygen
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
48690
-
48695
)
8
Ji
L. D.
Qiu
Y. Q.
Xu
J.
Irwin
D. M.
Tam
S. C.
Tang
N. L.
Zhang
Y. P.
Genetic adaptation of the hypoxia-inducible factor pathway to oxygen pressure among eurasian human populations
Mol. Biol. Evol.
2012
, vol. 
29
 (pg. 
3359
-
3370
)
9
Peng
Y.
Yang
Z.
Zhang
H.
Cui
C.
Qi
X.
Luo
X.
Tao
X.
Wu
T.
Ouzhuluobu
Basang
, et al. 
Genetic variations in Tibetan populations and high-altitude adaptation at the Himalayas
Mol. Biol. Evol.
2011
, vol. 
28
 (pg. 
1075
-
1081
)
10
Pagani
L.
Ayub
Q.
MacArthur
D. G.
Xue
Y.
Baillie
J. K.
Chen
Y.
Kozarewa
I.
Turner
D. J.
Tofanelli
S.
Bulayeva
K.
, et al. 
High altitude adaptation in Daghestani populations from the Caucasus
Hum. Genet.
2012
, vol. 
131
 (pg. 
423
-
33
)
11
Bigham
A.
Bauchet
M.
Pinto
D.
Mao
X.
Akey
J. M.
Mei
R.
Scherer
S. W.
Julian
C. G.
Wilson
M. J.
López Herráez
D.
, et al. 
Identifying signatures of natural selection in Tibetan and Andean populations using dense genome scan data
PLoS Genet.
2010
, vol. 
6
 pg. 
e1001116
 
12
Xu
S.
Li
S.
Yang
Y.
Tan
J.
Lou
H.
Jin
W.
Yang
L.
Pan
X.
Wang
J.
Shen
Y.
, et al. 
A genome-wide search for signals of high-altitude adaptation in Tibetans
Mol. Biol. Evol.
2011
, vol. 
28
 (pg. 
1003
-
1011
)
13
Aggarwal
S.
Negi
S.
Jha
P.
Singh
P. K.
Stobdan
T.
Pasha
M. A. Q.
Ghosh
S.
Agrawal
A.
Indian Genome Variation Consortium
Prasher
B.
Mukerji
M.
EGLN1 involvement in high-altitude adaptation revealed through genetic analysis of extreme constitution types defined in Ayurveda
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
18961
-
18966
)
14
Simonson
T. S.
Yang
Y.
Huff
C. D.
Yun
H.
Qin
G.
Witherspoon
D. J.
Bai
Z.
Lorenzo
F. R.
Xing
J.
Jorde
L. B.
, et al. 
Genetic evidence for high-altitude adaptation in Tibet
Science
2010
, vol. 
329
 (pg. 
72
-
75
)
15
Wang
B.
Zhang
Y. B.
Zhang
F.
Lin
H.
Wang
X.
Wan
N.
Ye
Z.
Weng
H.
Zhang
L.
Li
X.
, et al. 
On the origin of Tibetans and their genetic basis in adapting high-altitude environments
PLoS ONE
2011
, vol. 
6
 pg. 
e17002
 
16
West
J. B.
The physiologic basis of high-altitude diseases
Ann. Intern. Med.
2004
, vol. 
141
 (pg. 
789
-
800
)
17
Bartsch
P.
Mairbaurl
H.
Maggiorini
M.
Swenson
E. R.
Physiological aspects of high-altitude pulmonary edema
J. Appl. Physiol.
2005
, vol. 
98
 (pg. 
1101
-
1110
)
18
Allemann
y.
Sartori
C.
Lepori
M.
Pierre
S.
Melot
C.
Naeije
R.
Scherrer
U.
Maggiorini
M.
Echocardiographic and invasive measurements of pulmonary artery pressure correlate closely at high altitude
Am. J. Physiol. Heart Circ. Physiol.
2000
, vol. 
279
 (pg. 
H2013
-
H2016
)
19
Mishra
A.
Ali
Z.
Vibhuti
A.
Kumar
R.
Alam
P.
Ram
R.
Thinlas
T.
Mohammad
G.
Pasha
M. A. Q.
CYBA and GSTP1 variants associate with oxidative stress under hypobaric hypoxia as observed in high-altitude pulmonary oedema
Clin. Sci.
2012
, vol. 
122
 (pg. 
299
-
309
)
20
Ahsan
A.
Mohammad
G.
Norboo
T.
Baig
M. A.
Pasha
M. A. Q.
Heterozygotes of NOS3 polymorphisms contribute to reduced nitrogen oxides in high-altitude pulmonary edema
Chest
2006
, vol. 
130
 (pg. 
1511
-
1519
)
21
Stephens
M.
Smith
N. J.
Donnelly
P.
A new statistical method for haplotype reconstruction from population data
Am. J. Hum. Genet.
2001
, vol. 
68
 (pg. 
978
-
989
)
22
Barrett
J. C.
Fry
B.
Maller
J.
Daly
M. J.
Haploview: analysis and visualization of LD and haplotype maps
Bioinformatics
2005
, vol. 
21
 (pg. 
263
-
265
)
23
Kumasaka
N.
Nakamura
Y.
Kamatani
N.
The textile plot: a new linkage disequilibrium display of multiple-single nucleotide polymorphism genotype data
PLoS ONE
2010
, vol. 
5
 pg. 
e10207
 
24
Mourad
R.
Sinoquet
C.
Dina
C.
Leray
P.
Visualization of pairwise and multilocus linkage disequilibrium structure using latent forests
PLoS ONE
2011
, vol. 
6
 pg. 
e27320
 
25
Hahn
L. W.
Ritchie
M. D.
Moore
J. H.
Multifactor dimensionality reduction software for detecting gene-gene and gene environment interactions
Bioinformatics
2003
, vol. 
19
 (pg. 
376
-
382
)
26
Couvelard
A.
Deschamps
L.
Rebours
V.
Sauvanet
A.
Gatter
K.
Pezzella
F.
Ruszniewski
P.
Bedossa
P
Overexpression of the oxygen sensors PHD-1, PHD-2, PHD-3, and FIH Is associated with tumor aggressiveness in pancreatic endocrine tumors
Clin. Cancer. Res.
2008
, vol. 
14
 (pg. 
6634
-
6639
)
27
Mazzone
M.
Dettori
D.
Leite de Oliveira
R.
Loges
S.
Schmidt
T.
Jonckx
B.
Tian
Y. M.
Lanahan
A. A.
Pollard
P.
Ruiz de Almodovar
C.
, et al. 
Heterozygous deficiency of PHD2 restores tumor oxygenation and inhibits metastasis via endothelial normalization
Cell
2009
, vol. 
136
 (pg. 
839
-
851
)
28
Ali
Z.
Mishra
A.
Kumar
R.
Alam
P.
Pandey
P.
Ram
R.
Thinlas
T.
Mohammad
G.
Pasha
M. A. Q.
Interactions among vascular-tone modulators contribute to high altitude pulmonary edema and augmented vasoreactivity in highlanders
PLoS ONE
2012
, vol. 
7
 pg. 
e44049
 
29
Dehnert
C.
Grunig
E.
Mereles
D.
von Lennep
N.
Bartsch
P.
Identification of individuals susceptible to high-altitude pulmonary oedema at low altitude
Eur. Respir. J.
2005
, vol. 
25
 (pg. 
545
-
551
)
30
Kojonazarov
B. K.
Imanov
B. Z.
Amatov
T. A.
Mirrakhimov
M. M.
Naeije
R.
Wilkins
M. R.
Aldashev
A. A.
Noninvasive and invasive evaluation of pulmonary arterial pressure in highlanders
Eur. Respir. J.
2007
, vol. 
29
 (pg. 
352
-
356
)
31
Leonard
M. O.
Howell
K.
Madden
S. F.
Costello
C. M.
Higgins
D. G.
Taylor
C. T.
McLoughlin
P.
Hypoxia selectively activates the CREB family of transcription factors in the in vivo lung
Am. J. Respir. Crit. Care Med.
2008
, vol. 
178
 (pg. 
977
-
983
)
32
Kim
K. S.
Lee
M. K.
Carroll
J.
Joh
T. H.
Both the basal and inducible transcription of the tyrosine hydroxylase gene are dependent upon a cAMP response element
J. Biol. Chem.
1993
, vol. 
268
 (pg. 
15689
-
15695
)
33
Tsuchiya
H.
Iseda
T.
Hino
O.
Identification of a novel protein (VBP-1) binding to the von Hippel–Lindau (VHL) tumor suppressor gene product
Cancer Res.
1996
, vol. 
56
 (pg. 
2881
-
2885
)
34
Lin
C. H.
Chung
M. Y.
Chen
W. B.
Chien
C. H.
Growth inhibitory effect of the human NIT2 gene and its allelic imbalance in cancers
FEBS J.
2007
, vol. 
274
 (pg. 
2946
-
2956
)
35
Ameyar
M.
Wisniewska
M.
Weitzman
J. B.
A role for AP-1 in apoptosis: the case for and against
Biochimie
2003
, vol. 
85
 (pg. 
747
-
752
)
36
Ramsay
R. G.
Gonda
T. J.
MYB function in normal and cancer cells
Nat. Rev. Cancer
2008
, vol. 
8
 (pg. 
523
-
534
)
37
Ferreira
R.
Ohneda
K.
Yamamoto
M.
Philipsen
S.
GATA1 function, a paradigm for transcription factors in hematopoiesis
Mol. Cell Biol.
2005
, vol. 
25
 (pg. 
1215
-
1227
)
38
Cardinaux
J. R.
Allaman
I.
Magistretti
P. J.
Pro-inflammatory cytokines induce the transcription factors C/EBPβ and C/EBPδ in astrocytes
Glia
2000
, vol. 
29
 (pg. 
91
-
97
)
39
Usary
J.
Llaca
V.
Karaca
G.
Presswala
S.
Karaca
M.
He
X.
Langerød
A.
Kåresen
R.
Oh
D. S.
Dressler
L. G.
, et al. 
Mutation of GATA3 in human breast tumors
Oncogene
2004
, vol. 
23
 (pg. 
7669
-
7678
)
40
El Rouby
S.
Newcomb
E. W.
Identification of Bcd, a novel proto-oncogene expressed in B-cells
Oncogene
1996
, vol. 
13
 (pg. 
2623
-
2630
)
41
Droma
Y.
Hayano
T.
Takabayashi
Y.
Koizumi
T.
Kubo
K.
Kobayashi
T.
Sekiguchi
M.
Endothelin-1 and interleukin-8 in high altitude pulmonary oedema
Eur. Respir. J.
1996
, vol. 
9
 (pg. 
1947
-
1949
)
42
Pavlicek
V.
Marti
H. H.
Grad
S.
Gibbs
J. S.
Kol
C.
Wenger
R. H.
Gassmann
M.
Kohl
J.
Maly
F. E.
Oelz
O.
, et al. 
Effects of hypobaric hypoxia on vascular endothelial growth factor and the acute phase response in subjects who are susceptible to high-altitude pulmonary oedema
Eur. J. Appl. Physiol.
2000
, vol. 
81
 (pg. 
497
-
503
)
43
Kubo
K.
Hanaoka
M.
Yamaguchi
S.
Hayano
T.
Hayasaka
M.
Koizumi
T.
Fujimoto
K.
Kobayashi
T.
Honda
T.
Cytokines in bronchoalveolar lavage fluid in patients with high altitude pulmonary oedema at moderate altitude in Japan
Thorax
1996
, vol. 
51
 (pg. 
739
-
742
)
44
Cook
W. J.
Chase
D.
Audino
D. C.
Denis
C. L.
Dissection of the ADR1 protein reveals multiple, functionally redundant activation domains interspersed with inhibitory regions: evidence for a repressor binding to the ADR1c region
Mol. Cell Biol.
1994
, vol. 
14
 (pg. 
629
-
640
)
45
Wu
C.
Heat shock transcription factors: structure and regulation
Annu. Rev. Cell Dev. Biol.
1995
, vol. 
11
 (pg. 
441
-
469
)
46
Schultz
S. C.
Shields
G. C.
Steitz
T. A.
Crystal structure of a CAP-DNA complex: the DNA is bent by 90 degrees
Science
1991
, vol. 
253
 (pg. 
1001
-
1007
)
47
Hinnebusch
A. G.
Translational regulation of GCN4 and the general amino acid control of yeast
Annu. Rev. Microbiol.
2005
, vol. 
59
 (pg. 
407
-
450
)
48
Tsuzuki
S.
Sekiguchi
S.
Hayakawa
Y.
Regulation of growth-blocking peptide expression during embryogenesis of the cabbage armyworm
Biochem. Biophys. Res. Commun.
2005
, vol. 
335
 (pg. 
1078
-
1084
)

Supplementary data