Adaptive PRKAA1 variant in Andeans is associated with improved ventilation and sleep phenotypes

Abstract

Highland groups have adapted to the extreme selective pressures of hypoxia at high altitude via alterations in the oxygen-transport cascade. PRKAA1, which encodes the catalytic subunit of the AMP-activated protein kinase (AMPK), is a notable target of natural selection in Andeans and has been associated with protective fetal phenotypes in this population. AMPK is a universal cellular energy sensor involved in a multitude of physiological processes, including ventilation and the hypoxic ventilatory response (HVR) in animal models. We localized a signal of positive selection and identified a regulatory promoter variant (rs10035235, C>T) of adaptive significance that is associated with ventilatory and sleep phenotypes in male Andean highlanders as well as sleep phenotypes in publicly available lowland cohorts. This work identifies a functional, adaptive, and likely pleiotropic regulatory variant in PRKAA1 in Andeans that may accentuate hypoxia-induced ventilation and provide protection from sleep-disordered breathing in both high- and lowland populations.

Keywords: Biological sciences; Genetics; Health sciences; Human Genetics; Medical specialty; Medicine; Precision medicine; atural sciences.

Graphical abstract

Figure 1

LocusZoom of CMS results around PRKAA1 plotted against histone modifications in seven cell lines (ENCODE, obtained from UCSC Genome Browser) The left y axis shows CMS_BF score, where CMS_BF = 2ln(Bayes Factor). The right y axis denotes the recombination rate in hundred crossovers per million base pairs (cM/Mb). The top CMS marker within/near PRKAA1 (rs62357564) is indicated by a purple diamond, while other CMS markers are indicated by circles. Linkage disequilibrium (LD) with the top CMS marker is shown on a color-based scale from red to dark blue, indicating complete LD (r² = 1.0) to complete linkage equilibrium (r² = 0.0).

Figure 2

Visualization of putatively causal SNVs rs10035235 and rs3805490 across various epigenetic tracks: histone 3 lysine 27 acetylation (H3K27Ac), histone 3 lysine 4 trimethylation (H3K4Me3), histone 3 lysine 4 methylation (H3K4Me1), ENCODE chromatin states predicted by Hidden Markov Model (HMM), predicted CpG islands, and ENCODE DNase I hypersensitivity clusters in 125 cell types Histone modifications are from ENCODE in seven cell lines: lymphoblastoid cells (GM12878: light red), human umbilical vein endothelial cells (HUVEC: cyan), human skeletal muscle myoblasts (HSMM: light green), normal human lung fibroblasts (NHLF: pink), human immortalized myelogenous leukemia cells (K562: purple), normal human epidermal keratinocytes (NHEK: dark orange), and human embryonic stem cells (H1-hESC: yellow). Chromatin states (HMM) are also obtained from ENCODE in five cell lines: HUVEC, HSMM, human mammary epithelial cells (HMEC), NHLF, and K562. All tracks were obtained directly from UCSC Genome Browser (hg19): http://genome.ucsc.edu.

Figure 3

Allele frequencies and associations with PRKAA1 variant rs10035235 (A–G) Allele frequency of PRKAA1 variant rs10035235 (C>T) in Andean highlanders and 1000 genomes populations and subpopulations (African: AFR, European: EUR, East Asian: EAS, Southeast Asian: SAS, American: AMR, Colombian: CLM, Mexican: MXL, Perúvian: PEL, 1000 genomes phase III). PRKAA1 promoter variant rs10035235 and (B) end-tidal carbon dioxide (PetCO₂) at room air (4,340 m), (C) minute ventilation ($\dot{V_I}$) at 10% inspired O₂, (D) isocapnic hypoxic ventilatory response (HVR), (E) hypercapnic ventilatory response (HCVR), (F) mean O₂ saturation while asleep, and (G) percent time spent asleep with SpO₂ < 80% (TST₈₀) in males (see Figure S2 for females). All measurements were obtained at high altitude (Cerro de Pasco, Perú, 4,340 m) ∗p < 0.05. NS, not significant.

Figure 4

Forest plots of PRKAA1 variant in large publicly available datasets (A and B) PRKAA1 rs10035235(T) associations with (A) time- and (B) frequency-based sleep measurements in the hispanic community health study (HCHS) and the Framingham heart study (FHS) in non-rapid eye movement (NREM) and rapid eye movement (REM) sleep. Only apnea/hypopneas with desaturations ≥3% were considered.

Figure 5

Regulatory potential of rs10035235 (A) Sasquatch DNase I hypersensitivity shoulder-footprint ratio (SFR) scores for rs10035235 (C>T) in HUVEC, HMEC, NHLF, and K562 cells. A more negative $\sum Dmg$ score indicates a greater accessibility and higher probability for TF binding to that locus. (B) HOXA11 TFBS motif from JASPAR over PRKAA1 promoter sequence containing rs10035235 (C>T). Near identical TFBS motifs were observed for HOX[A9, A11, B9, C9, C10, C11, C12, D10, D11, D12] (Figure S3). (C) Luciferase assay expression levels in HEK293T cells untransfected or transfected with the pGL3 basic or pGL3 basic vector + PRKAA1 promoter containing either the wild type (C) or putatively adaptive (T) allele. ∗∗p < 0.008.