An integrated multi-omics analysis of sleep-disordered breathing traits across multiple blood cell types

Abstract

ABSTRACTBackgroundSleep Disordered Breathing (SDB) is characterized by repeated breathing reductions or cessations during sleep, often accompanied by oxyhemoglobin desaturation. How SDB affects the molecular environment is still poorly understood.MethodsWe studied the association of three SDB measures: the Apnea Hypopnea Index (AHI), average and minimum oxyhemoglobin saturation during sleep (AvgO2 and MinO2) with gene expression measured using RNA-seq in peripheral blood mononuclear cells (PBMCs), monocytes, and T-cells, in ∼500 individuals from the Multi-Ethnic Study of Atherosclerosis (MESA). We developed genetic instrumental variables (IVs) for the associated transcripts as polygenic risk scores (tPRS), then generalized and validated the tPRS in the Women’s Health Initiative (WHI). Next, we constructed the tPRS and studied their association with SDB measures (to identify potential reverse causal associations) and with serum metabolites (to identify downstream effects) in ∼12,000 and ∼4,000 participants, respectively, from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). Finally, we estimated the association of these SDB measures with transcript IV-associated metabolites in HCHS/SOL, to verify complete association pathways linking SDB, gene expression, and metabolites.ResultsAcross the three leukocyte cell types, 96 gene transcripts were associated with at least one SDB exposure (False Discovery Rate (FDR) p-value <0.1). Across cell populations, estimated log-fold expression changes were similar between AHI and MinO2 (Spearman correlations>0.90), and less similar between AvgO2 and the other exposures. Eight and four associations had FDR p-value<0.05 when the analysis was not adjusted and adjusted to BMI, respectively. Associations include known genes that respond to (PDGFC) and regulate response to (AJUBA) hypoxia. We identified a complete “chain” linking AvgO2, P2RX4, and butyrylcarnitine (C4), suggesting that increased expression of the purinergic receptor P2RX4 may improve average oxyhemoglobin saturation and decrease butyrylcarnitine (C4) levels.ConclusionsOur results support a mechanistic role for purinergic signaling and hypoxic signaling, among others, in SDB. These findings show differential gene expression by blood cell type in relation to SDB traits and link P2XR4 expression to influencing AvgO2 and butyrylcarnitine (C4) levels. Overall, we employed novel methods for integrating multi-omic data to evaluate biological mechanisms underlying multiple SDB traits.