Predicting mechanical ventilation effects on six human tissue transcriptomes

Multi-tissue transcriptional changes and core circadian clock disruption following intensive care

Objective: Both critical illness and current care have been hypothesized to upset daily rhythms and impair molecular circadian function. However, the influence of critical illness on clock function in different tissues and on circadian output genes are unknown. Here we evaluate the effect of critical care and illness on transcription, focusing on the functional organization of the core circadian oscillator. Methods: We downloaded RNAseq count data from the Genotype-Tissue Expression (GTEx) project. Treating mechanical ventilation as a marker for intensive care, we stratified samples into acute death (AD) and intensive care (IC) groups based on the documented Hardy Death Scale. We restricted our analysis to the 25 tissues with >50 samples in each group. Using the edgeR package and controlling for collection center, gender, and age, we identified transcripts differentially expressed between the AD and IC groups. Overrepresentation and enrichment methods were used to identify gene sets modulated by intensive care across tissues. For each tissue, we then calculated the delta clock correlation distance (ΔCCD), a comparative measure of the functional organization of the core circadian oscillator, in the both the AD and IC groups. The statistical significance of the ΔCCD was assessed by permutation, modifying a pre-existing R package to control for confounding variables. Results: Intensive care, as marked by ventilation, significantly modulated the expression of thousands of genes. Transcripts that were modulated in ≥75% of tissues were enriched for genes involved in mitochondrial energetics, cellular stress, metabolism, and notably circadian regulation. Transcripts that were more markedly affected, in ≥10 tissues, were enriched for inflammation, complement and immune pathways. Oscillator organization, as assessed by ΔCCD, was significantly reduced in the intensive care group in 11/25 tissues. Conclusion: Our findings support the hypothesis that patients in intensive care have impaired molecular circadian rhythms. Tissues involved in metabolism and energetics demonstrated the most marked changes in oscillator organization. In adipose tissue, there was a significant overlap between transcripts previously established to be modulated by sleep deprivation and fasting with those modulated by critical care. This work suggests that intensive care protocols that restore sleep/wake and nutritional rhythms may be of benefit.