Background: Lung aging is characterized by a number of structural alterations including fibrosis, chronic inflammation and the alteration of inflammatory cell composition. Chronic exposure to cigarette smoke (CS) is known to induce similar alterations and may contribute to premature lung aging. Additionally, aging and CS exposure are associated with transcriptional alterations in the lung. The current work aims to explore the interaction between age- and CS- associated transcriptomic perturbations and develop a transcriptomic clock able to predict the biological age and the impact of external factors on lung aging.
Results: Our investigations revealed a substantial overlap between transcriptomic response to CS exposure and age-related transcriptomic alterations in the murine lung. Of particular interest is the strong upregulation of immunoglobulin genes with increased age and in response to CS exposure, indicating an important implication of B-cells in lung inflammation associated with aging and smoking. Furthermore, we used a machine learning approach based on Lasso regression to build a transcriptomic age model that can accurately predict chronological age in untreated mice and the deviations associated with certain exposures. Interestingly, CS-exposed-mice were predicted to be prematurely aged in contrast to mice exposed to fresh air or to heated tobacco products (HTPs). The accelerated aging rate associated with CS was reversed upon smoking cessation or switching to HTPs. Additionally, our model was able to predict premature aging associated with thoracic irradiation from an independent public dataset.
Conclusions: Aging and CS exposure share common transcriptional alteration patterns in the murine lung. The massive upregulation of B-cell restricted genes during these processes shed light on the contribution of cell composition and particularly immune cells to the measured transcriptomic signal. Through machine learning approach, we show that gene expression changes can be used to accurately monitor the biological age and the modulations associated with certain exposures. Our findings also suggest that the premature lung aging is reversible upon the reduction of harmful exposures.
Keywords: Age prediction; Cigarette smoke exposure; Lung inflammation; Machine learning; Transcriptomics.