Identifying a gene signature for age-related hearing loss through machine learning and revealing the effect of the CTSS on the mice cochlea

Abstract

Age-related hearing loss (ARHL) is one of the most common health conditions among the elderly population. This study used machine learning to screen for a gene signature to predicts ARHL. Four ARHL mice cochlear transcriptome datasets and the mRNA sequencing of C57BL/6J mice were used for analysis. Machine learning was used to screen for gene signatures closely related to ARHL and validate them. Via qPCR, immunohistochemistry, and immunofluorescence confocal microscopy were used to assess the effect of key gene on the cochlea. The gene signature consisting of 38 genes constructed via Stepglm [forwards] had the best accuracy in the training group, with excellent accuracy and recall in the training and testing groups in predicting ARHL. The gene signature reflected active immune function. CTSS was selected as a key gene on the basis of its association with age and influence hearing loss severity. CTSS showed high expression in ARHL and enriched in the cochlear stria vascularis, which is significantly positively correlated with macrophage marker CD68 expression (R = 0.74, p = 0.006). The gene signature has good accuracy in predicting ARHL. CTSS is highly expressed in the cochleae of ARHL mice and may promote ARHL by inducing macrophage enrichment and causing low-grade inflammation.

Keywords: Age-related hearing loss; CTSS; Low-grade inflammation; Machine learning.