Identification of gene signatures and potential pharmaceutical candidates linked to COVID-19-related depression based on gene expression profiles

Acute and long-term mental health disorders correlate with coronavirus disease 2019 (COVID-19). The underlying mechanisms responsible for the coexistence of COVID-19 and depression remain unclear, and more research is needed to find hub genes and effective therapies. The main objective of this study was to evaluate gene-expression profiles and, identify key genes, and discovery potential therapeutic agents for co-occurrence in COVID-19 and major depressive disorder (MDD).

Initially, we identified differentially expressed genes (DEGs) in datasets from COVID-19 (GSE188847) or MDD (GSE101521). Subsequently, we employed two machine learning analyses-least absolute shrinkage and selection operator (LASSO) and random forest algorithms- to pinpoint shared hub gene between the two diseases. Furthermore, the LINCS L1000 characteristic direction signatures search engine (L1000CDS2) was utilized for drug repurposing studies based on the gene-expression signatures. Finally, molecular dynamics (MD) simulations were conducted to investigate the binding interactions between molecules and the target proteins.

We uncovered 60 DEGs that overlapped between the two datasets but exhibited distinct patterns of expression in each dataset. Subsequent machine learning analyses revealed EMILIN3, OPA3, and TFCP2 as potential shared hub genes underlying both diseases. Furthermore, L1000CDS2 analysis indicated that trichostatin A (TSA), a metabolite derived from Streptomyces, could potentially reverse the altered gene expression. Molecular docking and molecular dynamics simulations revealed that complexes of TSA-perturbed protein spontaneously form and are highly stable.

EMILIN3, OPA3, and TFCP2 are likely to be potential shared hub genes in both COVID-19 and depression. Meanwhile, TSA may serve as a therapeutic option for treating COVID-19-associated depression. Given the inherent constraints of computational modeling, further biological validation studies would help establish the significance of these preliminary findings.