Abstract
Given the severity of the SARS-CoV-2 pandemic, a major challenge is to rapidly repurpose existing approved drugs for clinical interventions. While a number of data-driven and experimental approaches have been suggested in the context of drug repurposing, a platform that systematically integrates available transcriptomic, proteomic and structural data is missing. More importantly, given that SARS-CoV-2 pathogenicity is highly age-dependent, it is critical to integrate aging signatures into drug discovery platforms. We here take advantage of large-scale transcriptional drug screens combined with RNA-seq data of the lung epithelium with SARS-CoV-2 infection as well as the aging lung. To identify robust druggable protein targets, we propose a principled causal framework that makes use of multiple data modalities. Our analysis highlights the importance of serine/threonine and tyrosine kinases as potential targets that intersect the SARS-CoV-2 and aging pathways. By integrating transcriptomic, proteomic and structural data that is available for many diseases, our drug discovery platform is broadly applicable. Rigorous in vitro experiments as well as clinical trials are needed to validate the identified candidate drugs.
Introduction
While many efforts have focused on repurposing drugs to treat COVID-19, most rely on matching drug signatures to viral infection patterns alone. However, SARS-CoV-2 pathogenicity is notoriously age-dependent: morbidity and fatality rates are significantly higher in older populations. A successful drug repurposing strategy must account for the "cross-talk" between viral response and the biological signatures of the aging lung. Furthermore, existing methods often fail to integrate multiple data modalities into a principled causal framework that can predict the actual effect of an intervention rather than just a correlation.
Data Integration and Hypothesis
The researchers hypothesized that identifying genes regulated by both SARS-CoV-2 and the aging process would reveal the most effective targets for the most vulnerable populations. The study integrated several large-scale datasets to build this intersection:
- RNA-seq (Infection & Aging): SARS-CoV-2 infected lung cells and human lung tissue samples from the Genotype-Tissue Expression (GTEx) project (comparing 20–29 year olds to 70–79 year olds).
- Drug Signatures: The Connectivity Map (CMap/L1000) database.
- Interactions: IRefIndex (protein-protein interactions) and DrugCentral (drug-target affinity data).
The Drug Discovery Platform
The authors built a multi-stage computational pipeline to bridge the gap between raw data and druggable targets:
- Overparameterized Autoencoders: To utilize the massive CMap database, the authors used a novel autoencoder approach. Unlike standard dimensionality reduction, overparameterizing the network helped align drug signatures across different cell types, allowing them to predict how a drug might behave in lung cells even if it was only tested in a different cell line.
- Steiner Tree Analysis: The intersecting "aging + infection" genes were mapped onto a human protein-protein interaction (PPI) network to find the most efficient subnetwork (the "interactome") connecting these genes.
- Causal Structure Discovery: Finally, the team applied the Greedy Sparsest Permutation (GSP) algorithm to single-cell RNA-seq data. This allowed them to move beyond correlation and identify which proteins are actually upstream of the disease-associated gene expression changes.
Key Insights and Results
The study found 219 genes significantly regulated by both SARS-CoV-2 and aging, often moving in the same detrimental direction. The causal analysis repeatedly pointed toward serine/threonine and tyrosine kinases as the most effective upstream intervention points.
RIPK1 (Receptor-interacting serine/threonine-protein kinase 1) was identified as a central regulator. Known for its role in necroptosis and inflammation, RIPK1 is a key driver of lung damage in severe COVID-19. The platform identified 142 FDA-approved drugs whose signatures align with the reverse of the SARS-CoV-2/aging signature, including protein kinase inhibitors like axitinib, dasatinib, pazopanib, and sunitinib.
Limitations and Future Work
As a computational discovery platform, the identified candidates require extensive in vitro experiments and clinical trials to prove efficacy. The model is currently constrained by the L1000 dataset's focus on ~1,000 landmark genes and relies on statistical assumptions for causal discovery that may be challenged by the complexity of living human lungs. However, the integration of aging data provides a critical filter for identifying drugs most likely to benefit elderly patients.
Summary updated