Computational Drug Repurposing for Alzheimer’s Disease Using Risk Genes From GWAS and Single-Cell RNA Sequencing Studies

Yun Xu, Jiming Kong, Pingzhao HuFrontiers in Pharmacology 2021 · DOI

TL;DR

This study leverages large-scale genomic data from GWAS and single-cell RNA sequencing to computationally identify existing drugs that could be repurposed for Alzheimer's Disease (AD). By integrating drug-induced transcriptional profiles with AD-associated risk genes and pathways, they identified eight promising drug candidates, four of which show significant preclinical evidence for AD treatment with moderate toxicity.

What Problem Does This Paper Address?

Alzheimer's Disease (AD) is a devastating neurodegenerative disorder with a complex and not fully understood pathology. Current treatments offer only modest symptomatic relief, highlighting the urgent need for more effective therapeutic strategies. Developing novel drugs is a lengthy and expensive process. Drug repurposing, which involves finding new uses for existing drugs, offers a faster and more cost-effective alternative. However, identifying suitable candidates requires sophisticated methods to connect disease mechanisms with drug actions. This paper addresses the challenge of systematically identifying potential AD drug repurposing candidates by integrating diverse genomic data sources (GWAS and single-cell RNA-seq) with a computational drug repurposing tool.

Methods

The researchers employed a multi-step computational approach:

  1. Gene List Generation: They constructed four distinct gene lists: one from a large-scale GWAS study of late-onset AD and three from single-cell RNA sequencing (scRNA-seq) studies, specifically focusing on differentially expressed genes in excitatory neurons (Ex), astrocytes (Ast), and oligodendrocytes (Oli) associated with AD pathology.
  2. Functional Analysis and Gene Overlap: The generated gene lists were analyzed using Enrichr for functional enrichment and to identify overlaps with known AD drug targets from clinical trials.
  3. Drug Repurposing with Gene2Drug: A key methodological contribution is the use of the Gene2Drug tool. This tool combines drug-induced transcriptional responses with pathway information. It analyzes gene expression data from the Connectivity Map (CMap) to identify drugs that can modulate pathways involving AD risk genes. This approach goes beyond directly targeting genes to also consider drugs that indirectly influence disease-related pathways.
  4. Prioritization and Filtering: Candidate drugs were ranked based on enrichment scores and adjusted p-values. Several filtering steps were applied, including removing databases that yielded fewer candidates and selecting top drugs based on median adjusted p-value and enrichment score ranks.
  5. Literature Validation: The top prioritized candidate drugs were further assessed through a literature review using PubMed, PubChem, and ClinicalTrials.gov to evaluate their preclinical evidence, toxicity profiles, and potential for AD treatment.

Datasets

  • GWAS Data: A large-scale SNP-based GWAS study of diagnosed late-onset AD (71,800 cases, 383,378 controls) from Jansen et al. (2019).
  • scRNA-seq Data: Single-nucleus transcriptomes from the prefrontal cortex of 48 individuals with varying degrees of AD pathology (80,660 transcriptomes) from Mathys et al. (2019), focusing on excitatory neurons (Ex), astrocytes (Ast), and oligodendrocytes (Oli).
  • Drug-Induced Transcriptional Profiles: Data from the Connectivity Map (CMap), comprising gene expression profiles of five cell lines treated with 1,309 different small molecules.
  • Clinical AD Drug Targets: Information on drugs in clinical trials for AD, retrieved from Therapeutic Target Database (TTD), DrugBank, and PubChem.

Key Insights and Ideas

  • Synergistic Power of Multi-Omics: The study demonstrates the value of integrating data from both GWAS (identifying genetic predispositions) and scRNA-seq (revealing cell-type-specific molecular changes) for a more comprehensive understanding of AD pathology and for identifying drug targets.
  • Pathway-Based Repurposing: The Gene2Drug approach is highlighted as a powerful method that doesn't just look for drugs directly hitting AD genes but also for those that can modulate the broader biological pathways these genes are involved in. This offers a more nuanced and potentially effective strategy.
  • Identification of Multi-Target Candidates: Several identified drugs appear to target multiple aspects of AD pathology (e.g., inflammation, Aβ production, tau pathology), suggesting they could be more effective than single-target drugs.
  • Preclinical Validation is Crucial: The study emphasizes that while computational methods can identify candidates, rigorous literature review and preclinical evidence are essential for validating their potential.

Key Results

  • The analysis generated four gene lists, including 255 risk genes from GWAS and cell-type-specific risk genes from scRNA-seq (532 from Ex, 103 from Ast, 82 from Oli).
  • Eight AD candidate drugs were identified: ellipticine, alsterpaullone, tomelukast, ginkgolide A, chrysin, ouabain, sulindac sulfide, and lorglumide.
  • Four of these drugs – alsterpaullone, ginkgolide A, chrysin, and ouabain – showed promising repurposing potential for AD based on preclinical evidence and moderate toxicity profiles.
  • The study also identified overlaps with known clinical AD drug targets, such as TREM2, CD33, and APP.

Limitations and Future Work

  • Limited Cell Type Inclusion: The scRNA-seq analysis focused on three major cell types (Ex, Ast, Oli), while six were identified in the original study. Future work could incorporate more cell types and other AD-specific scRNA-seq datasets.
  • Need for Experimental Validation: The identified drugs require further experimental validation in preclinical and clinical settings to confirm their efficacy and safety in humans.
  • Controversial Mechanisms: For some drugs, like lorglumide, the literature presented conflicting evidence regarding their role in AD, necessitating further investigation.
  • Severe Side Effects: Some drugs, such as tomelukast and sulindac sulfide, were excluded from further consideration due to severe adverse effects, highlighting the importance of toxicity assessment.
  • Broader Database Integration: The authors suggest extending the method to larger databases like the Library of Integrated Network-based Cellular Signatures (LINCS) to potentially improve drug discovery.
  • Personalized Medicine: The approach holds promise for facilitating personalized medicine by considering a patient's genomic composition and AD progression stage.

Summary updated

Abstract

<b>Background:</b> Traditional therapeutics targeting Alzheimer's disease (AD)-related subpathologies have so far proved ineffective. Drug repurposing, a more effective strategy that aims to find new indications for existing drugs against other diseases, offers benefits in AD drug development. In this study, we aim to identify potential anti-AD agents through enrichment analysis of drug-induced transcriptional profiles of pathways based on AD-associated risk genes identified from genome-wide association analyses (GWAS) and single-cell transcriptomic studies. <b>Methods:</b> We systematically constructed four gene lists (972 risk genes) from GWAS and single-cell transcriptomic studies and performed functional and genes overlap analyses in Enrichr tool. We then used a comprehensive drug repurposing tool Gene2Drug by combining drug-induced transcriptional responses with the associated pathways to compute candidate drugs from each gene list. Prioritized potential candidates (eight drugs) were further assessed with literature review. <b>Results:</b> The genomic-based gene lists contain late-onset AD associated genes (BIN1, ABCA7, APOE, CLU, and PICALM) and clinical AD drug targets (TREM2, CD33, CHRNA2, PRSS8, ACE, TKT, APP, and GABRA1). Our analysis identified eight AD candidate drugs (ellipticine, alsterpaullone, tomelukast, ginkgolide A, chrysin, ouabain, sulindac sulfide and lorglumide), four of which (alsterpaullone, ginkgolide A, chrysin and ouabain) have shown repurposing potential for AD validated by their preclinical evidence and moderate toxicity profiles from literature. These support the value of pathway-based prioritization based on the disease risk genes from GWAS and scRNA-seq data analysis. <b>Conclusion:</b> Our analysis strategy identified some potential drug candidates for AD. Although the drugs still need further experimental validation, the approach may be applied to repurpose drugs for other neurological disorders using their genomic information identified from large-scale genomic studies.
BibTeX
@inproceedings{xu2021computational,
  author = {Yun Xu and Jiming Kong and Pingzhao Hu},
  title = {{Computational Drug Repurposing for Alzheimer’s Disease Using Risk Genes From GWAS and Single-Cell RNA Sequencing Studies}},
  booktitle = {Frontiers in Pharmacology},
  year = {2021},
  doi = {10.3389/fphar.2021.617537},
}
Figure