Alzheimer's disease master regulators analysis: search for potential molecular targets and drug repositioning candidates

Abstract

Background: Alzheimer's disease (AD) is a multifactorial and complex neuropathology that involves impairment of many intricate molecular mechanisms. Despite recent advances, AD pathophysiological characterization remains incomplete, which hampers the development of effective treatments. In fact, currently, there are no effective pharmacological treatments for AD. Integrative strategies such as transcription regulatory network and master regulator analyses exemplify promising new approaches to study complex diseases and may help in the identification of potential pharmacological targets.

Methods: In this study, we used transcription regulatory network and master regulator analyses on transcriptomic data of human hippocampus to identify transcription factors (TFs) that can potentially act as master regulators in AD. All expression profiles were obtained from the Gene Expression Omnibus database using the GEOquery package. A normal hippocampus transcription factor-centered regulatory network was reconstructed using the ARACNe algorithm. Master regulator analysis and two-tail gene set enrichment analysis were employed to evaluate the inferred regulatory units in AD case-control studies. Finally, we used a connectivity map adaptation to prospect new potential therapeutic interventions by drug repurposing.

Results: We identified TFs with already reported involvement in AD, such as ATF2 and PARK2, as well as possible new targets for future investigations, such as CNOT7, CSRNP2, SLC30A9, and TSC22D1. Furthermore, Connectivity Map Analysis adaptation suggested the repositioning of six FDA-approved drugs that can potentially modulate master regulator candidate regulatory units (Cefuroxime, Cyproterone, Dydrogesterone, Metrizamide, Trimethadione, and Vorinostat).

Conclusions: Using a transcription factor-centered regulatory network reconstruction we were able to identify several potential molecular targets and six drug candidates for repositioning in AD. Our study provides further support for the use of bioinformatics tools as exploratory strategies in neurodegenerative diseases research, and also provides new perspectives on molecular targets and drug therapies for future investigation and validation in AD.

Keywords: Alzheimer’s disease; Drug repositioning; Hippocampus; Master regulators; Transcription factors; Transcriptional regulatory network reconstruction.

Fig. 1

Transcriptional regulatory network and master regulators. a Human hippocampus transcription regulatory network centered on transcription factors was reconstructed from normal brain dataset (GSE60862). The network shows 469 regulatory units of transcripts classified under transcription factor activity, sequence-specific DNA binding (GO:0003714), with more than 25 inferred targets; 132 of them showed more than 100 inferred targets and were considered tissue-specific regulatory units (blue container). These were then tested in AD case-control studies using master regulator analysis, resulting in 34 tissue-specific regulatory units significantly enriched with differentially expressed genes (red container). The 337 regulatory units with less than 100 targets are represented in black outside the blue container. b Tile plot representation of the MR candidates for each case-control expression dataset (GSE5281, GSE29378, GSE36980, and GSE48350). ns not significant

Fig. 2

Activation state of MR candidates and AD subregulatory network. a Tile plot representing the state of activation of MR candidates (two-tail gene set enrichment analysis) for each case-control expression dataset. b Subregulatory network showing the associations between the significantly activated and repressed MR candidates. Node size represents the number of inferred targets of the master regulator transcription factor candidate; node shape shows their activation state; node color maps their connectivity (subnetwork average degree = 5.75 ± 2.65); edge width shows the Jaccard coefficient of common targets between transcription factor pairs. ns not significant

Fig. 3

Connectivity map analysis and drug repurposing to AD therapy. a Schematic representation of connectivity map analysis: differentially expressed targets of repressed or activated MR candidates, for each case-control study, were ranked and used as query signature to the connectivity map webtool against gene expression profiles database of several cell lines treated with thousands of FDA-approved compounds. b Case-control associated drugs: consensus drugs consistently matched with at least two case-control studies. Drugs with negative AD association are assumed with therapeutic potential, and the ones with positive association are considered AD mimetic. ns not significant