Network pharmacology-based discovery and experimental validation of novel drug repurposing candidates in Alzheimer's Disease

This study utilizes a network pharmacology framework combined with experimental validation to identify and confirm the therapeutic potential of repurposing the drugs TUDCA and arundine for Alzheimer's disease by rescuing disease-relevant phenotypes through the regulation of G protein signaling.

The Gist

Imagine Alzheimer's disease as a massive, chaotic city where the power grid is failing, the streets are clogged with trash (toxic proteins), and the emergency services (immune cells) are panicking and causing more damage than they are fixing. For decades, scientists have been trying to build new fire trucks from scratch to put out these fires, but the process is slow, expensive, and often unsuccessful.

This paper proposes a smarter, faster strategy: Drug Repurposing. Instead of building new fire trucks, the researchers asked, "What if we just took existing fire trucks, ambulances, and police cars that are already approved for other jobs and see if they can help fix this city?"

Here is how they did it, broken down into simple steps:

1. The Digital Map (Network Pharmacology)

The researchers didn't just guess. They built a giant, digital map of the human body's "social network."

• The City Map: They mapped out how genes talk to each other (the "interactome"). In Alzheimer's, certain genes form a "bad neighborhood" (the disease module) where things go wrong.
• The Drug List: They took a list of 2,413 drugs that are already approved by the FDA or in late-stage testing.
• The Proximity Test: They asked a simple question: How close is a specific drug to the "bad neighborhood" on this map?
  • If a drug targets a gene right next to the bad neighborhood, it's a strong candidate (like a fire truck parked right next to the fire).
  • If a drug targets a gene three blocks away, it's probably too far to help.

They ran this computer simulation and found that many drugs usually used for cancer were surprisingly close to the Alzheimer's "bad neighborhood." This made sense because cancer and Alzheimer's both involve cells growing or dying in the wrong ways.

2. The Shortlist

The computer gave them a long list of potential candidates. The researchers then narrowed it down to three "superheroes" based on two rules:

1. Can it get into the brain? (It has to pass the "border control" known as the blood-brain barrier).
2. Does it make sense biologically? (Does it fit the story of what's going wrong in Alzheimer's?)

The three winners were:

• TUDCA (a derivative of Chenodiol): Originally a bile acid used to treat gallstones. Think of it as a "cleaning crew" for the gut that might also clean up the brain.
• Arundine: A compound found in plants (related to indole-3-carbinol), often studied for cancer. Think of it as a "peacekeeper" that calms down angry cells.
• Cysteamine: A drug used for a rare kidney/eye disease called cystinosis. Think of it as a "garbage collector" that helps cells get rid of toxic buildup.

3. The Lab Test (Experimental Validation)

Computers are great, but you can't trust them until you see them in action. The researchers took these three drugs into the lab and tested them on 33 different cell cultures (tiny petri dish versions of brain cells).

They looked for specific "good behaviors":

• Did it clean up the trash? (Removing Amyloid-beta, the sticky protein clumps that kill brain cells).
• Did it stop the panic? (Reducing neuroinflammation, where brain immune cells attack healthy tissue).

The Results:

• TUDCA and Arundine were the stars of the show.
• They successfully helped brain cells clean up the toxic protein trash (Amyloid-beta).
• They calmed down the angry immune cells, stopping them from causing inflammation.
• The statistical evidence (using a method called "Bayesian analysis") was very strong, meaning it wasn't just luck; the drugs were genuinely working.

4. The "Why" (The Secret Mechanism)

Why did these random drugs work? The researchers dug deeper and found a common link: a protein called RGS4.

Imagine RGS4 as a traffic light controller for the brain's cells. In Alzheimer's, this traffic light is broken, causing chaos.

• Both TUDCA and Arundine seem to fix this traffic light.
• By fixing RGS4, they regulate a signaling pathway (like the EGFR pathway) that controls how cells grow, divide, and survive.
• This is why cancer drugs (which often target these same pathways) showed up on the list. They are essentially "re-tuning" the brain's traffic lights to stop the chaos.

The Bottom Line

This paper is a blueprint for a new way to fight Alzheimer's. Instead of waiting 10 years to invent a new drug, the researchers used a digital map to find old, safe, approved drugs that can be repurposed immediately.

They found that TUDCA (a bile acid) and Arundine (a plant compound) are promising candidates that can clean up brain trash and calm inflammation. While more testing is needed before these become official treatments, this study lights a path forward, suggesting that the cure for Alzheimer's might already be sitting in our medicine cabinets, just waiting to be tried on a new problem.

Technical Summary

1. Problem Statement

Alzheimer's Disease (AD) remains a progressive neurodegenerative disorder with no effective disease-modifying treatments. While genetic risk factors (e.g., APOE4) and molecular pathways are increasingly understood, translating this knowledge into therapeutic interventions has been difficult due to the disease's genetic and cellular complexity. Traditional drug discovery is slow and expensive. Drug repurposing—identifying new uses for existing FDA-approved or late-stage clinical drugs—offers a promising alternative. However, current repurposing efforts often rely on correlative approaches that lack mechanistic depth. This study aims to address this by integrating multi-omic data with network pharmacology to identify and validate novel AD candidates.

2. Methodology

The study employed a hybrid computational-experimental workflow:

A. Computational Framework (Network Pharmacology)

• Data Integration: The authors integrated genomic, transcriptomic, and proteomic data from AD studies (GWAS, TWAS, PWAS) to define eight distinct AD risk gene sets (modules). These included curated knowledge-based genes and differentially expressed genes (DEGs) from APOE3/3 vs. APOE4/4 individuals across neurons, astrocytes, and microglia.
• Network Construction:
  1. Drug-Target Network: Constructed from ChEMBL (v29), containing 25,329 drug-target pairs for 2,413 FDA-approved or Phase 3/4 drugs.
  2. Gene-Gene Interaction Network (Interactome): A human interactome with ~11,000 nodes and ~217,000 edges.
• Network Proximity Scoring: The study utilized a topological distance metric ($d$) to quantify the proximity between a drug's target genes ($T$) and the AD risk gene sets ($S$). Drugs with shorter path lengths (higher proximity) to AD modules were prioritized.
  • Formula: $d(S, T) = \frac{1}{|T|} \sum_{t \in T} \min_{s \in S} d(s, t)$
• Rank Aggregation: Eight separate drug rankings (one per AD gene set) were aggregated into a single consensus list using the MC3 algorithm (selected via modified Kendall distance).
• Validation Metric: A "rediscovery rate" was calculated to ensure the screen could recover known AD drugs (positive control) while identifying novel candidates.

B. Experimental Validation

• Candidate Selection: Top-scoring drugs were manually filtered for Blood-Brain Barrier (BBB) permeability and prior AD relevance. Three lead candidates were selected: Chenodiol, Arundine (3,3'-diindolylmethane), and Cysteamine.
• Phenotypic Assays: The candidates were tested in 33 cell-based assays across 9 experiments.
  • Note: Chenodiol was tested as TUDCA (tauroursodeoxycholic acid), its taurine-conjugated epimer, due to prior evidence of neuroprotection.
• Statistical Analysis: Dose-response data were analyzed using Bayesian nonlinear regression.
  • Hypotheses: $H_1$ (protective), $H_0$ (neutral), $H_2$ (adverse).
  • Metric: Bayes Factors (BF) were calculated to quantify the evidence for protective effects ($2 \times \log BF$).

C. Mechanistic Follow-up

• Network path analysis was performed to trace connections between the drug targets and AD genes, identifying key mediator proteins.

3. Key Contributions

1. Multi-Omic Network Integration: The study successfully combined eight diverse AD risk gene sets (including single-cell iPSC data) into a unified network proximity framework, demonstrating that topological proximity is robust even when gene set overlaps are low.
2. Novel Candidate Identification: Identified TUDCA (bile acid derivative), Arundine (indole metabolite), and Cysteamine as high-priority repurposing candidates.
3. Mechanistic Insight (RGS4): Uncovered Regulator of G protein signaling 4 (RGS4) as a critical hub linking the identified drugs to AD pathology.
4. Bayesian Experimental Validation: Moved beyond standard p-values by using Bayesian hypothesis testing to provide robust evidence for drug efficacy in specific phenotypic contexts (e.g., amyloid clearance, neuroinflammation).

4. Key Results

• Computational Validation: The network proximity screen successfully rediscovered known AD drugs at a rate 2–3 times higher than random chance, validating the approach's biological plausibility.
• Experimental Findings:
  • Arundine & TUDCA: Showed moderate to very strong evidence (high Bayes factors) for enhancing Aβ1-42 clearance in BV2 microglial cells.
  • TUDCA: Demonstrated moderate evidence for reducing the release of Aβ40 and Aβ42 in H4-hAPP cells.
  • Neuroinflammation: Both Arundine and TUDCA showed strong evidence for reducing LPS-induced neuroinflammation in BV2 cells and iPSC-derived human AD microglia.
  • Cysteamine: While prioritized computationally, it did not show the same level of strong phenotypic rescue in the specific assays tested compared to the other two.
• Mechanistic Pathways:
  • Network analysis revealed that Arundine, Chenodiol, and Cysteamine target RGS4.
  • RGS4 interacts with AD genes via EGFR (ErbB1) and ERBB3 signaling pathways (MAPK/ERK and PI3K/Akt/mTOR).
  • This suggests the drugs may modulate AD pathology by regulating G-protein signaling and downstream proliferation/differentiation pathways often dysregulated in AD and cancer.

5. Significance and Conclusion

• Therapeutic Potential: The study provides strong computational and experimental evidence that TUDCA and Arundine are viable candidates for AD repurposing. TUDCA, in particular, has a history of safety in ALS trials, potentially accelerating its translation to AD.
• Biological Mechanism: The identification of RGS4 and the EGFR/ErbB signaling axis offers a new mechanistic hypothesis for AD treatment, linking drug targets to amyloid processing and neuroinflammation.
• Methodological Advancement: The paper demonstrates the power of combining systems biology (network proximity) with rigorous Bayesian experimental validation to overcome the limitations of purely correlative drug discovery.
• Future Directions: The authors suggest expanding the network to include newer chemical entities and updating interaction networks, while noting that the current findings support further clinical investigation of TUDCA and Arundine in AD populations.

Limitations: The study is restricted to FDA-approved or late-stage drugs and relies on the completeness of existing drug-target and gene-gene interaction databases, which may contain noise or gaps.