Artificial Intelligence Uncovers a New Cause of Alzheimer's Disease and Identifies a Potential Treatment

April 25, 2025 — Researchers at the University of California San Diego have leveraged artificial intelligence (AI) to deepen understanding of Alzheimer's disease, pinpointing a gene previously identified as a biomarker as a direct cause. This discovery not only clarifies the disease's underlying mechanisms but also paves the way for novel therapeutic strategies.

The study, published on April 23 in the journal Cell, focuses on phosphoglycerate dehydrogenase (PHGDH), a gene known for its role in amino acid production. Earlier research had shown that PHGDH levels in the blood correlate with Alzheimer's progression, with higher expression linked to more advanced disease stages. However, establishing a causal relationship remained elusive.

Using sophisticated AI models, the team analyzed both mouse models and human brain organoids. They found that reducing PHGDH expression slowed disease progression, whereas increasing its levels exacerbated symptoms, confirming PHGDH as a causal gene in spontaneous Alzheimer’s disease.

Further AI-driven structural analysis revealed that PHGDH has a previously undiscovered role beyond its enzymatic function. It contains a structural subdomain similar to DNA-binding domains found in transcription factors. This enables PHGDH to activate genes that disturb gene regulation processes in the brain, contributing to Alzheimer’s development.

This dual role explains the elevated levels of PHGDH observed in Alzheimer’s patients’ brains. Importantly, the research uncovered a pathway that links PHGDH's regulatory function to disease pathology — a discovery that sets the stage for targeted interventions.

Motivated by these insights, the researchers identified a small molecule, NCT-503, capable of inhibiting PHGDH’s gene-regulatory activity without impeding its primary metabolic role. Computational modeling showed NCT-503 can access the protein’s DNA-binding site effectively.

Testing this compound in two separate mouse models demonstrated promising results: treated mice showed significant improvement in memory and reduced anxiety, common symptoms in Alzheimer’s patients. While the study acknowledges the limitations of current animal models, these findings offer hope for translating this approach into human clinical trials.

Lead researcher Sheng Zhong emphasized that this work could lead to the development of a new class of therapeutics that target upstream pathways in Alzheimer’s disease, potentially preventing the formation of amyloid plaques, rather than just addressing symptoms. The next steps involve optimizing NCT-503 and conducting safety and efficacy studies in preparation for clinical trials.

This pioneering research illustrates how AI can reveal complex biological mechanisms, opening new avenues for treating neurodegenerative diseases. By targeting the regulatory function of PHGDH, scientists aim to develop effective therapies that could alter the course of Alzheimer’s disease in the future.

Source: https://medicalxpress.com/news/2025-04-ai-unravel-alzheimer-disease-therapeutic.html