From mechanistic modeling to AI-driven design: computational strategies for targeting the γ-secretase complex

Advancements in computational biology are transforming the study of complex membrane proteins and their therapeutic targeting. The γ-secretase complex, a quintessential intramembrane protease implicated in Alzheimer's disease (AD) and more than 150 other substrates, provides a powerful exemplar to illustrate this transformative shift. Traditional γ-secretase inhibitors have been constrained by off-target toxicity, particularly through disruption of Notch signaling, underscoring the need for deeper mechanistic insights, now increasingly enabled by modern computational methodologies. We evaluate the computational strategies driving next-generation drug discovery of γ-secretase. Integrative modeling frameworks, informed by cryo-electron microscopy (cryo-EM) and biophysical data, have facilitated atomic-resolution reconstructions of γ-secretase dynamics and substrate recognition. All-atom molecular dynamics (MD) simulations, supported by enhanced sampling techniques such as umbrella sampling, steered MD, replica exchange, and Gaussian accelerated MD, have mapped conformational landscapes and elucidated molecular determinants of substrate selectivity. Structure-function mapping of familial AD mutations further demonstrates how computational modeling translates genetic variation into mechanistic understanding. Beyond structural modeling, the integration of artificial intelligence (AI) including deep generative models, machine learning-based activity prediction, and high-throughput virtual screening has created accelerated pipelines for discovering modulators predicted to reduce pathogenic amyloid beta (Aβ) production while preserving essential signaling pathways. These approaches demonstrate how computational methods increasingly serve as predictive and design-oriented engines in drug development. Using γ-secretase, this review highlights how state-of-the-art computational techniques, from integrative structural biology to AI-driven drug design, are reshaping the discovery of safer, more selective modulators with broader relevance across diseases requiring precise modulation of protein function.