In a groundbreaking advancement at the nexus of artificial intelligence and medicinal chemistry, Insilico Medicine, in collaboration with Huadong Medicine Company, has unveiled pioneering small-molecule inhibitors designed to target the elusive protein–protein interaction between WD Repeat-Containing Protein 5 (WDR5) and the MYC oncogene. Harnessing the profound capabilities of generative artificial intelligence combined with physics-driven molecular modeling, this research marks a significant leap forward in drug discovery, as detailed in the latest publication featured in Chemical Biology & Drug Design.
The MYC protein, long recognized as a central oncogenic driver implicated in up to 70% of human cancers, has historically been labeled “undruggable” due to its lack of conventional binding pockets suitable for small molecule inhibitors. MYC functions primarily by regulating gene transcription and cellular proliferation, but its oncogenic activity stems from complex protein–protein interactions that have resisted traditional pharmacological intervention. Recent insights revealed that the interaction between MYC and WDR5 is indispensable for the maintenance of MYC’s oncogenic functions, thereby spotlighting WDR5 as a novel and promising target in therapeutic development.
Breaking new ground, the research team employed Insilico’s generative AI-driven platform, Chemistry42, creating novel small molecules that precisely engage the WDR5 interface critical for MYC binding. The platform enabled a ligand-centric and scaffold-hopping strategy enhanced by ’anchor points,’ which preserved pharmacophoric features essential for high-affinity binding. Among the AI-generated candidates, two compounds distinguished themselves: compound 8 exhibited inhibitory potency with an IC50 value of 16.35 micromolar, while compound 9 demonstrated a significantly improved IC50 of 1.91 micromolar. These findings indicated marked improvements over a reference molecule, which displayed an IC50 of 20.86 micromolar, signaling notable enhancement in targeting this challenging PPI landscape.
Recognizing the potential of these initial hits, further optimization was carried out through rigorous physics-based modeling facilitated by Chemistry42’s AlChemistry module. This approach enabled deep structural analysis and refinement of molecular interactions and binding conformations within the WDR5-MYC interface. As a result, lead compounds with sub-micromolar affinities were engineered, culminating in the identification of the standout molecule 9c-1. This lead showed a remarkable 35-fold increase in inhibitory activity relative to earlier analogs, specifically compound 3, showcasing exceptional binding strength and specificity against WDR5. Such potency positions 9c-1 as a trailblazer in the design of efficacious inhibitors capable of disrupting MYC-driven oncogenesis through direct interference with its protein–protein engagement.
The implications of this breakthrough are profound. The successful application of an AI-guided generative chemistry technique, integrated seamlessly with physics-anchored validation, underscores a paradigm shift in tackling traditionally “undruggable” targets. This study exemplifies how advanced computational platforms can rapidly generate candidate molecules with therapeutic promise, accelerating early-stage drug discovery timelines dramatically compared to conventional methodologies. Insilico Medicine’s innovative combination of machine learning and molecular modeling successfully circumvents longstanding challenges in drug design, especially for complex PPIs long deemed refractory to small molecule intervention.
Dr. Xiao Ding, Senior Vice President and Head of Chemistry & DMPK at Insilico Medicine, emphasized the significance of these findings, stating, “Our AI-powered platforms are transforming drug discovery by unlocking possibilities for targets previously considered inaccessible. This project demonstrates the synergistic power of generative chemistry aligned with physics-based modeling, delivering molecules that could herald new therapeutic paradigms for cancers driven by MYC.” The integration of computational creativity with empirical rigor has expedited the transition from conceptual targets to potent leads, offering hope for treating malignancies with profound unmet medical needs worldwide.
This achievement builds on a rich legacy of Insilico Medicine’s leadership in artificial intelligence applications for drug design. Initially conceptualized in 2016 within peer-reviewed literature as a pioneering use of generative AI for molecule creation, Insilico’s platforms have evolved to commercial maturity via Pharma.AI, a comprehensive digital ecosystem deployed extensively in early drug development pipelines. By uniting deep generative neural networks, reinforcement learning techniques, transformer architectures, and physics-based simulations, Insilico Medicine has optimized target identification and compound generation, significantly compressing drug discovery phases from an average 2.5–4 years down to 12–18 months per program.
Moreover, leveraging automated synthesis and high-throughput biological testing, Insilico Medicine has propelled over two dozen internal programs between 2021 and 2024, synthesizing and validating 60–200 molecules per candidate initiative. This integrated AI-drug discovery approach not only expedites lead identification but enhances molecular novelty and diversity—overcoming traditional attrition hurdles frequently encountered in medicinal chemistry campaigns focused on complex targets such as transcription factor PPIs.
The WDR5-MYC inhibitory compounds represent a new class of focused PPI disruptors, embodying a strategic shift to modulate oncogenic pathways at the protein interaction level rather than canonical enzymatic inhibition. Disrupting the assembly of oncogenic transcriptional complexes via WDR5 offers a promising intervention point with the potential to arrest cancer proliferation and survival mechanisms. Crucially, this approach illustrates the feasibility of rational PPI drug design supported by AI, challenging preconceived limitations in medicinal chemistry and expanding the therapeutic landscape for challenging targets across oncology and beyond.
Looking forward, the medicinal chemistry team aims to advance the 9c-1 lead through preclinical evaluations, exploring pharmacokinetics, toxicity profiles, and efficacy in cancer models. The translational potential of these findings opens avenues for addressing cancers driven by MYC dysregulation, including lymphoma, leukemia, and a spectrum of solid tumors. Furthermore, the AI-driven discovery methodology exemplified here serves as a model for future drug discovery efforts targeting other difficult proteins implicated in disease pathogenesis.
In conclusion, the collaboration between Insilico Medicine and Huadong Medicine Company showcases the transformative impact of integrating generative AI and physics-based modeling in uncovering novel therapeutic agents. This research not only delivers powerful WDR5 inhibitors with the potential to modulate the MYC oncogenic axis—a longstanding unmet challenge in oncology—but also validates an innovative drug discovery paradigm poised to revolutionize how next-generation medicines are designed and optimized.
Subject of Research: Discovery of small-molecule inhibitors targeting the WDR5-MYC protein–protein interaction using AI-driven generative chemistry and physics-based molecular modeling.
Article Title: (Not explicitly provided; refer to DOI link)
News Publication Date: May 28
Web References:
- Chemical Biology & Drug Design article: https://onlinelibrary.wiley.com/doi/10.1111/cbdd.70129
- Insilico Medicine website: https://insilico.com/
- Pharma.AI platform: https://pharma.ai/
- Previous Insilico concept article: https://pmc.ncbi.nlm.nih.gov/articles/PMC5355231/
References: DOI 10.1111/cbdd.70129 (journal article detailing the research)
Keywords
Medicinal chemistry, drug discovery, generative artificial intelligence, protein–protein interaction inhibitors, WDR5, MYC oncogene, pharmacophore modeling, molecular docking, physics-based molecular modeling, AI-driven chemistry, cancer therapeutics, small-molecule inhibitors
